Wikipedia:Bots/Noticeboard/Archive 2

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

PHP bot using curl

I am programming PHP bot for adding new settlement in slovenian wikipedia (some basic data). Is there any working PHP bot, I am interested about part which handle saving data do server (entire process), probably using cURL. tnx for any info --Slodave 19:18, 12 February 2007 (UTC)

I don't know of any off the top of my head, but they might exist. You can browse Wikipedia:Registered bots to look for one, if you'd like. —METS501 (talk) 19:57, 12 February 2007 (UTC)

My CbmBOT runs on PHP/cURL. –Dvandersluis 20:16, 16 February 2007 (UTC)

SatyrBot also runs on PHP/cURL. -- SatyrTN (talk | contribs) 05:12, 22 February 2007 (UTC)

My bot (Jayden54Bot) also runs on PHP, although it doesn't use cURL, but simply uses the fsockopen() function. I've written a complete framework to do most of the common tasks, and in the near-future I'll post the framework on SourceForge or somewhere else. Jayden54 10:05, 22 February 2007 (UTC)

Werdnabot down?

it seems like Werdnabot, a talk page archiver, is down and the user hasn't responded on his talk page in some time, anybody know what's up? --AW 19:22, 1 March 2007 (UTC)

Look at his contributions. He just did an archival job about three hours ago.--Dycedarg ж 19:27, 1 March 2007 (UTC)

Yeah, my mistake, I had the link pointing at the wrong page --AW 19:36, 1 March 2007 (UTC)

Caching the edit token

I've been used this technique for speeding up edits of my bot for a while, so let's see if someone else is already using it.

The idea is to try and avoid downloading the edit form, if possible. This form contains about 50k of HTML markup (in addition to the raw page content) which is mostly useless for a bot. The only things from it that are needed are: the start time, the edit time, the edit token, and the content. An alternative way to get this data is:

start time ; this is returned from live preview with content {{CURRENTTIME}};

edit time and (previous) raw content ; query.php allows getting both of them in a single request (in some cases the previous content may not even be needed).

The only part that is only available via the edit form is the edit token. However, the edit token only depends on the user and the php session, which means that the same token can be used for a sequence of changes. So the idea is to cache the edit token; the procedure to change a page is:

1. get the starttime via live preview
2. get edittime (and possibly the previous content) via query.php
3. change the content
4. submit using the cached edit token
5. if submission fails with an "expired session" error:
   1. load the edit form for the page
   2. get the edit token from it and cache it
   3. submit the page again

In the worst case (the token is expired), this is more work than just loading the edit form. However, in a sequence of consecutive edits, this will only be needed the first time. The problem with having the starttime before the edittime and content is that a false delete/edit conflict may result. I guess this is unlikely and does not create any problem anyway.

So, is anybody using this method already? Tizio 16:34, 11 March 2007 (UTC)

To be honest, I think that's too much hassle for very little bandwidth/speed gain. Most of the time the edit page will need to be loaded anyway to get the article content, protection status, etc. Another disadvantage with your method is that it takes several requests to get all the necessary data so it might not even save any bandwidth or speed. So I don't think it really helps. What we need is a (much) better MediaWiki API, but I don't know if anyone is working on that. Jayden54 18:53, 11 March 2007 (UTC)

That is a good idea, I may use it in future projects. I assumed the edit token was unique for each edit, Thanks. ^High_InBC^{(Need help? Ask me)} 18:54, 11 March 2007 (UTC)

I'm not using a caching method, but your idea sounds like a good one. However, what I would really like is the ability to get an edit token from the MediaWiki API. —Remember the dot (t) 18:59, 11 March 2007 (UTC)

You mean you can re-use the edit-token!? That's awesome! Thanks for pointing that out! -- SatyrTN (talk | contribs) 20:16, 11 March 2007 (UTC)

Thanks for the feedback, everybody! Yes, if one only needs to edit a couple of pages at time, all this is unnecessary. However, I have tested this system, and even on relatively few pages (4 or more) the improvement is noticeable. As for having the edit token directly from API, that was actually what I initially asked for at User talk:Yurik/Query API#Edit interface; all of this will be made unnecessary when the "edit" function of api.php will be complete, but there is no schedule for that, as far as I know. Tizio 19:09, 15 March 2007 (UTC)

"Undo" page

Hi, my bot did some mistakes and I would like to revert it's edits using the "undo" feature:

'/w/index.php?title=Title&action=edit&undo=2810979'

The problem is that the above feature will open the page for editing, and I want to immediately save the page. Is there a way to do this? Yonidebest 15:43, 18 March 2007 (UTC)

If you install popups, and add &autoclick=wpSave then the page will save itself automatically as well. Tra (Talk) 16:19, 18 March 2007 (UTC)

Well, (a) Doesn't this mean that by bot will need to use the popups for it to work? and (b) the link

'/w/index.php?title=Title&action=edit&undo=2810979&autoclick=wpSave'

does not save the page =/ Yonidebest 17:08, 18 March 2007 (UTC)

(a) Sorry, I thought you meant you wanted to undo the edits manually. If you intend to use a bot to do this, then you would want to get the contents of the form that appears and program the bot to submit it. (b) You weren't running popups so it didn't work. Tra (Talk) 23:14, 18 March 2007 (UTC)

It's likely easier and better to manually undo, as, aside from the problem you're writing about, any interim edits may cause the edit to not even be undoable. —Daniel Vandersluis^(talk) 23:00, 18 March 2007 (UTC)

Generally, you need to POST your edit to commit them. Haven't checked for undo, however. Tizio 12:43, 19 March 2007 (UTC)

Undo just brings up the normal edit screen, but preloaded with the undone revision; I think you then have to POST the edit form to actually do the edit. --ais523 12:44, 19 March 2007 (UTC)

WP:BAG Addition

An addition to WP:BAG is being held at Wikipedia_talk:Bots/Approvals_group#New_Member, please comment there if you desire. — xaosflux ^Talk 00:46, 22 March 2007 (UTC)

WP:BAG: Betacommand

There was a discussion on AN/I about whether or not Betacommand should be removed from BAG. I moved the discussion to Wikipedia_talk:Bots/Approvals_group in case there is future discussion. Perhaps this will just die off. In any case, I'm posting this notice here since it is quite applicable and a forced removal would be unprecedented. -- RM 13:38, 22 March 2007 (UTC)

Bots and external links → WP:WPEL

This is not a response to the note above, but somehow related. We have a WikiProject External links at Wikipedia. Unfortunately it has been quite inactive in the past but I'm trying to change this. Since external links seem to have been and be a controversial topic especially in combination with bots, I'd like to ask you to notify WP:WPEL about any bot request and any bot issues targeting external links. I'd also like to invite bot owners who already run or requested bots related to external links — join the project, your work is appreciated! This WikiProject is the appropriate place to discuss and coordinate EL stuff and to prevent disputes that nobody wants before they happen. Thank you! — Ocolon 17:27, 22 March 2007 (UTC)

DB lock

Is there a language-independent way to detect DB lock? Textarea can be read-only if page is protected, and lock message does not seem to be enclosed in any kind of unique formatting. MaxSem 19:24, 22 March 2007 (UTC)

No idea. I would need to see a db-lock message to check. —METS501 (talk) 20:22, 22 March 2007 (UTC)

AWB has long had a habit of stalling: on db-lock, on timeout, on disconnection. I would often wake to find my bot stalled, so I had to resort to putting a nudge timer into my plugin. This is a dirty hack I'd really like to get removed, so any progress on stopping AWB stalling when these things happen (or building the nudge timer into AWB) would be great. --kingboyk 23:00, 22 March 2007 (UTC)

I'd support the nudge timer built into AWB. Basically, I'm envisioning a timer that starts when AWB is ready to save, and if x seconds pass with no save then it tries to resave, and so on. —METS501 (talk) 23:07, 22 March 2007 (UTC)

That's pretty much how my plugin does it; it starts a timer and if 10 minutes elapse it fires AWB a stop event followed by a start event. --kingboyk 23:20, 22 March 2007 (UTC)

I would have it more like 60 seconds, but it sounds good. I would implement it. —METS501 (talk) 23:22, 22 March 2007 (UTC)

Discussion on the AWB part of the thread can continue at Wikipedia_talk:AutoWikiBrowser/Dev#Nudge_timer. --kingboyk 14:13, 23 March 2007 (UTC)

Reedy Bot (Contributions) blocked

Resolved

See my comments here. El_C 22:47, 22 March 2007 (UTC)

I was asked to do Category:Israel, and subcategories. The sub categories i presume are accounting for most of the bad edits. It was done as per Wikipedia:WikiProject_Israel#Scope.

So yeah, i'll agree, its my fault, i sorta nievely (sp?) assumed that if it was a subcategory of the main cateogry, it needed to be tagged.. Which TBH, makes sense to me - As if its in a sub category, surely, the majority are related?. The was no intention to tag articles that weren't needed.

I will for future, make sure i do like like User:Lostintherush suggested, - User_talk:Reedy_Bot#Bot_account_blocked, and just do it on a category by category basis, based on what is approved.

However, its my fault whichever way, and for which, i apologise

Hope this clears up the fault

Reedy Boy 08:26, 23 March 2007 (UTC)

Good man for holding your hands up. We all make mistakes, it's no huge deal and the bot was stopped before it went too wild. Just check each category is correct before adding it next time, or do them one by one. --kingboyk 14:11, 23 March 2007 (UTC)

The issue has now been resolved. El_C 20:18, 23 March 2007 (UTC)

New bot framework

I've written a rudimentary bot framework in Java which can be accessed here. I've tested it on the test wiki and it works. I haven't put it to use on the English Wikipedia yet. Comments and improvements are welcome. MER-C 11:36, 28 March 2007 (UTC)

Looks great as a start; I may even use it myself. How about query.php pagelist making? If you're looking for other ideas, see [1]. The dotnetwikibot framework could use some improvements itself, for that matter. —METS501 (talk) 16:29, 31 March 2007 (UTC)

I've just implemented the most useful of these: listing the pages in a category. I'll probably write the rest as I, or someone else prodding me on my talk page, requires it. MER-C 12:17, 1 April 2007 (UTC)

Block status

How does one tell if one's bot has been blocked? Or unblocked? Besides not being able to edit, that is :) -- SatyrTN (talk | contribs) 14:16, 5 April 2007 (UTC)

Check the block log at Special:Log/block, and fill in User:BotName as title. For example, my bot's block log would be available at [2] (my bot has never been blocked, so no items are shown in this case). Jayden54 14:23, 5 April 2007 (UTC)

And it appears that your bot (SatyrBot) is not blocked anymore, according to the block log. It did get blocked today at 01:19 but was unblocked 2 hours later. Jayden54 14:26, 5 April 2007 (UTC)

It does appear that way, but my bot is unable to edit. I can edit if I log in as the bot, but when it's running by itself it isn't able to. Thoughts? -- SatyrTN (talk | contribs) 14:47, 5 April 2007 (UTC)

Did you try resetting your computer and clearing your cache? Other than that, no idea. Are you using AWB or your own software for your bot? —METS501 (talk) 16:31, 5 April 2007 (UTC)

That would be my best guess as well; some sort of caching issue. If the usual steps fail (clear everything cached, re-start, etc) then you might have to do some debugging (if you're running your own software). Jayden54 18:04, 5 April 2007 (UTC)

You could be getting hit with the autoblocker. Try to log in as your bot and edit, and get the autoblock number. — xaosflux ^Talk 02:47, 6 April 2007 (UTC)

He said he can edit if manually logged as the bot. Without knowing at least the result header/page when the bot tries to submit an edit, diagnosis is almost impossible. Tizio 11:27, 6 April 2007 (UTC)

LOL - sorry about that :) It seems to be coding error and I haven't had time to debug since I posted above. Thanks for all the help! -- SatyrTN (talk | contribs) 13:34, 6 April 2007 (UTC)

Evidently it was a caching issue. Last night for no apparent reason, it started working again. <sigh> Isn't technology wonderful? :) Thanks again for everyone's help. -- SatyrTN (talk | contribs) 16:55, 6 April 2007 (UTC)

Ralbot's unmatched small tags

Recently when delivering the Signpost, Ralbot has been missing off the final <small> HTML tags, causing talk-page errors. Should this bot be blocked or not?? I'm not certain whether it should or not... --SunStar Net ^talk 09:47, 7 April 2007 (UTC)

No need for a block. Ral315 has been informed of the problem and it will be fixed for the next time that the bot delivers the signpost (hopefully). Until the bot is actually running with the error, a block is unwarranted. Martinp23 09:50, 7 April 2007 (UTC)

Agreed. I ran in to this, and found that it was already announced and that corrective measures would be in place. Blocking is not warranted at this time, but can be done if there are future malfunctions until they are repaired. — xaosflux ^Talk 13:53, 7 April 2007 (UTC)

A recent software fix caused the malformed HTML to fail; otherwise, it would have been gracefully ignored. Will be fixed next week. Ral315 » 20:39, 7 April 2007 (UTC)

Great! —METS501 (talk) 21:14, 7 April 2007 (UTC)

Replace a whole page

Anyone happen to have a pywikipedia mod that replaces a whole page (as opposed to a regex edit)? If not, I'll almost certainly code this myself. -- Rick Block (talk) 04:48, 15 April 2007 (UTC)

What do you mean? Once you generate the new pagetext for a page, you might use page.put(newtext, msg) to place it. Or do you need something else? Gimmetrow 04:56, 15 April 2007 (UTC)

I'm thinking something callable from a shell, like replace.py, that takes as input the page name and a file that has what I'd like the new contents to be (or, alternatively, the name of an external filter to run the existing content through to obtain the new content). -- Rick Block (talk) 05:03, 15 April 2007 (UTC)

The first part is easy enough to code. If you have any knowledge of python you should be able to do it in five minutes, or I could if you wanted. Although, why exactly would it be faster for you to tell a program the name of a page and a file that you want posted to it instead of just copying the file into the page yourself normally? As for the second, what precisely do you mean? What kind of external filter are you talking about, and what would it do?--Dycedarg ж 09:30, 15 April 2007 (UTC)

I have a number of scripts I currently run (see Wikipedia:Bots/Requests for approval/Rick Bot), that produce page content that I have been yanking and putting into the page manually. I'm intending to make some of these fully automatic (cron jobs). I do most things in unix shell, so like curl (or wget) can fetch a page (to stdout) what I really want is a pywikipediabot command (probably putpage.py) to put a page (from stdin or a file). The surround from replace.py (show the diff, prompt if ok, various params) would be good to preserve as well, so this isn't just a 5 line wrapper around page.put. The external filter idea would be another way to accomplish the same thing. What it would do is set up the existing page content as the filter's stdin and capture the filter's stdout as the "new" page content. This would allow the content manipulation to be done by an arbitrary external program (like awk or sed or perl or ...), extending the pywikipediabot capabilities to any string manipulation language. -- Rick Block (talk) 16:30, 15 April 2007 (UTC)

I have a modified version of replace.py that takes a new parameter (-content:filename) and replaces the content of a page with the content in the named file. I think doing this as a filter (per above) would actually be better (e.g. a new parameter like -filter:program), but this will suffice for now. If anyone wants my modified version or codes up a version invoking a filter, please let me know. -- Rick Block (talk) 03:51, 17 April 2007 (UTC)

Mass Link Replace

I've recently gotten a request that sorta scares me, so I wanted to get opinions from other bot owners. I should first say that I haven't gotten permission to run my bot in any namespace but "Talk", which this procedure would require — but assume for the moment that the bot has that permission.

The request is to change all wikilinks of WP:EA that currently point to the (inactive) Wikipedia:Esperanza so that the shortcut can be used by a new (active) Wikipedia:Editor assistance. It was followed up by a discussion at the Village Pump that yielded almost no comments.

A quick calculation showed a total of 47,000 pages that currently have links to WP:EA. At the moment I can't remember if that's just to WP:EA or also to Wikipedia:Esperanza. In any case, the sheer numbers make me nervous.

Any thoughts or comments? Thanks! -- SatyrBot 21:17, 16 April 2007 (UTC)

As long as you have a flag and edit lass than 10 times/minute, everything should be fine. You should also make sure the bot stops if you or the bot get a post on their talk page. Cbrown1023 talk 21:20, 16 April 2007 (UTC)

Not without specific approval for the task first, please, as it's a potentially controversial one. --kingboyk 21:21, 16 April 2007 (UTC)

Well, I assume that that would have been a given, but you are right, I should have mentioned it. Cbrown1023 talk 21:29, 16 April 2007 (UTC)

Definitely - as I said, assume I've gone through the process to get that permission. And I will before doing anything different from talk page edits. -- SatyrTN (talk | contribs) 21:33, 16 April 2007 (UTC)

I think it would be the easier if the new project chose a redirect name that wasn't already in use on 47,000 pages! (notwithstanding the fact that Esperanza is no longer active). --kingboyk 21:21, 16 April 2007 (UTC) (e/c)

The project's direct page is different - it's the shortcut that currently points to the inactive project. -- SatyrTN (talk | contribs) 21:33, 16 April 2007 (UTC)

(added) Sorry - I misread your comment. You're right, but you have to admit, "WP:EA" is easy and logically should point to something like "Editor assistance". But that's not my concern - that's up to the project :) -- SatyrTN (talk | contribs) 21:35, 16 April 2007 (UTC)

WP:EA is without doubt a good redirect for them, but I'm not sure how the community would feel about the usurpation and resultant 47,000 edits (mostly because Editor Assistance is new). Maybe we should just wait a few short weeks and see how the new project does? If, as I hope, it's successful, I can't see there being too much objection to this task. Better to use the redirect for a project that helps editors in distress than for a glorified MySpace, right? :) --kingboyk 22:21, 16 April 2007 (UTC)

This should go through a specific bot task approval due to it's huge scale. — xaosflux ^Talk 01:13, 17 April 2007 (UTC)

Given the potential controversy involved in changing these redirects, I thought it best to seek consensus on the matter at WP:RfD - see Wikipedia:Redirects for discussion/Log/2007 April 16. At present it seems unlikely that a change will be supported so the above discussion may be moot. WjBscribe 01:53, 17 April 2007 (UTC)

Prolonged absence of Bot owner

Hi, I seem to remember that Bot policy requires Bots to have an owner active on Wikipedia- how long does an owner have to be away for this to become an issue? Hagerman has now not edited since Feb 6. HagermanBot generates a lot of questions from people who don't understand how it functions. Also, a number of requests for possible improvements to the Bot's operation have gone unanswered over the last couple of months. It's a useful Bot and is functioning as expected but it is a concern that there is no one to address issues with it. Just wanted your thoughts on the matter. Cheers, WjBscribe 14:01, 16 April 2007 (UTC)

(note: the above was copied from my talk page) I've emailed the bot operator to get a status on them. Bots need to have respondant operators. — xaosflux ^Talk 01:15, 17 April 2007 (UTC)

Technically, a bot user is active if it responds to the queries brought up by his bot. (However, Hagerman bot is a different case then...). Cbrown1023 talk 02:10, 17 April 2007 (UTC)

February 6 is long ago in wiki time. Unfortunately we may have to consider this bot operator-less. I say "unfortunately" because HagermanBot is immensely useful.

Let's see if he responds to xaosflux's email; perhaps if the op isn't coming back soon he can allow a clone to operate? --kingboyk 13:04, 17 April 2007 (UTC)

I've received notice that this operator has NOT left the project, and will be responding to feedback soon, reported as being away for work. Thank you, — xaosflux ^Talk 01:06, 19 April 2007 (UTC)

I say let the bot run until there is an issue the requires the operator's attention. ^High_InBC^{(Need help? Ask me)} 01:25, 19 April 2007 (UTC)

Cydebot Block

(moved from my talk — xaosflux ^Talk 05:22, 22 April 2007 (UTC))

(Copied from User_talk:Cyde)

Your bot has been blocked

Your bot has been blocked as a malfunctioning bot. Your bot was editing in excess of 50 edits/min, in violation of the Bot policy. Bot accounts typically should run at up to 6 edits/min, unless performing tasks deemed "urgent", when a top rate of 15 edits/min has been approved. As blocks are not meant to be punitive and you are an administrator, please feel free to remove this block and resume bot edits once you have adjusted your bot's edit rate. You should also feel free to immediately remove any autoblocks or other collateral damage related blocks hit as a result of this block. If feel you must edit at this rate, please talk page me and we can bring this to a forum where more community input can be gathered. Thank you, — xaosflux ^Talk 01:32, 22 April 2007 (UTC)

Note from Mindspillage

Hey, I noticed you blocked Cydebot. So, my question is: why—what harm was it doing?

Yes, I've seen the guidelines. But what are they for?

The bot has about 45000 more edits to make this run. A massively large-scale renaming effort with almost no potential for misapplication seems to justify the high rate.

What reason is there to block it? It's already done about 40,000 without performing incorrectly (that many would surely have given people enough edits to go by to lodge complaints, if there were any to make), and it's not significantly affecting server resources. If anything, the higher speed seems like a good idea to make sure it will finish during Wikipedia's lower-traffic days where it won't affect load so much.

The renaming of the image license templates to reflect the renaming of the guideline pages is an issue very near and dear to my heart... or my spleen, or something... :-) and due to the large amount of time it will take to do, the lack of need for supervision, and the minimal effect on resources, I am strongly in favor of an unblock so it can continue. If this isn't something you are willing to do, I'd like to know why, and to get more input. Cheers, Kat Walsh (spill your mind?) 03:07, 22 April 2007 (UTC)

Reply

Mindspillage, I've got a few comments on this bot and my block:

1. I am not inclined to unblock this account until the operator is available to discuss it.
2. As to the block reason, it is editing in violation of the bot policy, as for the reasoning (besides OMG POLICY VIO!), it is because bots editing at extreme speeds make it harder for independant edit validation, and can cause flooding of logs.
3. Wikipedia:Non-free content/templates is only ~5 days old, but alread has rules in place requiring naming standards. The bot's replacements, while not modifying the displyed content of the page, do not appear to have been put through the normal discusion used when replacing template instances (especially 85000 of them), on WP:TFD (at least not one where I can find it).

   I find isssue with this all together if there hasn't been wide spread agreement on this new style, but not to the point where I'd block the account. Nonetheless, as the end result on the page is nothing other then renaming the template, I don't even see this as an urgent need waranting the 15e/min rate (but also would not have blocked if it was limited to that).

4. I've briefly looked through the prior bot approval's pages and can not find where Cydebot has been approved to process template replacements on the Image: namespace; (found one for the user: and userspace namespace), but as this bot has successfully processed countless CFD's and generally operates without much error I would not have blocked for that either.

— xaosflux ^Talk 03:32, 22 April 2007 (UTC)

Reply to reply

100,000 edits isn't going to make a small footprint however you consider it. At such a scale, I simply don't see the point. No one is going to go through and independently verify those edits to any degree that makes the difference in speed matter. As for flooding logs, no one has complained of it as far as I can tell; it is bot flagged, of course, so stays off of recent changes.

The template name change was proposed about a month ago on the talk page of the Non-free content page (formerly called "fair use"), and no one has contested it; it's largely motivated by the rename of the image licensing pages to make the names consistent and adhere with the requirement that non-free content be easily machine-identifiable.

The gist is this: I don't see any substantive difference at this scale between 6 edits a minute and 50. If it edits at all, any difference in editing rate isn't going to have any effect that anyone will appreciate. While it'd be nice if Cyde were around, I don't think it makes a difference. If it actually malfunctions, sure, cut it off and go message him about it, but if not I just don't see the point. Kat Walsh (spill your mind?) 04:22, 22 April 2007 (UTC)

This bot is approved to make several other types of edits, are these being made simultaneously? Or are they being suspended during this run, if not they will be hard to monitor. This and other reasons are behind the limit on bot edit speeds. I've left a message for the others in the bot approvals group to leave more input here on this. — xaosflux ^Talk 04:47, 22 April 2007 (UTC)

Cydebot was still doing its approved tasks, see [3]. Now it kinda can't update the speedy deletion page which is unfortunate. :-\ --Iamunknown 05:02, 22 April 2007 (UTC) Oh, and the contribs for which it is approved can be monitored by following that link (to answer your second question). --Iamunknown 05:11, 22 April 2007 (UTC)

Additions to reply to reply

Kat pointed out that the template rename discussion started almost a month ago, so the 5 day comment is a bit off the mark.. but it should also be noted that this is a move not a deletion, so you wouldn't expect a discussion at TFD. The /template subpage is a working page to keep track of the actual rename project. As you can see from the edit history it's had quite a few direct participants. Based on the lack of material objections to the rather simple and uncontroversial edits being performed, I'm going to recommend that the bot be unblocked right away so that it can complete its current tasks before it runs into heavy traffic times during the week. --Gmaxwell 04:29, 22 April 2007 (UTC)

I see the longer history on that page now, was only led to the /template page though, as that is what was in the edit summary. — xaosflux ^Talk 05:21, 22 April 2007 (UTC)

I suppose RFD would be more approriate. — xaosflux ^Talk 04:49, 22 April 2007 (UTC)

You mean RM? --Iamunknown 05:02, 22 April 2007 (UTC)

Except the redirects probably won't be deleted for a couple of months... not until people get used to the new templates. Until then we plan on having the bot go clean up all new use. --Gmaxwell 05:10, 22 April 2007 (UTC)

I meant RfD, as that is the action that is basically going on, if these don't need to be deleted, then this has become a redirect replacement only, and again, what's the hurry? Redirects are cheap. — xaosflux ^Talk 05:21, 22 April 2007 (UTC)

I wish you'd read more of the pre-existing discussions. The reason the pages need to actually be changed is that a primary driver for the change is machine readability of the license status from the wikitext of the image pages. Unless there is some indicator of the license status in the actual image page wikitext, someone working off the dumps or a database replica must operate a complete copy if the mediawiki parser, which is very slow, in order to have any hope of figuring out the license status of the images. The change addresses that issue, but not if the images are not changed. "Whats the hurry"?? This is a change that was discussed for a month... it's one that in addition to cleaning up a bunch of misconceptions, is needed to fulfill enwikipedia's obligation under the foundation licensing guidelines. --Gmaxwell 05:43, 22 April 2007 (UTC)

Just wondering... why was such a large task not proposed at WP:BRFA? —— Eagle101 ^{Need help?} 05:33, 22 April 2007 (UTC)

Good question. — xaosflux ^Talk 05:35, 22 April 2007 (UTC)

Um, because the bot approval group has no oversight over image licensing procedures on english wikipedia, at least I can see no reason why it should or why I would have expected it to. ::shrugs:: --Gmaxwell 05:43, 22 April 2007 (UTC)

Indeed we don't, which is why the task would get approved whether or not I or other BAG members personally agree with changing, say, the album cover template (I don't). Provided there is commmunity consensus - or at least little likelihood of complaints - we approve; this would be easy to approve then because consensus is clear and it's Foundation policy too. Where we do have jurisdiction (see above for current ArbCom case where this seems to be in the process of being confirmed) is over bot activity. I daresay this task can be given a special waiver, but (as below) it has to be applied for first. Nobody except perhaps Jimbo has carte blanche to operate a bot at such high speeds without getting it approved first. --kingboyk 12:20, 22 April 2007 (UTC)

Comment

So, basically, y'all have blocked Cydebot from doing obviously valuable and necessary work simply because someone didn't file Form 27-BXT? Can we please stop wasting time filing TPS Reports and get back to writing an encyclopedia? Kelly Martin (talk) 05:45, 22 April 2007 (UTC)

Nope, it was blocked due to fast editing rate. See WP:BOT. —— Eagle101 ^{Need help?} 05:46, 22 April 2007 (UTC)

Indeed. This is a demand for a shrubbery. Please get out of the way of getting real work done and unblock the bot already. If the devs have a problem with the edit rate, I'm sure they'll let us know. Kelly Martin (talk) 05:49, 22 April 2007 (UTC)

If that is the case, then I suggest that you make that recommendation to WT:BOT. I believe the edit rate is there for good reason... what would happen if the bot went nuts and did 1000 screwups before someone could grab an admin. If we trust bots not to screw up, and want them to go at faster rates, then bring it up on the bot policy page. —— Eagle101 ^{Need help?} 05:52, 22 April 2007 (UTC)

If you don't know what the reason for the edit rate guidelines, I suggest you refrain from enforcing (or even arguing) them. This bot has already made thousands of edits on this task without error; expecting it to suddenly start doing so is slightly paranoid. Again, I believe that you are demanding a shrubbery. Kelly Martin (talk) 05:57, 22 April 2007 (UTC)

I may be, I don't know, lets see how this whole bit plays out :) —— Eagle101 ^{Need help?} 06:00, 22 April 2007 (UTC)

You'd best ask your friends at ArbCom. BAG approval is needed, not to satisfy our "thirst for power" (do you think I actually enjoy this mundane task?!) but because it's technically necessary. --kingboyk 11:31, 22 April 2007 (UTC)

Point that bothers me

What bothers me about this, I suppose, is something that bothers me about discussing admin action in general that is often overlooked: lack of admin action is an admin action, too. How many admins saw the bot working or knew what it was going to do and thought "oh, there goes cydebot" and let it be? That's a decision, too. It made thousands of edits on this run already; there must have been at least a few who did. Those who don't act don't have to register their opinion or provide rationales for not acting, of course, which makes it hard to say if others care or not, and I don't want to undo someone's block in non-urgent circumstances without clearly showing that sentiment swings the other way. But I know of several admins who don't object, and currently of only xaosflux who does.

I don't, in general, follow bot goings-on; I'm paying attention to this one only because I care about the template renaming. What I see as someone not deeply involved in it is that something was happening that was outside the guidelines, but wasn't doing anything substantively rather than procedurally objectionable, and the reasons I've seen that the guidelines actually exist for don't seem to be a factor, as they either don't make a difference on this scale or, like the contributions by namespace filtering, aren't an issue.. Kat Walsh (spill your mind?) 05:47, 22 April 2007 (UTC)

I believe the block was made due to a violation of bot policy (fast edit rate, max is 15 edits per minute, it was going at 50). —— Eagle101 ^{Need help?} 05:49, 22 April 2007 (UTC)

Eagle 101, I approved it editing at a rate of one operation interval sleep per operation or 1 edit per second, whichever is slower. --Gmaxwell 05:52, 22 April 2007 (UTC)

Errr... where did you approve it editing that fast? Generally bots don't edit that fast to prevent too much damage should something error out.... Then again I'm not a member of bag. —— Eagle101 ^{Need help?} 05:55, 22 April 2007 (UTC)

If you run the bot at 15 edits per minute the task will complete in about 90 hours, which can be done easily if you host it on toolserv. Or you can just leave your computer on. —— Eagle101 ^{Need help?} 05:57, 22 April 2007 (UTC)

Um. This batch. By the end of the transition every non-free image on enwiki will be be touched. Also add time lags from pipeline bubbles in the decisions to keep or get rid of templates that have to happen before the bot can run, and such.. and you're taking about making the process take months for no particular reason. It was my understanding that Cyde intended to run at higher speeds during low traffic times. The fact that this is unproblematic is demonstrated by the fact that it ran for many hours without incident, and also made similar edits at low speeds all week last week without trouble.

Cyde asked me via email what would be as safe rate to conduct this rather large set of edits from a server load perspective. As far as it 'going nuts' goes, with it simply doing a 1:1 template name replacement there is pretty much no risk of that, and with some 40k edits already made it's not likely to do any harm at this point. --Gmaxwell 06:04, 22 April 2007 (UTC)

Last I checked, bot operations were approved as a result of a bot approval request, how did this approval process run? — xaosflux ^Talk 05:56, 22 April 2007 (UTC)

Additionaly, it appears the operator has set this to run even faster then that delay [4]. — xaosflux ^Talk 06:01, 22 April 2007 (UTC)

No it wasn't, it's running with a higher interval.. above.. i.e. slower. :) --Gmaxwell 06:07, 22 April 2007 (UTC)

Yes, that's what it says, but that doesn't address my comment. I know it did not follow the letter of procedure, but what I want to know is how it was actually causing harm, and if anyone who saw it edit has objected to any of the edits it was making or thought its going slower would be a substantive improvement. Kat Walsh (spill your mind?) 06:08, 22 April 2007 (UTC)

Note, This thread is getting in to many issues that may be easily resolved once the operator is available, is there an assertion of immediate urgency on why this can't wait? — xaosflux ^Talk 06:01, 22 April 2007 (UTC)

Please let me know where to send the shrubbery. Kelly Martin (talk) 06:05, 22 April 2007 (UTC)

Gmaxwell how many images was this task going to hit? —— Eagle101 ^{Need help?} 06:09, 22 April 2007 (UTC)

Most recent estimate I heard was that there were around 400,000 non-free images on enwiki... but that sounds a bit off to me but they are hard to count because our current tagging isn't very machine readable (I can't just query the database to find out). It's in that general ballpark. Cyde won't be editing them all right now, because some templates are up for deletion rather than being renamed. Only the ones that people are sure aren't changing are being renamed now.--Gmaxwell 06:15, 22 April 2007 (UTC)

Ok, I did not know, in any case, when Cyde gets on he can explain why he did not do a simple WP:BRFA, and went past what our current policy says is the fastest you can go. I'm sure its all just a mistake :). Unblocking the bot now probably won't do anything as it likely has errored out. —— Eagle101 ^{Need help?} 06:31, 22 April 2007 (UTC)

At this point I'm tempted to put the BAG pages up for deletion, I do not believe that the community would approve of this sort of disruption of actions which are harmless and uncontested. Nor do I think the community would approve of the makeup of the bag or the methods for their selection. There is no technical or editoral reason for this disruption. And as far as I can tell at this point the BAG is simply obstructing valid activity in order to make a point about their authority to do so. If that is the case, it is reasoning made in error: no users group has the authority to disrupt valid edits just to show off their power to do so. --Gmaxwell 06:46, 22 April 2007 (UTC)

That's a bit much, considering I was in bed asleep at the time! You might want to see Wikipedia:Requests_for_arbitration/Betacommand/Proposed_decision#Automated_editing too. --kingboyk 11:05, 22 April 2007 (UTC)

I think it was the actions of one admin, I don't think he minds if it is unblocked, and was probably expecting cyde to be around. I really don't know much more then that. All I've been doing is explaining why he probably did what he did. Perhaps this was a case of WP:IAR, on the part of Cyde? I mean the block is valid per WP:BOT, but it might have been a case where WP:BOT should be ignored. —— Eagle101 ^{Need help?} 06:56, 22 April 2007 (UTC)

As a random member of the Community passing through, I'd like to point out that as I see it the reason we have BAG is that most of us do not understand BOT policy (or more precisely the rationale behind it) and so defer to those who do. Unless someone can show that BAG has completely lost the plot I suspect it would survive MfD. In any event, as only one BAG member has commented in this thread it seems a little premature to draw sweeping conclusions. WjBscribe 07:02, 22 April 2007 (UTC)

I find myself with little option but to agree with you, Greg. There was no reason for Xaosflux to believe that Cydebot's edits were harmful; there is no reason for him to insist on waiting to "speak to the operator" before unblocking either. As far as I can tell, the reason for this block is nothing more than "I was not consulted and I have granted myself the right to be consulted". This is, quite simply, disruption of important and legitimate efforts to improve the encyclopedia, for the sole reason of protecting personal power. As such, it is totally unacceptable. If this is reflective of the normal behavior of the bot approval group, then the bot approval group needs to be either restructured or disbanded. I am reasonably certain that has never been any demonstrated community consensus for the bot approval group's rules, procedures, or authority. I am also reasonably certain that many of the bot approval group's policies are arbitrary and have no technical or editorial merit, but exist merely to exist. The edit rate rule is quite certainly one of these arbitrary policies. Kelly Martin (talk) 07:05, 22 April 2007 (UTC)

Technical problems introduced by bots editing at extreme paces go beyond the processing power of the database and caching servers. Linus's Law teaches us that given enough watchers, all problems will be uncovered. Unfortuantley we do not have an infinite supply of watchers. When a multi-purpose bot goes off to make thousands and thousands of edits in a short time, it makes it even harder for the watchers to determine if it's other purposes are having issues. It also makes it hard to detect other problems (e.g. Recent Image Changes (warning large link) now is only availble for the last 7 hours of edits. Running at full speed (as listed above) would make this log max out at 83 mins.) While some options (i.e. hide bots in the RC log) are in place to help alleviate this, they have limits (i.e. you have to hide ALL bots). My block is not "just because I can" but becuase I have good faith concerns over this operation, and at the LEAST feel that further community input is warranted. — xaosflux ^Talk 07:13, 22 April 2007 (UTC)

FWIW, I would have blocked this bot for editing at this speed even had I not been in the approvals group. — xaosflux ^Talk 07:27, 22 April 2007 (UTC)

Alright, so we looked at it, it's fine, lets move on. -- Ned Scott 07:17, 22 April 2007 (UTC)

While you all get hot under the collar about your precious free image crusade, what exactly is wrong with {{albumcover}} anyway? Why waste server resources renaming it, when 99.999% of album covers are copyright?! --kingboyk 11:07, 22 April 2007 (UTC)

I'm still waiting to be informed why moving {{albumcover}} to {{Non-free album cover}} is worth the server resources. The bast majority of album covers aren't freely licenced. Why not leave albumcover as is and just create a new {{Free album cover}} template? Are people not capable of reading the blurb on the template or what?! Is this a prelude to removing all album covers from Wikipedia? --kingboyk 15:02, 22 April 2007 (UTC)

It allows third part users to remove non free images simply by killing anything with a template starting with "Non-free" on it.Geni 15:53, 22 April 2007 (UTC)

I see. Thanks. --kingboyk 15:54, 22 April 2007 (UTC)

Another bot for now?

While we're waiting for word from Cyde, would it be alright to have another bot work on the template renaming (at a much slower pace, of course)? -- Ned Scott 07:31, 22 April 2007 (UTC)

Is there another Bot flagged to do this sort of work? WjBscribe 07:36, 22 April 2007 (UTC)

This debate has brought up many things! I won't personally enforce any blocks on bots performing this task at the approved rate while this debate is ongoing, but others may if the proposed bot is not approved for the task. FWIW, new bot approval requests are handled at WP:RFBOT. — xaosflux ^Talk 07:38, 22 April 2007 (UTC)

This is such a minor housekeeping task, though.. -- Ned Scott 07:41, 22 April 2007 (UTC)

If it is the request to be approved should go by quickly, and a case for going faster then what WP:BOT allows can be made. Anyway hopefully this will get sorted out by the time I wake up. ;) —— Eagle101 ^{Need help?} 07:44, 22 April 2007 (UTC)

Hypothetically if Xaosflux speedy approved the replacement Bot there is a crat around who might be willing to flag it. But per Xaosflux's question above- where's the urgency if its just "minor housekeeping" that this Bot would do? WjBscribe 07:46, 22 April 2007 (UTC)

Like I said, I won't block it unless it's racing. I would not be inclided to speedy approve this task, but to maintain neutrality while this debate is in progress I wouldn't deny it either. Other bot approvers are availble though. — xaosflux ^Talk 07:50, 22 April 2007 (UTC)

MartinBotIII is approved for "template substing and renaming per consensus" (or, it should be :)), so if it's urgent, and the bot is needed, I can have it complete the task at an edit rate of 9-10/min by Tuesday/Wednesday (guess). However, it would probably be better to wait for Cyde to get back and finish the job for himself. As for the 50epm rate - I find it surprising that a bot can possibly fetch the wikitext of a page, and put the new wikitext, at such a rate (though this is probably the fault of my internet connection). The bot was editing in deifance of policy, both on the maxed out edit rate and the lack of approval, both of which could be valid reasons for a block. However, it seems that the task was simple (and the operator experienced), and not something that one would expect any admin to block for (BAG or not is irrelevant - any admin has the power to block bots). This does not excuse the edit rate, which should have been specifically requested if required, but again it is at the discretion of the particular admin whether to block or not. For 40000 pages, an edit rate of just 8/9 per minute will get them done in a couple of days (MartinBotIII is doing such a lengthy task now) - the concern that it would take months is nonsense :) Martinp23 09:14, 22 April 2007 (UTC)

You missed a zero. It was 400,000 pages, which I had at 18 days at 15 e/m. I agree that 50 e/m should have been specifically requested, but at any rate, I had an idea that might fix some of the shrubbery concerns. Now, it's getting late where I am and I reserve the right to change my mind in the morning, but it's posted at Wikipedia Talk:Bot policy. --Selket ^Talk 10:05, 22 April 2007 (UTC)

Ooops - sorry. I read 40k a few times above, and it stuck :) Martinp23 10:09, 22 April 2007 (UTC)

Yes, as for the speed, I do have a server on a nice campus network. I actually could go a lot faster than one edit per second ... but the devs assured me that wouldn't be such a good idea :-P Cyde Weys 13:55, 22 April 2007 (UTC)

Heh. /me is jealous. --kingboyk 13:58, 22 April 2007 (UTC)

Endorse block

I endorse the block. Nobody - except perhaps Jimbo Wales - is excluded from the bot rules. If Cydebot were running at a little over 15ppm, I would personally have turned a blind eye. 50ppm is, however, over 3 times the limit.

I didn't impose the limit; I don't know who did and on what basis. What I do know is that however noble the task, folks can't go and unilaterally decide to ignore it. By all means let's have a discussion and quite possibly agree that Cydebot can go faster for this task; for now I endorse the block. --kingboyk 11:35, 22 April 2007 (UTC)

I fully agree with kingboyk, whilst this task might be fine to run at a higher speed it should be discussed before it is allowed to continue doing so. Adambro 11:58, 22 April 2007 (UTC)

Good block; non-urgent task that doesn't need to be run faster than the agreed-on rate. I see no reason to break the rule for this specific case, and a change of the general rule (for which there are good non-technical reasons) should be discussed beforehand. Kusma (talk) 12:53, 22 April 2007 (UTC)

You people are being completely pompous idiots. The only reason for the rate limit is to protect the wiki. A developer - that is, someone competent to decide - has approved it. You can "endorse" the block all you like, but you are in fact not competent to decide. Stop acting like you know more about the systems than the people who run them - David Gerard 20:06, 22 April 2007 (UTC)

That's a personal attack. Please retract it. --kingboyk 20:08, 22 April 2007 (UTC)

Agreed, that is out of line. If you cannot make your point without name calling, then perhaps you need to think longer before posting. ^High_InBC^{(Need help? Ask me)} 20:10, 22 April 2007 (UTC)

Excuse me. You are acting like completely pompous idiots in this instance. You blocked it, he pointed out a dev okayed it as harmless, and rather than the obvious thing to do - "ok, no worries, let us know in advance next time" - you spend thousands of words defending the policy against the people who are actually responsible for the servers the policy is supposed to protect. It's spectacular, and evidence for the urgent need for severe rationalisation and ground-up rewrite of bot policies. Preferably by devs - David Gerard 20:30, 22 April 2007 (UTC)

Get your facts straight. I unblocked. Also, at no point in time did anybody say "gmaxwell is a dev and he approves it". Then it would have been discussion over, wouldn't it? --kingboyk 21:53, 22 April 2007 (UTC)

I agree that the "idiots" comment is uncivil and uncalled for. One of the big problems here has been lack of communication. Cyde didn't make it clear where he had gained approval for his bot and it wasn't made clear that Greg was a developer nearly soon enough. I find the assumption that members of BAG have some empire building agenda an incredible assumption of bad faith when they seem to have been nothing but helpful. Bot policy at present requires administrators to block Bots acting outside the scope of their permission or exceeding the present edit rate limit. If the Bot policy needs changing, fine- that can be discussed and done. But it appears to me that those who have responded from BAG have been amazingly civil and helpful given the barrage of critism and accusations they have been receiving. WjBscribe 01:19, 23 April 2007 (UTC)

I'm back

Okay, I'm back. What is needed of me now? As far as I can tell, I did everything correctly. If anything, I am only guilty of going over the BAG's heads and asking the CRO and devs directly about what kind of editing rate I should use. I'm sorry I didn't notify — I didn't realize it would somehow be a problem — but now that everyone knows, I think Cydebot should get back on its task. There is a lot more work to do and this is a change that is definitely wanted by the Foundation. --Cyde Weys 13:53, 22 April 2007 (UTC)

Good question. Best submit a task approval request I suppose; approval should be a formality if the Foundation want this done. As for the edit rate, we've not had a case like this before. If you have some "official" sanction perhaps provide details of that and we can rubber stamp it? BAG don't proclaim to have authority over the Foundation or the devs :), it's just as per above these things need to be cleared on wiki first otherwise other folks will think "Cyde can do it, my task is just as important, I'm going at an edit a second too").

In the meantime, I'll unblock your bot (if it's not been done already). --kingboyk 14:02, 22 April 2007 (UTC)

I've unblocked. Off the record etc etc, if I were you I'd carry on but keeping the edit rate down for now. --kingboyk 14:05, 22 April 2007 (UTC)

It appears that this bot is still racing. Currently running at an average speed of: 20e/min on this task alone, and an aggregate rate of 25e/min as it appears to be running multipe task threads. As this isn't to the extreme that it was at before I'm not going to wheel war over the block, but really would like to know why it needs to run at 166% of the maximum approved rate. — xaosflux ^Talk 15:38, 22 April 2007 (UTC)

As I said on my user talk page, I unblocked it because Cyde was back (which is normal procedure for bot blocks unless they're totally rogue), and unofficially recommended he carried on at 15ppm max, pending a proper bot task application. If you reblock because it's racing that's fine by me (and presumably by the other editors who agreed with my endorsement of the original block). No wheel warring is necessary or implied if you reblock due to changed circumstances :) --kingboyk 16:00, 22 April 2007 (UTC)

(moved from my talk — xaosflux ^Talk 19:25, 22 April 2007 (UTC))

Lets be clear, I think bot policy is important and needed. But what we do not need are meat-bots enforcing the bot policy. If the BAG is going to disrupt harmless activities because they run at 50e/min vs 15e/min, without any clear technical or editorial reason beyond "we said not to" then I think it doesn't deserve to exist and it should be replaced with something more competent and less dysfunctional. And there really is no technical reason the editing can't run at that rate.... Most tasks shouldn't run at that rate for editorial reasons and sometimes technical ones, but most tasks will not need to touch anywhere near this number of pages. Most tasks are also not making an utterly trivial change to image pages which is almost impossible to do incorrectly. --Gmaxwell 17:41, 22 April 2007 (UTC)

See my response on:WP:BOWN (ultra short summary: Technical issue:log flooding; blocking I would have blocked this even if I wasn't on WP:BAG for the same reason). — xaosflux ^Talk 17:55, 22 April 2007 (UTC)

What log flooding? It is bot-flagged. And why would you think that something showing up as a continuous drone for a week is better than it showing up more frequently for a day? So if I push a temporary change to mediawiki to suppress the creation of RC entries in the image namespace for cydebot, you will have no more objections? --Gmaxwell 18:02, 22 April 2007 (UTC)

That would be one of the logs, another would be anyone trying to review the contributions of this bot during it's run's through Special:Contributions, as this bot does work on things other than this task. If this would have been brought to bot review I would have suggested that it run under a unique account for this special task, especially if it will be running at abnormal rates. Note, that these concerns are purley from the technical nautre, but the actual reasoning for the edits appears to be getting questioned by other above (I really don't care myself which template is used). — xaosflux ^Talk 19:23, 22 April 2007 (UTC)

Regarding special:contributions, even 15e/min will quickly scroll them off unless you're looking by namespace. Once you look by namespace there is no issue.

"actual reasoning for the edits appears to be getting questioned by other above", difflink?

I think it's also interesting to note that relative to the overall editing rate rambot's runs[5] were at a much greater speed than cydebot's edits here. I can't help but see Xaosflus's complaints as nothing more than a pointless power game.

Since I have already approved the higher rate the argument is moot. --Gmaxwell 19:44, 22 April 2007 (UTC)

You still haven't answered the simple question of what position gives you the authority to approve of it. I'm not meaning to be disrespectful, and am probably showing my ignorance in even asking, but if you're approving as a dev or a board member or whatever please just say so :) --kingboyk 19:55, 22 April 2007 (UTC)

A process to approve bots exists not only to ensure that they comply to technical specifications, but to ensure they are performing tasks that the community wants / needs. An important aspect of that is documenting the task so that anyone can understand it upon later review. That is why the bot approval process is done on-wiki, in the open, with a community discussion period. I am infering (perhaps wrongly) from your reply that you are declaring yourself to be an absolute authority when it comes to approving this, where has the community vested you with this role? As this is getting heated, I am going to go have a nice cup of tea and revist this later. — xaosflux ^Talk 19:58, 22 April 2007 (UTC)

You appear to be empire-building, rather than protecting the wiki. Stop assuming competence of areas you are not competent to decide upon, especially when directly contradicted by a developer, who is competetent to decide upon said area - David Gerard 20:08, 22 April 2007 (UTC)

The edits per minute rule was never subject to the devs opinion, while it's clearly a techcnical matter. If the developers are saying that there's no harm in going over 15 epm, then the rule becomes automatically arbitrary without any strong argument other than the BAG feud wanting it to be. -- drini ^[meta:] _[commons:] 20:11, 22 April 2007 (UTC)

And yes kingkboy, greg's is a dev. -- drini ^[meta:] _[commons:] 20:11, 22 April 2007 (UTC)

Where on gmaxwell's user page does it say that he is a dev? Why have I had to ask several times in what capacity he approved it, before getting an answer? (and then get shot down in flames for asking). I can't be expected to know everybody and I've politely asked more than once. The real answer, of course, is that I'm not a member of the IRC cartel, no doubt this is playing out in that arena right now.

As for empire building, I can disprove that notion by resigning from the group, which I am happy to do right now. --kingboyk 20:15, 22 April 2007 (UTC)

I think this fuss might have been avoided had there been a note on Cydebot to the effect: "Approved by dev:whoever to run at 50edits/min for task whatever". This would have at least explained the situation to anyone questioning the unusual edit rate. Is that empire building or common-sense public notice? Gimmetrow 20:26, 22 April 2007 (UTC)

And, this is sort of tangential but way more important: can I go back up to the faster editing rate without being blocked again? Have we pretty much worked it out now, regardless of whatever prior notification could have been given previously? I'd really like to get this task finished within a week rather than within a month. --Cyde Weys 20:33, 22 April 2007 (UTC)

Cyde - consider the bot approved for the task and edit rate requested and and allowed by the developer above (who, if common sense previals, is the authoritiative figure here). I do invite any BAG member to overrule me, but for the time being, lets all try to de-escalate the situation. Martinp23 20:48, 22 April 2007 (UTC)

Of course that's correct. I don't want a turf war with devs, and would - in the absence of any evidence otherwise - assume them to have greater authority over the servers than the bot approvals group. --kingboyk 21:55, 22 April 2007 (UTC)

As a complete outsider to the bot groups and the Foundation group and the developer group and any other groups involved here, can I just agree with what Gimmetrow said above: "I think this fuss might have been avoided had there been a note on Cydebot to the effect: "Approved by dev:whoever to run at 50edits/min for task whatever". This would have at least explained the situation to anyone questioning the unusual edit rate." - it is painfully clear to anyone watching this from the outside that what has happened here is a lack of communication. Carcharoth 20:54, 22 April 2007 (UTC)

Indeed - more communication is needed. When this communication isn't present, we get several (often well respected) users finding out about the issue, thinking that the actions involved were completely reprehensible, and making comments attacking the process. Of course, all of these comments are made in hindsight, with the benefit of a lot of information not available to those taking the steps they felt neccessary at the time. This is a recurring pattern on Wikipedia, where one small action causes the wheels of outrage to trundle on, until the next slightly-controversial action comes along... Martinp23 21:45, 22 April 2007 (UTC)

Nail. Head. Hit squarely. --kingboyk 01:42, 23 April 2007 (UTC)

The role of the BAG

If everyone could stop bickering for a minute and actually have a real conversation we might actually accomplish something.

I can assure you one thing: the BAG is not trying to build an empire over the bots that run on Wikipedia. If everyone was perfect and no one would possible run a bot that doesn't following the bot policy and help the wiki, then there would be absolutely no need for the group, and I'm sure no one would have a problem deleting the bot approvals page and abolishing the bot approvals group. The group was established precisely because this is not the case. Everyone in the group has technical and policy knowledge enough to approve or reject bots. It is also an open forum to promote discussion and consensus. If any developers would like to join the BAG and have enough policy knowledge as well, they are more than welcome. Most people think developers can trump consensus, and this has some truth, but only in half the cases. The developers can stop actions that they deem harmful, but they cannot approve actions that they deem harmless; that role is delegated to the community.

One thing fundamental is ignored often on this wiki, and perhaps this should even be a policy. No one, and I mean no one, is immune to the rules. It just doesn't work like that. If some people were immune and others weren't, everything would collapse. It just doesn't work. Even if you think your bot task is uncontroversial and save, you still need approval from a member of the bot approvals group in a forum open for discussion. It is impossible to predict the opinion of everyone, so even if you and a developer think something is OK, the rest of the community may not. This task is very uncontroversial, and would no doubt have been speedily approved, but there would be a record of this approval for everyone to see. Also, there was strong opposition when discussed of raising the bot edit rate over 15 edits per minute. I pushed for this change a few months ago, and it was controversial raising it from 6 to 15. Now imagine 50, over 8 times the original maximum edit rate. What if the bot messed up for 10 minutes on little-used pages which meant no one noticed. That's 500 edits, a MASSIVE amount to clean up afterwards. 15 edits per minute is a good compromise, reached by community consensus. If you want a change in the policy to get 50 edits per minute, then discuss it, don't just do it. I hope I've made this point clear.

If George Bush went ahead and just suddenly declared war on Canada and said afterwards that he thought it was necessary, and that he had the agreement of one other high-level official, the American people would be insulted and feel violated. Cyde, even though you may not think it, you're trying to be the President of Wikipedia, and the rest of the people feel they have no control over your actions, and because you have a few people supporting you, you feel that you are in the right and everyone else is wrong. Sorry, but that's the truth. People are afraid to approach you because you don't talk constructively, and you think that you never err. Everyone has to obey consensus, and when there is none, they cannot simply create a 2-person consensus.

People claim there is too much bureaucracy here, but really, the problem is that people spend more time trying to get around it thereby causing a huge amount of discussion, during which which everyone cites the rules. If people would just follow the rules, and not feel like they are the king, things would just get done faster.

Sorry for the stupid analogy above, and I hope I haven't offended anyone; I certainly didn't mean to, I just needed to get my point across.

Now, back to my Wikibreak :-) —METS501 (talk) 21:04, 22 April 2007 (UTC)

But why was BAG immune to the rules when it was created without consensus? ... or was Essjay immune to the rules, but now he's gone so no one is? --Gmaxwell 21:20, 22 April 2007 (UTC)

It's funny and inappropriate that you blame cyde here, since he wasn't the one in all the discussion claiming that the bag was off the mark here... it's been a lot of other people, including myself. So if you're going to make statements like this you need to target them at the right people. --Gmaxwell 21:16, 22 April 2007 (UTC)

I have no idea how or when BAG was created; I joined some time after creation. Was it created without consensus? It's quite intruguing that the arbitrators don't seem to think so (see current Betacommand case, where they are coming to the view that BAG has authority over bot issues).

Why is it okay to slag us off for doing a volunteers' job that we were under the impression we had the mandate to do? Why is xaosflux getting stick for blocking a high speed bot that didn't have any approval, and why am I getting stick when the only action I took was to unblock it?

Where in Wikipedia policies and guidelines does it say that consensus is formed on IRC?

What is the precedent for saying that rules can be overridden because an influential group of Wikipedians say so? (I'm not talking of WP:IAR here, which AFAIC doesn't apply to automated operations.

All that said, clearly the process needs to be rethought, and I am no longer convinced the process has community support. Therefore I am going to resign from the group. I have better things to do with my time. --kingboyk 22:04, 22 April 2007 (UTC)

No responses. Speaks volumes. --kingboyk 01:47, 23 April 2007 (UTC)

On wiki notice is important, this whole thing would never have happened with it. Right now the method is through bag (to notify about new bot tasks whatever). If that is to change, someone needs to start some discussion somewhere, perhaps back to the old bot noticeboard that we used to have before bag took its place. —— Eagle101 ^{Need help?} 21:19, 22 April 2007 (UTC)

I have always maintained that the maximum bot edit rate is mostly decided on non-technical considerations. While there may be a technical limit, I don't believe anyone has reached it yet. We routinely run bots on the server side that edit with a rate only limited by the speed of the database servers.

The main technical concern is that an excessive write rate will cause the slave servers to lag. To avoid this on the server side, all of our scripts monitor lag and pause if it goes above 5-10 seconds. If client-side bots did this too, I would have no problem with edit rates on the order of 100 per minute. -- Tim Starling 21:11, 22 April 2007 (UTC)

Server-side bots? Couldn't this image template renaming thing have been done by a server-side bot? Carcharoth 21:31, 22 April 2007 (UTC)

Hey, if they'll give me an account ... :-D Cyde Weys 21:33, 22 April 2007 (UTC)

That did occur to me too; if the task is so important and so massive why isn't it being run by devs on the server? I did wonder though whether the API is much less efficient than anything the devs have, unless it were done by way of an SQL update query. --kingboyk 21:58, 22 April 2007 (UTC)

We have better things to do. Editing via HTTP is good enough. - Tim Starling 22:23, 22 April 2007 (UTC)

Thanks for the reply. --kingboyk 22:24, 22 April 2007 (UTC)

Isn't that what toolserver is for? Actually, I don't know if Cyde has a toolserver account but I could run the bot from mine - 0.1 ms ping to the wikipedia servers, and no need to waste Tim's time. --Draicone ^(talk) 13:35, 3 July 2007 (UTC)

Toolserver's database is read only. --ST47Talk 14:32, 3 July 2007 (UTC)

It appears to be "OK" for certain people to slag off the work of (other) volunteers, because evidently in our constitutional monarchy (with no constitution, and indeed no constitutional conventions), they've received some sort of implicit patronage to do so. If the powers that be think that the bot policy as currently written sucks in general, or simply needs to set aside in a particular case (and there might be an excellent argument for either position), couldn't this at least be communicated (to the community in general, and to the BAG in particular) in advance of such bunfights as this? Simply going ahead and doing it anyway, and justifying it after the fact on the basis of "IARing all rules because our empire trumps your empire-- er, because it's the right thing, that's it" is deeply unhelpful. I don't see much chance of "No one is immune to the rules" becoming policy, since the practical outworking of the IAR "policy" is precisely the reverse. Alai 23:27, 22 April 2007 (UTC)

Thank you Alai. I do believe you're right. BAG were never empire builders, we were/they still are working for the community. We should have been informed of this decision; communicating it to us would have avoided this mess, and we don't deserve the resultant criticism. If anything, criticism should be flowing the other way for making decisions off wiki and scapegoating us when it went pearshaped. We were doing a volunteers' job in good faith and get personal attacks for our efforts. Seems a tad unfair to me and I'm quite unhappy about it :( --kingboyk 01:47, 23 April 2007 (UTC)

Look, rules are irrelevant. What is important is the intentions behind those rules. Limiting edit rate is in place to ensure that nothing can go wrong and that bots can't go mad and make thousands of hard-to-reverse edits before someone stops them. But this is a professional developer who has been trusted with modifying the system managing the information on this wiki - heck, he could probably even figure out your password. Yet he cannot be trusted with the simple task of approving a trivial bot? Since when was BAG the all powerful community group with regards to bots? In fact, has anyone even looked at the source code in question? If we look at CydeBot's scripts and have twelve experienced developers approve them as perfectly harmless, would people agree to let the bot edit at a decent rate? While we're at it, why rate limit at all once we know that nothing could possibly go wrong, just to get it over and done with? --Draicone ^(talk) 13:28, 3 July 2007 (UTC)

Maxlag parameter

On thinking about the maximum edit rate problem, I've introduced a hidden parameter called maxlag. You can set it for any request to index.php, e.g. [6]. If the specified lag is exceeded at the time of the request, a 500 503 status code is returned, content type text/plain with a response body with the following format:

Waiting for $host: $lag seconds lagged\n

Recommended usage is as follows:

• Use maxlag=5 (5 seconds). This is an appropriate non-aggressive value, used by most of our server-side scripts. Higher values mean more aggressive behaviour, lower values are nicer.
• If you get a lag error, pause your script for at least 5 seconds before trying again. Be careful not to go into a busy loop.
• It's possible that with this value, you may get a low duty cycle at times of high database load. That's OK, just let it wait for off-peak. We give humans priority at times of high load because we don't want to waste their time by rejecting their edits.
• Unusually high or persistent lag should be reported to #wikimedia-tech on irc.freenode.net.

-- Tim Starling 21:36, 22 April 2007 (UTC)

Tim, that is great, any chance of this on query.php? ^High_InBC^{(Need help? Ask me)} 21:39, 22 April 2007 (UTC)

On further thought, query.php is read only, so it is not the same issue. Great idea for index.php though. ^High_InBC^{(Need help? Ask me)} 21:43, 22 April 2007 (UTC)

The implementation of this functionality is such that it could be easily done with query.php if it made sense. Mike Dillon 22:23, 22 April 2007 (UTC)

I'll start experimenting with this in pyWiki shortly and will hopefully have it committed to the main CVS in not too long. --Cyde Weys 21:43, 22 April 2007 (UTC)

How about making it return 503 Service Unavailable to distinguish it from other servers errors (rare as they may be) and providing an X- header with the current lag? Mike Dillon 21:49, 22 April 2007 (UTC)

Actually, it looks like there is a purpose-specific Retry-After in the HTTP spec. Mike Dillon 21:49, 22 April 2007 (UTC)

OK, I'll make it 503. -- Tim Starling 21:59, 22 April 2007 (UTC)

I point out for the suggestion to pause for at least 5 seconds that bot coders might consider an aggressive backoff such as I've been using. When response time increases, my bots increase their delay-between-request time by more than the response time. The delay time is then decreased slightly by each request which gets a fast reply, so as to gradually approach the desired maximum rate. I'm not giving the exact values because it is best if each bot has different characteristics. (SEWilco 04:01, 23 April 2007 (UTC))

This is an excellent idea. Thank you for implementing it so quickly. --kingboyk 14:46, 23 April 2007 (UTC)

I've put some documentation for this feature here. -- Tim Starling 09:00, 25 April 2007 (UTC)

Bot policy rewrite

OK, I've read this whole page and am pretty much convinced that many people (if not most) are unhappy with the current bot policy and some have lost complete respect for the BAG. So I've created a straw-pole-like thing below. I ask that the BAG doesn't !vote on the first heading ("Too much power for BAG"), because it's really up to the community if they think that we have too much power, not us. Also, if you want to add any more headings feel free. —METS501 (talk) 22:19, 22 April 2007 (UTC)

Too much power for BAG

Comment below (preferably with an explanation) if you agree or disagree that the BAG has too much power (or thinks they have too much power) and that some of this power should be delegated to others or left as responsibilities of bot owners.

• BAG seems to do their job well, though I would like to see a more standardized way to join than just bringing it up on a talk page(something with more community input). ^High_InBC^{(Need help? Ask me)} 23:13, 22 April 2007 (UTC)
  • I agree with that entirely. I don't think BAG has "too much power", and I note that ArbCom seems to be recognising BAG as legitimate in the ArbCom case. However, if the group is to be taken seriously and it's opinions respected it needs a more formal process for membership. I'll toss out a first idea, it can surely be improved on: Perhaps applications could be made at WP:RFA and be closed by a bureacrat? (I'm not suggesting that only admins need apply). --kingboyk 23:17, 22 April 2007 (UTC)
    • The only problem with that is that most of the community is unable to judge whether a person has enough technical understanding to be in the group, and also many people think RfA is failing, so it might not be a good idea to go there... —METS501 (talk) 23:18, 22 April 2007 (UTC)
      • Yes, I know. In fact I've just been over there talking about whether RFB is failing or not (I think it is). Any other ideas? We can't have a situation where it's ok to call members "idiots" without it being a personal attack; that means, we need more rigorous recruitment standards and a more formal assignation of authority. --kingboyk 23:21, 22 April 2007 (UTC)

Approvals process is too strict

Comment below (preferably with an explanation) if you agree or disagree that the bot approvals process is too strict and that there is too much bureaucracy involved in BAG.

• I have never had a problem getting a well articulated project approved. If you follow the instructions then approval is very fast and reasonable. ^High_InBC^{(Need help? Ask me)} 23:11, 22 April 2007 (UTC)
• As an outsider stumbling across this discussion, I agree that the process is too strict. rspeer / ɹəədsɹ 00:26, 23 April 2007 (UTC)

Edit rate

Comment below (preferably with an explanation) if you agree or disagree that for well-established bots (which we can come up with a good definition for) who are running simply takes and have literally zero chance of making a mistake should be allowed to edit at a rate of 50 edits per minute.

• I think edit rates above the standard rate should need special approval, and should be taken on a case by case basis. ^High_InBC^{(Need help? Ask me)} 23:11, 22 April 2007 (UTC)
• I might also support a policy whereas with a broader consensus than a normal bot approval, a bot can run at very high edit rates. —METS501 (talk) 23:14, 22 April 2007 (UTC)

• Likewise. I don't know where the 15ppm rate came from, but if the devs are happy with higher rates in special cases, so am I. --kingboyk 23:19, 22 April 2007 (UTC)

• Lets be clear..., Are we talking about experienced bot operators operating bot flagged bots making safe edits (ones which are clearly uncontroversial and can be easily undone)? If so, I believe the rule should be that the bot should edit slowly (6epm or so) for the first 10 minutes of its run and after that it should be able to operate at whatever speed the experienced bot operator believes is needed to complete the task in a reasonable amount of time up to the limits imposed for technical reasons.. which is on the order of 1-2 edits per second while watching serverlag. As for the bots actual task, thats up to the community and the subject area editors to decide. Anyone running around making lots of bad edits will suffer the communities wrath already, so we don't need to add another layer of approval. Anyone who makes a large scale bot run which is later shown to be non-consensus and whom isn't willing undo the changes will be subject to blocking like anyone else who is harming the project.--Gmaxwell 23:23, 22 April 2007 (UTC)

• There are two issues here: the ability of the WMF software to sustain a very high edit load from one account is not tested. It would make sense to investigate this with the developers. I saw a few very strange bugs recently (deletion of an article that does not go to the log, diff that does not make sense etc). I wonder if it maybe a result of a racing condition on a server. The second is the probability of an error. If something goes wrong and you have to undone 400K edits then it require at least another 400K edits. Even if a single revert is trivial 400K reverts are not. I would suggest first to ensure complete consensus for the concept on WP:VP, WP:AN, WP:BAG, etc. Then, say, 8h of running at 6epm, then 50epm if no objections. I guess 400K edits will pollute the recent changes anyway whether we apply then at 3epm for a year or at 50epm for a month Alex Bakharev 23:46, 22 April 2007 (UTC)
  • What do you think a very high edit rate is? We routinely take editing rates in well excess of what were are talking about. Enwiki takes over 200epm over the entire 4 hour busiest time of day every weekday. The "servers can't handle it" argument is basically FUD and misinformation. :) As for 8h at 6epm thats still around 3k edits, if a problem isn't obvious and doesn't get noticed right away it's probably not going to get noticed until it is well into it in any case. --Gmaxwell 02:18, 23 April 2007 (UTC)

    Well, we can't start handing out 50epm rates to everyone, we have many bots running simultaneously - barring special circumstances, individual bots shouldn't be much higher than 20epm, if that - one bot shouldn't be compared to our peak rates, you're weighing an apple against a cherry tree. Weigh the apple tree against the cherry tree so we know how many edits our bots have to work with, and decide from there. ST47Talk 03:42, 23 April 2007 (UTC)

No, but if everyone used the new maxlag parameter above, that would not be an issue. —— Eagle101 ^{Need help?} 04:06, 23 April 2007 (UTC)

• Personally, I believe any trivial task shouldn't be rate limited, as long as the lack of rate limiting doesn't affect the database performance (and unless the bot is run from toolserver, it can't anyway), once the bot is verified to operate perfectly. Here, trivial tasks include tagging pages, basic moves, fixing links and so on. Of course, ff a bot is going to analyse wikitext with a regexp, tokenize placeholders and mass modify pages, I'd be worried about it running faster than 5 epm. What we need is input from developers during the bot approval process, and preferably some of the people who maintain the WikiMedia technical backbone or MediaWiki. It would be interesting to see who came up with 15epm in the first place - has anyone ever examined the effect of automated HTTP requests on wikimedia server load? --Draicone ^(talk) 13:45, 3 July 2007 (UTC)

Complete rewrite

Comment below (preferably with an explanation) if you agree or disagree that the bot policy needs a complete and total rewrite, and the bot approvals process needs a complete rewrite as well, possibly getting rid of the BAG.

• I have not seem a demonstration that the policy is failing, so any improvement should build and improve on the existing one, I see no fundamental flaw. ^High_InBC^{(Need help? Ask me)} 23:12, 22 April 2007 (UTC)

Return to noticeboard format

From one of the other bag noticeboards... (that I posted at and was suggested to post here), An idea, I'm sure there are others but, lets make this back to what it used to be, a simple noticeboard where people who want to run a bot can post a message. If ther are no complaints in say a day or two, just run the bot.—— Eagle101 ^{Need help?} 22:28, 22 April 2007 (UTC)

• As a bot operator, I'm very tempted by that. As somebody who's seen various hair-brained approval requests (as a BAG member) and heated discussions on the admin boards when bots have gone wrong (as an admin), I'm not. On balance, I think we do need a formal process, alas. --kingboyk 23:23, 22 April 2007 (UTC)

  If people aren't going to notice it listed on a notice board, why are they going to notice in some formal process? Avoid-bot-paranoia. We need to avoid the situation where bot approval is the longest part of the process for making a change on wikipedia. --Gmaxwell 23:25, 22 April 2007 (UTC)

A second note, we need to address why User:Cyde seems to be unwilling to bring tasks for his bot to a WP:BRFA. Finding that reason may help you all in "fixing" this process. I still urge something back to what we used to have. Post tasks, if nobody complains or has questions in a day or so let them run the bot. No need to have "elections" or what not, if you know things from a technical standpoint you would be more then capable of commenting on a standard notice board. —— Eagle101 ^{Need help?} 23:27, 22 April 2007 (UTC)

Maybe Cyde doesn't like process? :) (I'm quite sure he doesn't, actually, but let's wait for him to answer that :))

OK, User X proposes a task on the noticeboard. User Y objects. What happens next?

What about if User X is Cyde, and User Y is a casual editor, not an admin, doesn't have friends in high places, but has some kind of objection which might be reasonably valid? What happens?

What about if Cyde goes ahead anyway? What happens then?

Aren't these things better done within a framework of rules and guidelines, and a transparent process which everybody must follow? --kingboyk 23:34, 22 April 2007 (UTC)

I agree with kingboyk here: A self-determined group is the best way to proceed: Community-wide support isn't going to work for those very reasons, and even though I'm no fan of instruction creep, we need to prevent a washing-out of the process by people who may not be experienced in the procedure. A vote will simply suck - we'll end up with an RfA-like system that doesn't even address the issues that need to be addressed. As for elections into the group, our current system is also superior to votes: those who have been on the 'front lines', so to speak, know what needs to be done and are able to choose those who have shown than they can do it (can you imagine a system in which anyone who writes a featured article can approve bots?). Community-wide elections - while the system does currently allow anyone to comment - seems like the opinions of the BAG would be washed out and overridden - we need a small, mostly self-elected, group with members who know what they're doing and who have the trust of the community - that is the main thing, the community needs to be able to trust that the people were chosen for a reason. ST47Talk 00:47, 23 April 2007 (UTC)

Well, it need not be self-elected, I've tried to get outsiders to look at some of the elections but few bothered. I'd agree that elections are probably better than if it was just like RfA and we need to discuss the issues rather than dissolve and have a huge vote. And we do encourage outside opinions on bots, in fact that often is why it took so long, perhaps we could cut down on that. If a bot is possibly controversial we can wait for outside views for 2 days, posting on AN/I or such. Otherwise, things should be pretty linear and in-and-out, especially for trusted bot operators. Voice-of-All 01:09, 23 April 2007 (UTC)

I think I said self-governing where I meant self-contained, as in, BAG IS the BAG, they are responsible for electing their own - it's more of semantics than anything, but it should be considered to be a BAG process in which the community is involved (Like BRFA - BAG makes decision, community expresses opinion) rather than a community process (Like RfA, where the outcome is directly tied to the set of all users). ST47Talk 01:58, 23 April 2007 (UTC)

I'm not sure whether you are implying this to be a good or bad thing. The point is that the community has given BAG the authority to make these decisions and does not want to be otherwise bothered. There are times when BAG specifically seeks outside community input, but otherwise works autonomously. Community opinion is given significant weight and definitely not ignored. Afterall, BAG has its commission from the community and exists to serve it. It could be a disaster if people were approving bots who didn't understand the implications. My favorite example is the automated spell check bot. If you think it is a good idea that can be properly implemented, chances are you should not be approving bots. -- RM 13:47, 23 April 2007 (UTC)

• I prefer a notice board to a formal approval process with designated approvers. If there is any more process it should be reserved for obtaining bot-flag. Once someone has proven themselves to be generally trustworthy with the bot bit, we should give them enough rope to accomplish whatever they want to accomplish. Sometimes that will also be enough rope to hang themselves, but thats their problem, not ours. The community is able to police the actions of editors, automated ones are no exception. --Gmaxwell 23:30, 22 April 2007 (UTC)

  I think you make a good point; I could be swayed by that. What would you propose for the initial process though? (The process whereby a bot flag is procured). --kingboyk 23:44, 22 April 2007 (UTC)

Simple, a post to the B-crat noticeboard, with evidence of why you need the flag and that community accepts the tasking. Do we really need anything more then that? —— Eagle101 ^{Need help?} 23:46, 22 April 2007 (UTC)

Do bureaucrats want that task? Are they capable of it? Do we even have enough bureaucrats? I don't know if you're aware or not, but it's just about impossible to get anybody promoted to the role at the moment. --kingboyk 23:48, 22 April 2007 (UTC)

There's an issue here that selection of the BAG gets very little community input; RFB, if one were being cynical, could be said to suffer from getting too much. One possible approach would be to split up the current BC roles at the technical level (if the devs wouldn't mind), and have them be requested separately (if the community'd be OK with that), so at least the different trust (and "plaform", and what have you) issues can be factored out a bit more. Alai 00:06, 23 April 2007 (UTC)

King, bcrats already have to flag bots anyway, that simple load won't increase. —— Eagle101 ^{Need help?} 00:56, 23 April 2007 (UTC)

Not as measured by number of clicks on the BC tools, no. But they'd then have the (sole, as opposed to in theory joint with the BAG, and in practice rather secondary) responsibility of questioning the operator to determine their competence and suitability, assessing consensus for the task, etc. Alai 01:02, 23 April 2007 (UTC)

Right. At the moment, all the crat does is check that a BAG member gave the approval, and click the button (or whatever they do to add a flag). --kingboyk 01:05, 23 April 2007 (UTC)

Right, but its not that hard for the crat to look and see... bot needs to edit fast, here is proof of the tasking, and nobody complained at the board after XX days (say 5). Clickety click, approved for the flag. This is what it was before BAG came in and attempted to simplify things. (around when essjay started this) Also its not like bots get flags every day or anything. —— Eagle101 ^{Need help?} 01:09, 23 April 2007 (UTC)

I think that "no one complained after X days" is a pretty decent model, in fact (though not infallible, a trait it'd have in common with all others). Faster approval for genuinely simples cases can be handled by acclaim, invocation of precedent, or by explicit approval by designated individuals (whether that be a BC or BAG (or equivalent). It's desirable that a reasonable number of eyes see any given request, though, and ask the obvious (or less obvious) questions, flag any gotchas that a single-point-of-failure might miss, etc. Alai 02:24, 23 April 2007 (UTC)

(undent) I'm not saying that going back to the board style is perfect, but it would put the onus for good bot operations on the bot operator, rather then an authorization group. Really either way is fine with me, and I will continue to comment on new bots in whatever format that is chosen. —— Eagle101 ^{Need help?} 12:40, 23 April 2007 (UTC)

The onus has always been on the operator to perform good actions. The BAG is there to verfiy it and in the case of new bots there is theoretically a bureaucrat to perform another verification step. -- RM 14:22, 23 April 2007 (UTC)

How the French do it. --kingboyk 13:10, 27 April 2007 (UTC)

Simple suggestion from an Insider.

I was part of BAG and I think that BAG or some form of simi-tight control is needed. I have seen some completely ass-backward bot request for Approval/Bot request. On the other hand it is a real pain in the ass as a bot op to come back for every task. I propose a idea. Bot op's MUST apply for their first approval (and get bot flag). once a bot and operator is trusted let them do their work and any other similar work. if a task is a major change in what it does have the bot op leave a note on the BRFA talk and if there are any issue file a BrfA otherwise let the bot op work. (Keep It Simple, Stupid) the motto of all bot writers Betacommand ^{(talk • contribs • Bot)} 00:01, 23 April 2007 (UTC) (I will crosspost this)

I agree with that totally. It's the bot flagging which I think needs scrutiny; we should then trust flagged ops to do a responsible job, or have a self-policing noticeboard for new tasks. --kingboyk 00:11, 23 April 2007 (UTC)

Why is the oversight even needed, B-crats already have to flag bots, so that won't change. (in reply to above section). Guys lets remember that this used to be a simple noticeboard, you posted your task here, waited a day or so, and then provided nobody found any problems, ran the task. There is no need for oversight, thats every admins job (not all of them do focus here, but out of 1200 admins enough do). If you are qualified to be a "bag" member, then you are more then qualified to raise issues at a simple noticeboard. —— Eagle101 ^{Need help?} 00:55, 23 April 2007 (UTC)

Eagle: B-crats flag bots for technical reasons, they are the ones who can. While in RfA the bcrats determine an outcome and there is no group other than them who is tasked with closing RfAs, in this case those powers are split: bots aren't as powerful as admins, it's a community task to run a bot, so we don't need the busy bcrats to need to handle and process all these requests, it can be done by a group of trusted users. I don't see anything wrong with the process or with the people performing it, so let's just leave it alone. ST47Talk 02:03, 23 April 2007 (UTC)

B-crats don't see the majority of the requests now anyway, and under the proposal below they won't see the majority of the requests should bag disappear. Do remember that the community never really discussed bag in the first place ;) —— Eagle101 ^{Need help?} 02:28, 23 April 2007 (UTC)

Is there any need to try and score points? "Do remember that the community never really discussed bag in the first place" What? Policy develops organically on wiki. People have been free to comment here whenever they like, too. I'm also told that only a few days ago you were asking about joining BAG! Let's keep the discussion on target and discuss rather than attempt to impress our IRC friends, alright? --kingboyk 10:49, 23 April 2007 (UTC)

Ah nah, that point is irrelevant, but what about the others that I have raised? I don't think switching formats would matter all that much, at least from the standpoint of flagging bots with a bot flag. B-crats don't seem to care about non bot flagged bots anyway. Its just an impression, and just so you know I do agree with the original block, but I also wonder if there is a better way then bag. —— Eagle101 ^{Need help?} 12:37, 23 April 2007 (UTC)

Thanks for clearing that up, I appreciate it. I know where you're coming from, and deciding if there's a better way is why we're in this discussion after all :) Cheers. --kingboyk 14:35, 23 April 2007 (UTC)

Bag alternative

Well here it is, I dug out a copy of the old noticeboard out of WP:BRFA. This is what we did before BAG came into place. As bag never had a formal !vote, I suggest that we comment on this alternitive method and perhaps adopt it. I don't recall any problems under the old system. Check out my sandbox for the header, and see if you can think of any decent modifications User:Eagle_101/Sandbox/4. I'd like to emphasize that the community never really approved of bag to start with vie any large community discussion, so I'd like to see a justification of why the older noticeboard format is invalid, and why this current format of BAG is better if that is the case. :) —— Eagle101 ^{Need help?} 01:34, 23 April 2007 (UTC)

Err well that one actually is part of the BAG as it started, but it looks like User:Betacommand has modified it so that mentions of BAG are not there. Again this is an alternitive to what we do now, I welcome any suggestions, and hope that this at least provokes thought. What is really needed when it comes to bots? Any bag member can comment on bots at a simple noticeboard, and heck I've been commenting on bots under this current format anyway. This is a community discussion, we must not forget that. The point of this is not to see "will it get approved" its "will the community accept my task?". —— Eagle101 ^{Need help?} 01:49, 23 April 2007 (UTC)

With all this talk about a "BAG alternative", I thought I should weigh in, but I have not yet had time to read anything of the above discussion. Let me first post by saying that I've written on Wikipedia talk:Bots/Approvals group a little about the history of BAG and why I think it has always had consensus. I'll read the rest of the discussion and post more. -- RM 13:03, 23 April 2007 (UTC)

RM, I appreciate the history bit, there were things in there I did not know. I'm starting to think that this process can be improved... at least for existing bot operators. There was nothing wrong with the task that User:Cyde chose to run, but the bot was blocked as it was going at 50 edits per minute without prior notification. So... that makes me wonder why did Cyde not notify BAG, was it too time consuming to draft up a BRFA? I'm wondering if for experianced operators they could simply post to a noticeboard like we did in the old days... Again just food for thought :) —— Eagle101 ^{Need help?} 13:08, 23 April 2007 (UTC)

Historically speaking, I've always been in favor of letting bot operators do whatever they feel like doing without any approval at all. Cyde just did what I would have done a couple years ago. My bot runs consisted of approximately 30,000 edits at a time and editing as quickly as possible was the only way to perform the tasks in a reasonable amount of time. Having a bot run for, say, a month really is just too long. I agree that the process could be changed or improved, but I don't think there is an overwhelming need to rush to a change. I've read the comments from the dev above regarding edit speed and our policy should reflect those thoughts and instructions as appropriate. I would agree that the approvals process is currently broken due to lack of people approving. It takes too long. But more members would solve that problem. Requests for new tasks really doesn't otherwise take that long and we allow established existing bot operators to ignore the process for certain urgent tasks. I like to think of it this way: putting a bot through approvals not only shows good will but shows that the operator wants to get constructive input to make sure that things get done right. -- RM 13:36, 23 April 2007 (UTC)

Problem - timeliness of repsonse

THe old system suffered from requests that did not get answered at all, the new system still has overly long delays. The principles of both are sound

1. disclosure of intent
2. testing
3. approval

These protect the wiki, the community and the bot owner. However we could perhaps find a more active solution by increasing the numbers in BAG, we can't expect those few meebers to be constantly available, I would volunteer, and I'm sure there are others with experience. How they are appointed doesn't really matter that much, as the process of task approval is fairly objective (but perhaps could be made clearer). Rich Farmbrough, 13:10 23 April 2007 (GMT).

I was also thinking about this overnight, that perhaps the task approval process could be made more objective. e.g. Commnunity consensus? Yes, check. No, go get it and come back. Operator has a reasonable contribs history/block log? Check. --kingboyk 14:44, 23 April 2007 (UTC)

Interesting, I never really thought of it this way... in any case I'm wondering if it would be a decent idea to ask Cyde why he did not bother with the BRFA before running the task. As I mentioned above there was nothing wrong with the task that he chose to run. But it got blocked for reasons of high edit rate, something that would not have happened had an announcement of intent (through bag or a noticeboard whatever) had occurred. In either case the block was correct per our existing WP:BOT. —— Eagle101 ^{Need help?} 13:16, 23 April 2007 (UTC)

You're absolutely right. He only needed to notify us. We're not empire builders, and we have always acted in what we perceived to be the best interests of the wiki. There was nothing wrong with the task (I personally am a bit skeptical about it but would have approved because of clear consensus). There's nothing wrong with the edit rate if a dev has okayed it. What was wrong was going ahead without telling anybody on wiki, and then blaming the uninformed parties for questioning it. --kingboyk 14:44, 23 April 2007 (UTC)

The major problem I see with bot approvals is the speed because there are too few members. Having a productive member resign over this is only going to make the problem worse. I think the process works really well when there are enough people working on the process and there isn't a backlog. Only in the last few months have has the backlog grown substantially. I used to approve about 60% of all bot requests before I became less active for reasons off-wiki. We need more members in a way that we did not in the past. That would help solve most of our problems. Finding qualified individuals is complicated because approving bots generally requires broad policy understanding to approve bots correctly. If someone wasn't paying adequate attention to policy they might, for example, approve an automated spell checking bot which would be VERY BAD. As objective as they are, if you don't have a good policy understanding, you'll make bad approvals. I think that Rich Farmbrough should nominate himself on Wikipedia talk:Bots/Approvals group. I'll support the nomination. -- RM 13:28, 23 April 2007 (UTC)

As I see it, BRFAs go through 2 stages, questioning and testing, then most are approved. The first stage should only take a day or two, and then the bot is able to run. Perhaps I can throw something together with BAGBot that can contact BAG members who choose to be contacted under certain situations - such as after the trial or when one of the {{BAGAssistanceNeeded}} templates is placed on the page - or something. Also, when and if my nomination passes - I nominated to try to speed things up - I will help with that. The LAST thing we need is people resigning, and the BEST way to speed up the questioning phase and ensure that what needs to be asked is asked is community input. I think that most of the community knows about BRFA, less so about this page, and most of them either don't care, don't think they have the knowledge, or don't know they're allowed do comment. We regularly have 100+ people comment on RfA, but if someone posts one line on an admin's talk page, the admin doesn't blank every page he edits. If a bot has a line that says $newtext="";, then we have a bit of a problem, no? (I'm not saying that bots are more important than admins, but if someone watches WP:RFA and the Village Pump, BRFA is the next logical place.) ST47Talk 14:34, 23 April 2007 (UTC)

Just to clear one thing up RM: my resignation is until such time as community support for the process and it's mandate is clear and unambiguous, at which point (be it next week, next year, or next decade) I may well stand again if I have the spare time. I have specifically ruled out just adding myself back to the list, though.

I certainly agree BAG (if it continues) needs more members. I think we need to make a decision about how those members are appointed, however.

I would certainly approve of Rich joining the group, and tentatively (as I haven't researched the guy yet) Eagle101 too. Indeed, anybody with a modicum of common sense and a knowledge of bot operations is welcome to apply and will likely get my support. --kingboyk 14:40, 23 April 2007 (UTC)

I don't get it, I have gotten approval for several bots, and they all were very fast. If you want approval faster, get all your ducks lined up in a row before making your request. State exactly what it does, providing source code helps, get consensus ahead of time and provide a link. Approval is most often delayed due to people not reading the instructions before asking for approval. ^High_InBC^{(Need help? Ask me)} 14:47, 23 April 2007 (UTC)

All my approvals have been really fast except for one comtroversial one which ended up being a bad idea anyway. ST47Talk 14:53, 23 April 2007 (UTC)

Which one was that? Birthday Bot or something else? --kingboyk 14:54, 23 April 2007 (UTC)

I was referring to one which essentially cleaned up internal links - foos to foos and other stuff that doesn't need doing. ST47Talk 15:03, 23 April 2007 (UTC)

Ok, well while we are all guessing... should someone ask Cyde what he feels is wrong with the BAG approval process? I mean this whole thing started because of a bot running very fast without on wiki notification after all. I really do think that Cyde might be able to bring another very good point of view to this discussion. There has to be some way to make it so the bot operators don't "skip" this, whatever format it ends up in. —— Eagle101 ^{Need help?} 15:02, 23 April 2007 (UTC)

Cyde is (without meaning to offend, and I'm quite sure he wouldn't take offence at this) a bit of a renegade, a bit of a shoot first and ask questions later kind of character :) That's cool, and ordinarily WP:IAR applies to actions here, but I think there has to be a limit to how far IAR applies to bots. No harm in asking him though; on the contrary I'd like to see some input from him here as he's conspicuous by his absence :) --kingboyk 15:07, 23 April 2007 (UTC)

In terms of approval, the process seeks to establish several things, I think:

1. No technical harm
2. No content harm
3. No community harm
4. Some measure of improvement however tiny.

There are things we look for which we see as certificates of the above, but we don't necessarily need the same set of certificates in each case.

1. Approver knowledge of WP (including the community)
2. Comments by third parties
3. Requests for the run by projects, or editors
4. Results of test runs
5. Users previous record

Clearly factors which "worry" us and make us look for higher standards of proof include number of edits, edit speed, difficulty of reversing, near "dangerous ground", maverick operators.

But really, absent other contraindications, a small test run, followed by progressively larger runs should flush out problems of most types without causing significant or irreparable problems.

Just thinking outloud... Rich Farmbrough, 16:20 23 April 2007 (GMT).

edit rate

(undent) Also not to get picky or anything but I'm digging for where this bot started editing... and the only way to figure that out is to click "previous 500" about 100 times or so... I'm still going, but I noticed this section of the bots diffs where the bot is going at a speed of 71 edits per minute, over the developer set criteria of slower then one operations's pause or 1 edit per second, whichever is slower. See ( [7] Notice on this one there is approx 6 minutes of edits in here, which lends to an edit rate of ~83. [8], ~7 minutes [9], ~7 minutes [10], ~7 minutes [11], ~7 minutes)

Do note that those are 2500 consecutive edits, for a total time of about 38 minutes in which it was going in excess of the developer set speeds. (See the approval by gmaxwell here). To qoute him "I approved it editing at a rate of one operation interval sleep per operation or 1 edit per second, whichever is slower".

Please do notice I was not really looking for this, just happened to notice it while digging for the start time. Probably now its a moot point, but these speeds were in excess of the developer set amounts that were reported on wiki. Note I'm still looking for that elusive start time! :) —— Eagle101 ^{Need help?} 15:24, 23 April 2007 (UTC)

Not to be picky or anything, but why say this here? This section is about bot approvals being too slow, not cyde being too fast. ST47Talk 15:29, 23 April 2007 (UTC)

Yeah yeah I know, I was too lazy to make a new section, feel free to refactor I will refactor. In any case this bot was running from 15:40, April 19, 2007 (hist) (diff) m Image:1943 arcade.png (Replacing template name per Wikipedia:Non-free content/templates.). So its been going for about 4 days now, Xaosflux only noticed it on day 3 of the tasking. Of course nobody could have known the start time without digging it out like I did :), but this may or may not be useful. (See: for the starting contribs [12])

—— Eagle101 ^{Need help?} 15:43, 23 April 2007 (UTC)

I don't see why it matters when it was spotted? We don't sit with recent changes refreshing all the time (actually, I often watch recent changes to Talk: to keep an eye on usage of my plugin but that's a different issue). We have WP:ANI, WP:AIV etc to keep an eye on these things. --kingboyk 15:54, 23 April 2007 (UTC)

Tangential to the main point, once you have gone back to "older edits" once, you can see a clearly encoded date and time in the URl, which you can modify to your hearts content. Rich Farmbrough, 16:08 23 April 2007 (GMT).

Figures, I just happened to notice it and was not sure if it would be relevant to the discussion now or not. —— Eagle101 ^{Need help?} 16:16, 23 April 2007 (UTC)

Manually approved edits and welcome bot

User:Matthew Yeager is interested in running a bot-assisted welcome campaign (but he insists the edits are manually approved, so no bot approval is needed). The bot account was blocked and the blocking admin asked for comments here. The people watching this noticeboard might be in a better position to comment. CMummert · talk 02:28, 26 April 2007 (UTC)

Isn't a welcome bot the perpetually mentioned archetype of a really, really bad bot idea? --Cyde Weys 03:29, 26 April 2007 (UTC)

If it's a bot. If the user were to do it under his own account, and somehow review the contribs, I would not be opposed to a script-assisted welcomer. ST47Talk 03:58, 26 April 2007 (UTC)

That's the complicating factor. It's apparently some AWB-like setup where he approves a large set of messages and then the script goes through and commits them. I don't know how things like that have historically been treated. CMummert · talk 04:02, 26 April 2007 (UTC)

Resolution has been reached, thank you all for your input. Matthew Yeager 04:43, 26 April 2007 (UTC)

Well, more comments are always better. The suggested resolution is to run the setup with a non-bot name and a slow edit rate. CMummert · talk 04:48, 26 April 2007 (UTC)

I used to do some rather heavy welcoming. I made sure for every person I welcomed I:

1. Signed/edited with my own name/account
2. Put their page on my watchlist
3. Reviewed their existing contributions
4. Responded to any vandalism
5. Gave advice regarding any mistakes
6. Gave encouragement for being useful
7. Answered the questions they leave on my talk page afterwards

As long as you don't skip these things I don't see why any reason why you cannot welcome 200 people a day. ^High_InBC^{(Need help? Ask me)} 04:59, 26 April 2007 (UTC)

Policy rewrite proposal

I have started a (hopefully insightful) discussion on the possible rewrite of our bot policy here. Comments are welcome. Миша13 09:51, 28 April 2007 (UTC)

Transmitting in UTF-8

I am having some trouble with UTF-8 encoding with my bot. I am trying to write text to wikipedia that contains UTF-8 characters (pulled from wikipedia in the first place), but am having inconsistent results – some characters are translating properly, whereas others are not (and are coming across as ?'s). All of the characters look proper in my browser on the original page, so it's not a browser issue.

I am using PHP. I tried calling utf8_decode() on the incoming text, and utf8_encode() on the outgoing text, which lead to the condition described above where only some of the characters come across properly. I'm not sure what I'm missing or what I can do to fix it; any help would be greatly appreciated.

For an example of the problem, see this diff. The left side is what it is supposed to look like (manually inserted), and the right is what the bot is putting. Note that while the à (in "U.S. Città di Palermo") is being retained, ā, ō, and ū are turned into ?'s. —Daniel Vandersluis^(talk) 15:29, 1 May 2007 (UTC)

Never mind, I figured it out with help from the Village post (technical). Thanks anyways —Daniel Vandersluis^(talk) 16:29, 1 May 2007 (UTC)

Cydebot's new task of GFDL standardization

Just a heads-up, Cydebot is now taking on the large task of Wikipedia:GFDL standardization at a rate of one edit per second. I should be able to get through all of this in two days, maximum. So don't worry, nothing's gone wrong, and I'll be able to handle all of it soon enough. This is one of those wiki-wide tasks where it's helpful if all of the edits came from a single account, for tracking purposes. --Cyde Weys 03:52, 7 May 2007 (UTC)

Sounds good, and thanks for alerting the community beforehand :-). I'll trust that there's consensus somewhere. Also, don't forget the #Maxlag parameter. —METS501 (talk) 04:32, 7 May 2007 (UTC)

Roomba to run again.

Just FYI: I haven't run roomba for a long time due to the lack of a suitable live slave database to pull its input data from. I've fixed that issue tonight, so I'll soon be running it again to tag all the non-free images which are not used in the main namespace as orphaned. This is the first step in the deletion process for non-used non-free images. This is the same task that roomba has always done but since it has been so long since I've done it roughly 7.5% (26k) of our non-free images will need to be tagged. It won't be running at high speed or anything like that, though I check maxlag. --Gmaxwell 04:47, 7 May 2007 (UTC)

Ah, betacommand tells me that he's finally got around to coding similar functionality. I'm always glad to let someone else take heat, so I'll leave it to him for now. --Gmaxwell 05:16, 7 May 2007 (UTC)

Children of Curpsbot

Please see Wikipedia:Administrators'_noticeboard/Incidents#Children_of_Curpsbot regarding the secret use of adminbots. Dragons flight 03:58, 8 May 2007 (UTC)

Logins seems to stop working in Perlwikipedia

It looks like logins stopped working in Perlwikipedia (probably because of latest changes in security).

In sub _get

when doing

my $res = $self->{mech}->get($url);

where $url is http://en.wikipedia.org/w/index.php?title=Special%3AUserlogin&action=edit It gives 403 Forbidden.

Any ideas how to make it work?

Alex Bakharev 10:33, 9 May 2007 (UTC)

The devs have blocked the Perlwikipedia user-agent. A dev (User:Simetrical) tells me it could be because it doesn't contain descriptive contact information, so I've changed mine to 'Bot/WP/EN/ST47/BotName', as that gives necessary identification and contact information. ST47Talk 10:43, 9 May 2007 (UTC)

Sorry, I have to change

$self->{mech}->agent("Perlwikipedia/$VERSION")

to

$self->{mech}->agent("Bot/....)?

Is this correct? Alex Bakharev 11:05, 9 May 2007 (UTC)

It works! Thank you very much! You are great! Alex Bakharev 11:09, 9 May 2007 (UTC)

LDBot creating pages from....where?

Note: I asked this on lightdarkness's talk page, but he does not edit often anymore so I am crossposting here.

Where is LDBot pulling the source for the next AFD page from? It doesn't appear to be importing Wikipedia:Articles for deletion/current, because these edits occurred hours before this page creation. Please advise. -- nae'blis 15:42, 11 May 2007 (UTC)

It might just have memorized what the correct layout of the page is, like Sandbot. --ais523 16:09, 11 May 2007 (UTC)

Oof, that seems like a shortsighted maneuver when the bot runs automatically. -- nae'blis 17:46, 11 May 2007 (UTC)

CAPTCHA

What is exactly sets off a CAPTCHA requirement to log in? Sometimes I log in no problem, but then suddenly my bot is being challenged with a CAPTCHA even though it never got the pass wrong.

Does a single wrong password make it so you need to pass a CAPTCHA for X minutes? Is there no way bots can be exempt from this? ^High_InBC^{(Need help? Ask me)} 18:52, 11 May 2007 (UTC)

just log in with the web interface and you'll fix the CAPTCHA. Betacommand ^{(talk • contribs • Bot)} 18:54, 11 May 2007 (UTC)

That is what I thought, but I found I was still being challenged after a successful log in(just for a few minutes), so I am thinking now there is a timer. ^High_InBC^{(Need help? Ask me)} 23:11, 11 May 2007 (UTC)

You have to confirm your e-mail address. Then it shouldn't bug you anymore. – Quadell ^{(talk) (random)} 16:24, 24 May 2007 (UTC)

Incidentally, either I'm overlooking something or e-mail confirmation is down right now. When logged in as my bot User:polbot, I tried confirming my e-mail, but I never got an e-mail. I left a message at Help talk:Email confirmation, but I don't know if anyone ever looks there. Does anybody know what's up with this? – Quadell ^{(talk) (random)} 16:28, 24 May 2007 (UTC)

Exclusion code for User talk pages

From looking through some of my bot's edits and the edits of other bots that primarily issue notifications on user talk pages, I every so often see a user removing these notices. Sometimes they even put in their edit comment something like: "remove unwanted bot notices AGAIN". I know that a few bots already have their own exclusion codes, but I think it would be convenient if a standard way is devised to exclude certain, or all bots, from user talk pages. This would only apply to bots issuing notices, such as duplicate image warnings, image copyright warnings, speedy deletion warnings, etc. Bots that are warning a user to stop vandalizing or something of that nature should probably ignore any exclusion codes for the most part. My proposal for the exclusion code is a simple HTML comment of the form:

<!--Diallow:BotName-->

and to disallow all bots:

<!--Diallow:ALL-->

How does this idea sound? I created a topic on this here but this is probably a better place to have the discussion. --Android Mouse 17:01, 21 May 2007 (UTC)

{{bots}} is what your looking for but not all bots are compliant. (mine is not on purpose) Betacommand ^{(talk • contribs • Bot)} 17:24, 21 May 2007 (UTC)

Thanks, I'll make mine compatible when I get the time. But why are there so little bots compliant? Lack of publicity to this template? How would a user even find out about this? --Android Mouse 17:40, 21 May 2007 (UTC)

most bots need to leave their messages, Mine has to per policy. Betacommand ^{(talk • contribs • Bot)} 17:46, 21 May 2007 (UTC)

Not wanting a bot to post a message to your page is often not reason enough to not allow for the posting that message. ^(H) 17:57, 21 May 2007 (UTC)

You have a point, but how do you distinguish what should be disallowable and what should not? --Android Mouse 21:40, 21 May 2007 (UTC)

I would say that messages intended to help the user should be disallowed, while warning and similar messages telling the user something should not. For example, a warning about an orphaned fairuse image is not necessary, as the user is not mandated to take any action. On the other hand, an antivandalbot message shouldn't be disallowed. Tizio 16:21, 22 May 2007 (UTC)

That seems pretty reasonable. Anything where the party MUST be notified cannot be blocked, but most bots should allow users to opt out through {{nobots}}. -- nae'blis 16:24, 22 May 2007 (UTC)

That is true, H, but it's reason enough to remove it. Gracenotes^T § 16:11, 13 June 2007 (UTC)

Sandbot

Sandbot (talk · contribs) appears to have ceased editing. The last time that this happened, I enabled sandbox raking in User:Uncle G's 'bot (which rakes sandboxes on several projects). I have done so again. Uncle G 15:52, 22 May 2007 (UTC)

Interwiki question

Hi, are there any bots active doing a full namespace sweep starting from this wikipedia? I noticed it took quite some time (about 3 months) for an interwiki link to propagate from the English wikipedia to the Dutch wikipedia. multichill 10:19, 30 May 2007 (UTC)

As it takes about two weeks to do the 300k articles in nl.wp, it will take a little less than three months to complete the whole en.wp namespace. I guess it's frustrating for the interwiki namespace count, so that no one actually tries to complete a full en.wp namespace sweep. Will you be one of the first to try? :) Siebrand 10:23, 30 May 2007 (UTC)

I could give the category namespace a try to see how much new links i encounter. multichill 10:49, 30 May 2007 (UTC)

Server Issues

We just found out that we're having some server issues. If you run a non-essential bot (ie: If it's not anti-vandalism, or working with AIV, etc), please take it offline for now to help. Archiving, etc can wait a few hours. ^demon^{[omg plz]} 02:07, 4 June 2007 (UTC)

Alright, I think everything's back to normal. —METS501 (talk) 02:37, 4 June 2007 (UTC)

the servers have been fixed, but lets give the servers ~5 hours to get back to normal. Betacommand ^{(talk • contribs • Bot)} 02:39, 4 June 2007 (UTC)

Preemptive open sourcing

I've seen a few bots disappear over time because their owner had to take a wikibreak for whatever reason. Generally this means to other users that some expected functionality no longer works. It would be kind of neat if, when bot X drops out, some other bot takes over its functionality; in order to make this work best, it would be nice if bot owners posted the source code to their bot somewhere on wiki. What do people think of this? >Radiant< 08:17, 19 June 2007 (UTC)

I think it's a fine idea; I do it myself. Example: User:MadmanBot/Source/Task2. It would also help new bot programmers learn some of the basic techniques. — Madman bum and angel (talk – desk) 08:26, 19 June 2007 (UTC)

I do this as well. At Wikipedia:Bot_policy#Good_form, our policy suggests that bot-owners "Publish the source code of their bot (unless it's a clone)". – Quadell ^{(talk) (random)} 10:39, 19 June 2007 (UTC)

Does the BAG ask for a bot source before approving it? >Radiant< 11:52, 19 June 2007 (UTC)

Not generally. But if the user is inexperienced, and doing something complex, regex's and such may be requested. Reedy Boy 12:28, 19 June 2007 (UTC)

There are various options for this. Here you can find a wiki dedicated to the development of bots that serve MediaWiki and its users. Once a bot has been found to be stable there, and if it was written in Python, we/I usually commit it to the pywikipedia bot framework, as was recently done with welcome.py, imagecopy.py and several other tools. Proper documentation and readable code is a requirement before code is committed. Please participate there if you require feedback on your bot code and if you would eventually like to have your bot code committed to the SourceForge project pywikipediabot. Bots in all languages are welcome, although we currently only have a formal project for bots in Python. As a last resort: each toolserver user can commit code to a publically available Subversion respository. You could ask any tollserver user to add your code to it, so that you would have a stable distribution platform for it. Cheers! Siebrand 12:35, 19 June 2007 (UTC)

I dont openly post my code because I know in the wrong hands it can be very dangerous. But if a trusted user would like the code I will give it to them, under the condition that they dont pass the code out. Betacommand ^{(talk • contribs • Bot)} 15:29, 19 June 2007 (UTC)

Half the code for my bot Bot523 (the bit that does the editing) is available on the wiki as User:Bot523/monobook.js (one of the advantages of a JavaScript bot). The other half is nonportable and a bit of a mess (it generates a shortlist of pages that might need editing and opens the relevant page so that the monobook.js side can edit it if needed), so I haven't published it online, but would email it on a good-faith request. --ais523 16:32, 19 June 2007 (UTC)

Are there any essential daily/weekly bots that aren't open-source, where if the operator (God forbid) were to die suddenly, we'd be up a creek? – Quadell ^{(talk) (random)} 16:39, 19 June 2007 (UTC)

Well, the loss of HagermanBot has made an impact, and it seems he did not publish his source code; I believe E is trying to get in touch with him now. As I understand it, the code for MartinBot is available... is the code for the MiszaBots and the AIV helperbots available? Most of them seem to be available on demand, and not published, which doesn't solve the problem in your hypothetical situation. — Madman bum and angel (talk – desk) 17:10, 19 June 2007 (UTC)

I plan to do a re-write of MartinBot at some point, and will make that open-ish source (ie - access will be freely available to those who request it, and who have a positive history). I need to get the source of RefDeskBot out there too soon, and MartinBotIV perhaps? Anyway - if anyone ever wants source, I am usually happy to pass it on when I have the time to clean it up. Martinp23 19:40, 19 June 2007 (UTC)

BAG Joining

Hey, I have been asked to post a notification of my request to join the Bot Approvals Group on here. It can be found at Wikipedia talk:Bots/Approvals group#Joining. Thanks! Matt/TheFearow (Talk) (Contribs) (Bot) 02:08, 22 June 2007 (UTC)

Oh good. We need more BAGgers. – Quadell ^{(talk) (random)} 03:10, 22 June 2007 (UTC)

Python help

I'm interested in creating a bot called SpebiBot, but I can hardly understand Python. The bot's tasks are to substitute unsigned and vandalism warning templates. This doesn't seem like such a complicated idea, however, if you can help out, any help will be much appreciated. Thanks, –Sebi ~ 02:11, 23 June 2007 (UTC)

I can write this, But it will be a few days. Betacommand ^{(talk • contribs • Bot)} 02:14, 23 June 2007 (UTC)

If you don't know python and want to learn programming, then I'd recommend you learn C instead. It seperates the real men from the boys :p --Android Mouse 02:18, 23 June 2007 (UTC)

No, that's puberty. Mike Dillon 02:32, 23 June 2007 (UTC)

yep the boys use C and men use python. I use the m:Pywikipedia bot framework Betacommand ^{(talk • contribs • Bot)} 02:26, 23 June 2007 (UTC)

A good language for learning programming in general may or may not be the language best suited to do some specific task. So in a sense you're both right. Gimmetrow 02:59, 23 June 2007 (UTC)

Thanks for your help :) –Sebi ~ 03:15, 23 June 2007 (UTC)

Edit token structure changed

The edit token string now includes a plus sign near the end; that is, "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+\" instead of "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa\". This means that the edit token must be encoded when sent to the server in a POST request, whereas previously an edit would have worked without encoding. If not encoded (+ → %2B), the edit will not be made due to "loss of session data". Gracenotes^T § 22:08, 1 July 2007 (UTC)

Ah, I see that. I encoded it anyway, but thanks for the heads-up! I wonder why the change and why there wasn't any discussion that I could see. — Madman bum and angel (talk – desk) 03:13, 2 July 2007 (UTC)

Ah!!! That must be it!! I've been killing myself trying to figure out why I keep getting a damned loss of session data message on every edit! Thanks. —METS501 (talk) 03:15, 2 July 2007 (UTC)

Yes, thanks for the notice. —Remember the dot ^(talk) 04:12, 4 July 2007 (UTC)

It was changed here, Madman. (In rev:23287, which also had a host of other changes.) However, Tim Starling (talk · contribs) also added a more specific error message. Gracenotes^T § 18:40, 3 July 2007 (UTC)

BAG

Wikipedia_talk:Bots/Approvals_group#BAG_Joining. Thanks! ~ Wikihermit 05:32, 4 July 2007 (UTC)

Interwiki bots forcing unwanted links

For some time now I've had problems with various interwiki.py bots repeatedly adding links to pages where local human editors consistently remove them. I find the bots are currently poorly designed to handle such a situation. Bots should defer to humans, always, so there should be an easy way for local editors to tell the bot not to touch a certain link. On meta:Talk:Interwiki.py and on the pywikipedia mailing list I suggested to the developers a simple solution: make the bot check whether a given link has been commented out on a page; if it has, leave it alone. Unfortunately I got zero response on either venue.

The case that has triggered this is at Ingria, where bots have been adding a link to the highly contentious "ru-sib" wiki, to which local editors strongly object. Please look at the page history to see what's going on: an army of bots stubbornly revert-warring against a consensus of local human editors. -- Let me make it very clear, though, that the issue is really independent of this particular political conflict, and I'm not going to discuss here the merits of whether or not one should link to ru-sib or not, or whether editors should be entitled to "boycott" links to a contentious page in this way. There can be any number of other legitimate reasons why an editor might conclude a certain link that interwiki.py considers a good match is inappropriate. Errors made on another wiki proliferating; mismatches in how topics are distributed across pages; disagreements about which of several pages is the better link target; whatever. The matter is simply that in each such case it should be up to humans to work out such a conflict, it shouldn't be settled by bots mechanically revert-warring.

On Ingria, it's gone so far that I've become exasperated enough to threaten bots with blocks, and actually blocked one bot today (made the bot owner mighty angry, which I can understand to some degree...) I know that's a poor solution. I posted previously on several other fora and got no response at all. Can somebody advise how to solve this? (For the time being, I implore bot owners to run their bots with "-neverlink ru-sib", until this has been technically worked out.) Fut.Perf. ☼ 19:47, 12 July 2007 (UTC)

There are the {{bots}} and {{nobots}} templates, which were made for that. You could just put the correct template on the page that shouldn't be modified, and then if a bot modifies it anyway, you can ask the bot operator to modify the bot so that it is exclusion compliant. – Quadell ^{(talk) (random)} 20:45, 12 July 2007 (UTC)

Thanks for the suggestion. {{nobots}} was tried, but the pywikipedia bots apparently won't honour it ([13], [14]). And, really, it would be a poor solution too. It's not that people want to keep all bots out of the whole page. They want a specific type of bot to stop doing one specific edit. Fut.Perf. ☼ 20:56, 12 July 2007 (UTC)

You can use {{bots|allow=<botlist>}} or {{bots|deny=<botlist>}}. This would work for you (if bots respected it), right? If a bot causes repeated consternation for not adhering to the {{bots}} standard, then I would support blocking it until it complies. This is a tough situation, since bots usually aren't required to be exclusion compliant (though in my view they should be), but if it's causing consternation, it should be stopped in my opinion. All the best, – Quadell ^{(talk) (random)} 21:08, 12 July 2007 (UTC)

This would still only work half way. First, both the set of all interwiki bots and the set of all non-interwiki bots are essentially open-ended. I wouldn't know how to exhaustively enumerate either what to allow or what not to allow. Second, it's not that people want to ban all interwiki bots either. There's plenty of non-controversial stuff done to that page by interwiki bots. It's just that one link. I still think the change in the bot code that I suggested would be the best way to go forward. It can't be difficult to do, probably just three or four lines of code. Fut.Perf. ☼ 21:18, 12 July 2007 (UTC)

Bout time to archive this page?

This page is getting a bit long, anyone mind if I archive it? —The preceding unsigned comment was added by SXT40 (talk • contribs) 12:33, 18 July 2007 (UTC).

How about using a bot for this? After all it's a bot-related noticeboard. That said, I've set my bot to archive threads inactive for more than two weeks. Миша13 14:54, 18 July 2007 (UTC)

Excellent Idea (I use your bot too...) What's everyone else think? BTW, thanks for fixing my sig... I forgot :)--SXT4^{$\color {Red}\oplus$} 15:21, 18 July 2007 (UTC)

Question about a possible clone?

Forgive me if I'm posting this in the wrong place, and if I am, please correct me. I'm running MediaWiki software on my own website and it is getting big enough that I now need a bot that will automatically add a stub tag to articles shorter than a certain size. I know XHTML and some CSS, but no "hard" coding like perl or C++. Is there a bot here that I can clone and run on my wiki? I can follow basic installation/deployment instructions. Thanks for your time. Archer904 17:27, 19 July 2007 (UTC)

Do you have perl installed on your machine? – Quadell ^{(talk) (random)} 18:25, 19 July 2007 (UTC)

I'm hosted by phpwebhosting.com...and they allow Perl, python, ruby, C, and C++. Archer904 22:04, 19 July 2007 (UTC)

Not PHP? It'd be an easy bot to write in PHP, and that's my best language. — Madman bum and angel (talk – desk) 00:31, 20 July 2007 (UTC)

I'm sure it runs PHP or else the the MediaWiki software wouldn't run and the web hosting company would have to change their name to something more relevant. :)--I already forgot 00:51, 20 July 2007 (UTC)

(edit conflict) Ah, good! Assuming you have recent-ish versions of all the required modules, I can probably help you. I'll take this to your talk page. – Quadell ^{(talk) (random)} 00:41, 20 July 2007 (UTC)

Bot Approvals

We're trialling a change to the Bot Approvals process allowing anyone with bot experience to join the approvals group. There are some details on that page, and discussion can go towards the talk page, WT:BAG.

In related news, there's a change to the technical side of posting bot requests, when you initially post a request to WP:BRFA, rather than transcluding, there's a template that is used, Template:BRFA. --ST47Talk·Desk 14:43, 27 July 2007 (UTC)

There's a problem with that template. When you click the edit links for an request on the WP:BRFA page, it is trying to edit the template & not the subpage. This means you have to actually go to the individual bot request page to make comments which is annoying. Thanks. -- JLaTondre 17:05, 27 July 2007 (UTC)

Working on it. —METS501 (talk) 17:25, 27 July 2007 (UTC)

Oooh, i see Mets is back...? Reedy Boy 17:38, 27 July 2007 (UTC)

Yep :-) And I've fixed the edit section link problem (well, not really fixed, but this and this seems to do the trick; if anyone else sees a better way, let me know) —METS501 (talk) 17:47, 27 July 2007 (UTC)

Category:Emergency shutoff compliant bots

This is a self-reference category name, which is bad per WP:ASR. I am planning to rename it to Category:Wikipedia emergency shutoff compliant bots, unless there are strong objections or someone else comes up with a better name. — Carl (CBM · talk) 19:18, 8 July 2007 (UTC)

Well, there have been a few very similar speedy renames in the past. Personally, I would have no objections to having the category listed at WP:CFDS even though there's really no harm in going the regular route . --S ^up? 19:37, 8 July 2007 (UTC)

Just tag it with {{Wikipedia category}}, and forget about it. (→zelzany - review) 19:39, 8 July 2007 (UTC)

I would prefer the rename; speedying it would be fine by me (although there's always a chance that someone would object on 'out of process' grounds), but there's nothing particularly urgent about it. --ais523 16:07, 10 July 2007 (UTC)

That's fine and all, but ASR is for article content, which excludes the bot category. -- Ned Scott 05:58, 20 July 2007 (UTC)

Honest question: What's the point of this category? All it lists are users that use a template -- something that can easily be realized from the WHATLINKSHERE on that template. Bots that don't use the giant button can be shut off just as easily with the "block user" link. Ral315 » 21:06, 20 July 2007 (UTC)

I agree. Bots don't need to give their permission to be shut off. If they malfunction in a harmful way, they will be blocked regardless of their "compliance." Andre (talk) 05:45, 1 August 2007 (UTC)

I have nominated this category for deletion. Ral315 » 00:49, 8 August 2007 (UTC)

Bot question

If I wanted to run a bot in Perl continously, how would I keep it running? My internet connection isn't active all the time, so how would the bot stay connected and continue running? GrooveDog (talk) 19:33, 6 August 2007 (UTC)

use a indef loop, As for your internet connection I have no clue how to help, unless you get a host that has constant internet access β^_command 19:36, 6 August 2007 (UTC)

For a possible constant internet connected host, see m:Toolserver. — E ^talk_bots 20:47, 6 August 2007 (UTC)

Just a note - if you do not have a toolserver account, and want to run it without waiting for approval, or you don't have any UNIX/Linux experience, you can ask another user to run it from the toolserver for you. Matt/TheFearow (Talk) (Contribs) (Bot) 00:12, 7 August 2007 (UTC)

FYI: Both myself and TheFearow have accounts. Feel free to contact either of us if you want it temporarily hosted. — E ^talk_bots 07:03, 7 August 2007 (UTC)

Hm...Well, I'm just going to be writing it this week, so thanks for the offer! I'll probably take you up on it temporarily until I can get approval for an account. GrooveDog (talk) 12:28, 7 August 2007 (UTC)

Joining the BAG

As part of the BAG policy when joining, I must notify certain pages about my request for BAG membership. Please see my request at Wikipedia talk:Bots/Approvals group#Nomination to join the BAG. Thank you for your time, — E ^talk_bots 12:09, 10 August 2007 (UTC)

Is there any bot issuing xfD or prod notices?

I think I remember a bot that did notices for prods (although it may have been AfDs) but I've looked at articles nominated in the past days for both prod and AfD and haven't found it issue any notifications. --Android Mouse 02:01, 12 August 2007 (UTC)

Twinkle users automatically notify creators, but I am not aware of any explicit bot activity. --After Midnight ⁰⁰⁰¹ 02:20, 12 August 2007 (UTC)

Maybe I'm mistaken and am confusing an image notification bot with this. Still, I think it would be useful since not everyone uses TW and those that do often close the window that pops up after tagging an article without giving any notification. --Android Mouse 08:32, 12 August 2007 (UTC)

I think that Twinkle does the notification automatically, it does not give the user a choice to not leave that particular message (other Twinkle messages are optional). As you mention however, not everyone uses Twinkle. I agree that people should be notified when the pages they created are nominated, but I can't image a bot successfully doing this. Among other things, how would the bot know if the author were already notified manually before the bot got to them? We don't want people to receive duplicate notifications any more than we want them to not receive any. --After Midnight ⁰⁰⁰¹ 13:25, 12 August 2007 (UTC)

Indeed, and we've already had problems with bots notifying "creators" of pages, when for one reason or another the author of the first revision was not the creator of the page. — madman bum and angel 16:36, 12 August 2007 (UTC)

I'd have the bot do as my other notification bot does, if it finds any occurance of the article name on the user's talk page it would skip it. --Android Mouse 17:53, 12 August 2007 (UTC)

OK, that would work as long as the notified person didn't remove it from their page before the bot got there. I imagine that there is no way that it would work with DRV, since you would really want to notify the closing admin in that case, instead of the author. --After Midnight ⁰⁰⁰¹ 20:48, 12 August 2007 (UTC)

Interwiki bots forcing unwanted links, again

I'd like to draw people's attention again to the problem of interwiki.py-based bots making repeated reverts of controversial interwiki links against the wishes of local editors, essentially revert-warring, as reported previously here. This problem continues to be unsolved and to create a lot of friction on the article talkpage.

This needs to stop. I am hereby putting all operators of interwiki bots on notice that their bots will be blocked if they make this particular edit repeatedly. Bots must not engage in revert-wars. Please fix this software, it's broken. Fut.Perf. ☼ 02:00, 14 August 2007 (UTC)

You want to complain at m:interwiki.py, where it is located. You should try and get the link blacklisted there. But out of curiosity, why do you not want that particular link? It's just an inter language link... Matt/TheFearow (Talk) (Contribs) (Bot) 02:20, 14 August 2007 (UTC)

I reported it there, weeks ago. Zero response. I reported it also at the pywikipedia mailing list. Zero response. -- The reasons for the controversy are discussed at length at Talk:Ingria, the background is the trainwreck at m:Proposals for closing projects/Closure of Siberian Wikipedia. Short summary: it's an interwiki link to a project that represents an artificial "language" used by and associated with only a tiny anti-Russian political fringe group, extremely offensive to Russian editors, and whose legitimacy as a project is in serious doubt. The only reason the whole wiki hasn't been shut down yet is that nobody finds the courage to take the decision. But really, it's none of my business, I'm only trying to stop an edit-war conducted by bots. Fut.Perf. ☼ 02:31, 14 August 2007 (UTC)

Perhaps enabling Template:Bots will work here? (If interwiki.py supports it—I'm not sure if it does.) Gracenotes^T § 02:25, 14 August 2007 (UTC)

As discussed earlier, it doesn't, and even if it did it wouldn't really solve the problem well. My suggestion is and has been that the bots should be changed to recognise a commented-out link and leave it alone. Clean and simple. Fut.Perf. ☼ 02:31, 14 August 2007 (UTC)

• This seems to be getting a little heated. Questions:
  1. Is this only a problem with this one article?
  2. Have you tried editing out the en: link on ru-sib: ? (most of these bots are crawlers, and as they are coming from a foundation wiki- that is linked here, they are just trying to link back)

  Thanks, — xaosflux ^Talk 04:42, 14 August 2007 (UTC)

As I understand interwiki.py (and I don't use it myself), suggestion 2 wouldn't work. You would need to remove the ru-sib: link on all Wikipedias. From what I've seen, I agree with Fut.Perf. on all points. I'll have a look whether I can change interwiki.py as suggested, but I think it's okay to block bots that change pages against the wishes of (human) editors, even in this situation. -- Jitse Niesen (talk) 05:13, 14 August 2007 (UTC)

I don't. OrphanBot and ImageRemovalBot frequently edit pages against the wishes of some human editors. Bots editing against community consensus are another matter, but I don't see consensus here that we shouldn't be making interwiki links to ru-sub:. --Carnildo 05:51, 14 August 2007 (UTC)

To Xaosflux: The Ingria article is the only one I'm aware where there is currently an active edit war ongoing, but there are several others where the ru-sib target articles are even more offensive or ridiculous, people are just not paying as much attention to the issue there. But, as I keep saying, the problem is really one of poor software design and as such goes beyond the ru-sib issue. Think of this constellation:

• xx:User:X wants an interwiki link from xx:Fu to yy:Bar.
• yy:User:Y doesn't want a link from yy:Bar to xx:Fu. For whatever reason, good or bad.
• The bots are currently designed to decide that xx:User:X is right. Always. And they'll mechanically revert-war yy:User:Y into submission, unless xx:User:X can be convinced to remove the link on their side.

This is simply wrong. It's the same problem that occurs in those everyday cases where it's just due to a factual error on the part of X (like here, which drove me to distraction until I figured out how to stop it.)

To Carnildo: the comparison with the image bots is flawed. The image bots are enforcing a policy whose overriding importance and non-negotiability has been determined beforehand in extensive community-wide debate. There is no such overriding policy concern in the case of interwiki links, certainly none that would override the prohibition of revert-warring. There is no foundation policy that says: interwiki links must be used wherever possible without any exception. I'm not saying such a principle might not be reasonable or it might not find consensus if you asked the community. But nobody did the asking. And it's not for the bots to decide, nor for the bot developers to decide, nor for the bot approval people to decide, nor for this noticeboard to decide.

I'm not asking a lot. I'm asking for the addition of probably just two lines of code to the bot software, something equivalent to:

if (! ($PageText =~ /\<\!--\s*$ProposedLink\s*--\>/)) {
    # ... do whatever ...
};

I'm also asking to keep two issues apart: (1) should we (Wikipedia editors) link to ru-sib articles? and (2) should we (bot operators) acknowledge the possibility that some interwiki links might be unwanted? (1) is an issue that has to be solved through dispute resolution, not by a revert-warring bot army; (2) is an issue that has to be solved by fixing the bot code. Fut.Perf. ☼ 07:02, 14 August 2007 (UTC)

Interwiki links are a bit of a special case: they are usually maintained by bots, and most human editors don't do much with them. Nevertheless, I think there is a huge consensus that interwiki links between all topic-equivalent articles in all language Wikipedias should always exist. They should exist regardless of the quality of the articles (stub here, FA there, etc.). Special casing links for reaons like "article in language XX is poor and has POV" won't help anybody find and correct that POV. We also do not want to have authors going around and remove interwiki links to perfectly good articles that they happen to not like. The problem here seems to be only that ru-sib Wikipedia sucks, and that the Foundation does not have an effective way of dealing with projects in foreign languages that produce problematic content. Kusma (talk) 07:19, 14 August 2007 (UTC)

I'm wondering how the heck they got the project started. -- Ned Scott 07:43, 14 August 2007 (UTC)

Reading more about this, I'm very shocked that this project hasn't been closed down. The Foundation might not know what to do, but I suggest developing a consensus on en.wiki to add "ru-sib:" to the spam filter. -- Ned Scott 07:49, 14 August 2007 (UTC)

Addendum: Independent of the ru-sib issue, I just wish to point out again that my proposal is also related to the everyday occurrence of bot reverts in cases where unwanted links are simply the product of factual errors proliferating through the different wikis. A brief look at any botowner's talkpage confirms that this is happening all the time: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14

In a lot of these cases, users were justifiably upset because bots were making these mistaken edits repeatedly, and the users couldn't figure out why they were doing it and how to stop them. Come to this, many interwiki bots are poorly documented - their userpages don't state what software they are running on, where to find out about their algorithm, who approved them, where to go with complaints, etc. The documentation of pywikipedia on meta is difficult to find.

I propose:

1. interwiki.py should offer an easy method for local editors to tell the bots to leave a specific interwiki link alone. Most transparent and probably easiest to implement, as I said: Presence of a commented-out link on a page automatically prevents reinsertion of the same link by bots.
2. All interwiki bots should be required to observe this convention.
3. All interwiki bot owners should be required to clearly document this method, and any related processes for fixing errors, on the bot's userpage.
4. All interwiki bot userpages should be required to link prominently to a suitable documentation of the pywikipedia framework and the interwiki.py system (algorithm, logfile etc.)

Fut.Perf. ☼ 10:27, 14 August 2007 (UTC)

Prohibiting the bot from re-adding a certain interwiki link is just a band-aid over the problem that some other Wikipedia has its links wrong. The link should be fixed (by making the edits on all wikis), not just disabled on one wiki. I agree that this should be better documented. Kusma (talk) 11:00, 14 August 2007 (UTC)

Two comments: (a) this puts an undue burden on the person noticing the mismatch. It's a lot of work going through two dozen wikipedias, getting an account on each (in case you don't wish to display you IP all over the place), and then edit pages in scripts you might not even be able to display on your computer, let alone read. (b) if the mismatch originates in some Wikipedia I can't read, I may have no way of even understanding what has to be fixed. (c) It still doesn't address the problem that in some (though certainly rare) cases the mismatches may be due to genuine differences of opinion. What if editors on wiki A explicitly want the link, for whatever reason, and editors on wiki B explicitly don't want it (because the match between the topics is not close enough, or they have a different project-specific policy about linking, or whatever? I propose that in such cases, local consensus must trump the artificial ideal of total uniformity across all wikipedias, and the current bot behaviour just makes that impossible. Fut.Perf. ☼ 11:17, 14 August 2007 (UTC)

Additional suggestion: If the bot finds A has a link to B but editors on B have objected against the reverse link to A, a good thing to do might be for the bot to post a warning to A's talk page, with a request for humans to check the links. Bots should not attempt to be more clever than humans and pretend to know better than them what links are appropriate. Fut.Perf. ☼ 11:25, 14 August 2007 (UTC)

Some of this discussion has been brought up at Help talk:Interlanguage links before, FYI. — xaosflux ^Talk 11:30, 14 August 2007 (UTC)

D'oh. Another indication of how poorly documented this whole thing is. Yet another page I wasn't even aware of. Fut.Perf. ☼ 11:37, 14 August 2007 (UTC)

I'm not sure exactly which section of the page Xaosflux is talking about. But one of the things noted there (three years ago) is that interwiki links are not necessarily transitive. That is, if en:XXX links to fr:YYY and fr:YYY links to es:ZZZ then en:XXX shouldn't necessarily link to es:ZZZ. An hypothetical example which I'm plucking out of thin air in which this would happen, is if we at the English Wikipedia have an article "Nihillissimo", which describes a book called Nihillissimo, and the French have an article describing the book and also the film made from the book, and the Spanish have an article about only the film. Then we wouldn't want to link from the English article to the Spanish article. Anyway, leaving the blocking aside, is there any reason not to give editors of a page a possibility to override the bots? -- Jitse Niesen (talk) 12:16, 14 August 2007 (UTC)

In some cases the bots already seem to handle such things properly. For example, most Star Wars people (like Darth Vader) are interwikilinked to a section of de:Figuren aus Star Wars, which is linked to List of Star Wars characters. I guess your example could be handled by using section links. In any case, we should try to get more interwiki bot operators in this discussion. Kusma (talk) 12:22, 14 August 2007 (UTC)

In terms of precedents of "boycotting" entire Wikipedias for wiki-political reasons, that page also mentions the case of the Klingon wiki, which in 2004 it was decided shouldn't be linked to from en. More recently, it seems that ru-sib and ru already have been systematically boycotting each other. Interestingly, the Russians apparently found it impossible to do technically, probably because of the insistence of the bots, and used a CSS hack as a workaround. They tolerate the links in their page source, but just won't display them. Fut.Perf. ☼ 12:40, 14 August 2007 (UTC)

OK, so I don't use the Pywikipedia bot framework myself, but I'm familiar with the code, and I fail to see why these Pywikipedia bots are apparently not obeying {{bots}} and {{nobots}}. You've tried, right? I know that wikipedia.py includes a botMayEdit() function, which does a loose check for those templates. Perhaps a call to botMayEdit() should be made mandatory from interwiki.py? (and all other non-essential modules, for that matter? *cringes in fear of angry operators*) When writing an encyclopaedia is the goal, editors' wishes should have at least that much control over bots' operation. — madman bum and angel 18:54, 14 August 2007 (UTC)

Yeah, ST47 tried it with his bot a couple weeks ago when I talked to him about the problem. In fact, I guess making the bot search for an already commented-out link, as I've been suggesting, would be just as easy to implement as making it search for the "nobots" template, wouldn't it? And far less invasive to other, potentially uncontroversial, editing activity the bots might do. As I said during the last round of discussion, nobody wants to exclude all bots or even all interwiki bots from that page. They just don't want that one link. Fut.Perf. ☼ 19:00, 14 August 2007 (UTC)

Indeed. I was just commenting on the more grievous omission, considering botMayEdit() is already built-in, and if an editor says that they do want a certain bot to keep out, it should. — madman bum and angel 19:07, 14 August 2007 (UTC)

AIV helperbots needed

The original AIV helperbot has been down since its owner left and the others are both been down for a few hours and have increasingly inactive owners. Would anyone already setup to run bots mind setting up another AIV helperbot? The source code is here and it's already gone through approval, of course, so all we need is a willing bot owner with the necessary resources. Even if the other two come back online, a little redundancy isn't a bad thing.--Chaser - T 20:22, 18 August 2007 (UTC)

I'll put one up on emergency basis and seek approval. --ST47Talk·Desk 20:54, 18 August 2007 (UTC)

Hm. I need to find someone with the Mediawiki.pm module, that may take a little longer. --ST47Talk·Desk 21:00, 18 August 2007 (UTC)

I have all the requirements for a HBC AIV helperbot as I was a former runner of #2. — E ^talk_bots 21:51, 18 August 2007 (UTC)

I have one up, had a small bug with the edit token regex that had me running frantically through code I'd never seen before for much longer that it should have. Requesting approval, but E, you may want to get one approved as well for as needed operation. --ST47Talk·Desk 22:13, 18 August 2007 (UTC)

Ya'll seriously rock. Thanks. - Philippe | Talk 22:27, 18 August 2007 (UTC)

It just looks like something caused one or both of the running helperbots to lose their login sessions. I've kicked my instance and it's running normally again. The most helpful thing that could be done would be for someone to look into "am I logged in" verification and add it to the code. It's a rare situation, which makes it somewhat annoying to code for, and I simply don't have much time to work on Wikipedia coding at the moment. You're correct that I don't actively contribute to Wikipedia much anymore, but I do actively monitor my e-mail virtually 24/7 (with today, of course, being an exception because I was having a housewarming party) and check my talk page at least once daily, usually more.

ST47, I got your e-mail, and will send you the updated copy of the module shortly. Though it's already been done now, I might suggest I look into taking over the HBC AIV Helperbot user, and we re-use HBC AIV Helperbot3, rather than jumping in with HBC AIV Helperbot4, if we can. Of course, that's kind of moot since Helperbot4 has already been created. *shrug*. Anyway, 3 is up, contributions to prevent this from recurring are always welcome, and anyone who is now running the bot and plans to continue doing so, please get in touch with me to get on our private mailing list for operations coordination (mostly for discussing updates to the bot, etc). —Krellis (Talk) 02:07, 19 August 2007 (UTC)

I would also like to note, for the record, that no one seems to have even attempted to contact me about this, via my talk page or e-mail, until ST47 did so, quite a long time after the bots had stopped functioning, and even then, it seemed the request was more for the updated MediaWiki module than for getting the bot back up. Had I actually been around today, and even if I had been more actively contributing, there is no telling that I would have noticed random comments about the bots on the talk pages of some bot-serviced pages, especially since even some of those comments seem to have come only after the bots were offline for several hours. I'm certainly not trying to lay any blame on anyone, but I don't like the insinuation that this problem went on longer than it should have because I am less-than-totally active on Wikipedia. I run the HBC* bots I run for the greater good of the Wikipedia community, and I'm happy to help anyone get additional clones of the helperbot up and running if if is felt they are necessary and the BAG sees fit to approve them. (This sounds needlessly defensive, and for that I apologize, as it is not my intent, but I've had a long day, and now it's time for bed.) —Krellis (Talk) 02:15, 19 August 2007 (UTC)

Unligitimate blocking

Recently User:Future Perfect at Sunrise unlegitimately blocked my bot. Here you could see the discussion about it. Please help me to do something, or give me a link where I could request for unblocking --A4 11:00, 23 August 2007 (UTC)

Bots shouldn't make controversial edits or edit war, after being asked not to. I'll certainly unblock you if you agree to have it ignore the Ingria page. – Quadell ^{(talk) (random)} 12:57, 23 August 2007 (UTC)

Please submit a patch containing the AI so the bot can decide if an edit is controversial or not. I will commit it personally. ValHallASW 14:20, 23 August 2007 (UTC)

"...after (their operator) is being asked not to" is the key. I believe operators do not need AI to understand that an edit is controversial after being told so. Tizio 14:28, 23 August 2007 (UTC)

Correct. However, telling them to use -neverlink ru-sib using a comment on the Ingria page is not asking the operator in my opinion. However, the page should be left alone by bots after they update to r4096+, which now respects {{nobots}} by default (but it may be overridden). ValHallASW 14:43, 23 August 2007 (UTC)

The "AI" I've been proposing for a while is simple: check whether the link is already on the page but has been commented out. That means human editors are aware the link target exists, but have made a conscious decision to not use it. Which is all the bot needs to know. Shouldn't be difficult to implement. Fut.Perf. ☼ 17:30, 23 August 2007 (UTC)

Should be possible to implement. A template would probably be easier for the bot, something along the lines of {{iwignore|[[ ru-sib:Ингерманландия]]}}. I have added it to the feature tracker here. ValHallASW 23:04, 23 August 2007 (UTC)

Sorry A4, bots should be used for non-controversial tasks and unless it is a vandal bot it should not be reverting people, at least not more than once. ((1 == 2) ? (('Stop') : ('Go')) 14:33, 23 August 2007 (UTC)

RTFM. It's not *reverting*, it's *adding a missing link*. As far as I know adding interwiki links is hardly controversion. Again, be my guest and submit a patch that determines the controveriality of an edit. Any ETA? ValHallASW 14:43, 23 August 2007 (UTC)

ValHallASW, I did read the discussion, and the problem is the bot made the same edit more than once after it was undone. That is a revert. If it only did it once that would be fine. This[15] is just adding a link which is fine, but this[16] and this[17] are reverts. ((1 == 2) ? (('Stop') : ('Go')) 17:36, 23 August 2007 (UTC)

Technically, there is a difference. Reverting is opening an old revision and replacing that one. Adding a link is getting all the known iw links and placing them. ValHallASW 23:04, 23 August 2007 (UTC)

I'm not saying you did anything wrong on purpose. I'm saying your bot (inadvertently) made a controversial edit. If you make sure the bot doesn't edit the same article again, there's no problem. Just say you'll do that, and your bot will be unblocked, and the problem goes away. (Or just continue to be sarcastic, if you prefer, but I doubt it will work well for you.) – Quadell ^{(talk) (random)} 15:11, 23 August 2007 (UTC)

Unblocked

I'm not really sure where we need to go with this matter, but the only complaint here is regarding apparent inter-wiki edit warring on Ingria, and not anything that is technically malfunctioning with the interwiki program. The issue also appears to go much beyond this particular bot, in such that blocking it will not resolve the issue. I've unblocked this bot so it can continue working, but I've full-protected Ingria as an article being edit warred over. This protection is for 30 days. This appears to be entering an impasse, so perhaps an RfC is in order along the lines of "interwiki bot operators v.s. Ingria editors". — xaosflux ^Talk 00:48, 24 August 2007 (UTC)

A bot keeps edit-warring with humans and still there is nothing technically malfunctioning in it? Please define a "malfunction" then.

I agree that the issue is wider than just the Ingria article. However, the place to discuss whether "bots must not revert humans (apart vandals)" is IMO the policy page WP:BOT, not an RfC. Tizio 00:58, 24 August 2007 (UTC)

If it were A bot I'd completely agree with you, but it's not it's tons of bots, and we can't pretend that blocking this one bot is going to resolve the issue. No evidence that this bot is actively edit warring is in place. While this is certainly a bot issue, if this was group of editors this would be going to dispute resolution, not blocking one of them. — xaosflux ^Talk 01:07, 24 August 2007 (UTC)

(ec)I have decided to start a general discussion about bots reverting humans here. This is about policy, not the specific case. Tizio 01:16, 24 August 2007 (UTC)

As far as I can see, the problem is not a group supporting an article version and another group supporting another version. Here we have a group supporting the iw links apparently because that's the simpler technical solution, not because they want the links. Tizio 01:26, 24 August 2007 (UTC)

Why don't...

we just add

#interwiki-ru-sib { display:none; }

to MediaWiki:Common.css? MaxSem 06:18, 24 August 2007 (UTC)

Because it's only wanted to be hidden on that one page. Maybe:

.page-Ingria #interwiki-ru-sib {display:none;}

That would only hide it for that page. Matt/TheFearow (Talk) (Contribs) (Bot) 09:30, 24 August 2007 (UTC)

• Oppose this, common.css should not be a creeping featurism. — xaosflux ^Talk 11:59, 24 August 2007 (UTC)

Totally opposed. We should not be putting anything in the site-wide css/js that deals with a single page, with the possible exception of extremely widely viewed pages such as Main Page. --Cyde Weys 13:19, 24 August 2007 (UTC)

I think we should link to all ru-sib pages, or to none. Given what I hear about ru-sib, I think we should link to none until the Foundation (or whoever) decides to close the m:SIB discussion. I think the decision can't be more than two or three years away. Kusma (talk) 13:25, 24 August 2007 (UTC)

Why don't folks consider Fut.Perf.'s suggestion and add

if (! ($PageText =~ /\<\!--\s*$ProposedLink\s*--\>/)) {
    # ... do whatever ...
};

or some similar algorithm. Additionally, an "all or none" approach is silly. This is a single page, and a single interwiki link, where editors have reached a consensus that it is simply best to leave it out due to the inevitable edit warring when it is added back in. Keep the problem in scope. --Iamunknown 21:51, 24 August 2007 (UTC)

For what it's worth, I think an enwiki-wide decision to not link to ru-sib at all would be a good thing. The conflict over this single page seems to be really just a test case for both sides; almost all of the arguments exchanged over it really apply to the whole wiki. But the hack of hiding the links by css is certainly ugly. Fut.Perf. ☼ 07:12, 27 August 2007 (UTC)

Image-matching regular expression

I've been trying to make a regular expression that tests if a excerpt of wikitext contains an image. Very annoying is the fact that an image description may itself have a link, or just a single bracket ([ or ]). I am pretty sure that the following definitely works, if not attaining the title of Regular Expression from Hell:

• /\[\[[Ii]mage:[^\|\[\]\{\}]*(?:\|[^\[\]]*)?(?:(?:\[\[.*?\]\]|\][^\]]|\[[^\[])[^\[\]]*)*\]\]/

I'm sure there must be bots that find images in wikitext. For their authors: how do you do it? A simpler regular expression, hopefully? Is the task just not a one-liner? Thanks, Gracenotes^T § 21:47, 26 August 2007 (UTC)

Hm... it appears as though this matches something like [[Image:Im[g.png]]. Most unfortunate. *thinks* Gracenotes^T § 22:10, 26 August 2007 (UTC)

Uh, never mind that. (In the second character class, I left out [) Gracenotes^T § 22:16, 26 August 2007 (UTC)

Gracenotes answering himself :-p. ~ Wikihermit 00:17, 27 August 2007 (UTC)

Although, I can't answer myself on whether the regex works [and believe me] :) Gracenotes^T § 17:00, 27 August 2007 (UTC)

That's got the same problem that OrphanBot has: it doesn't match things like [[Image:Foo.jpg|[http://google.com From Google]]] properly -- it will match the first two closing brackets, not the last two.

ImageRemovalBot uses a two-stage process to match images: First, it converts all [[ and ]] groups to single non-printing characters to make regular expression matching easier, then it uses the regular expression

\x01${image_regex}[^\x01]*?(?:\x01[^\x02]*?\x02[^\x01]*?|)+\x02[ \\t]*

where ${image_regex} is a regular expression that matches the name of the image it's removing (an exciting topic in its own right), \x01 is a non-printing character that represents an opening double-bracket, and \x02 is a non-printing character that represents a closing double-bracket. --Carnildo 03:08, 28 August 2007 (UTC)

May Be of Some Interest

Didn't know if anyone would be interested in bugzilla:11078. Just as an FYI. ^demon^{[omg plz]} 14:16, 28 August 2007 (UTC)

User:ImageRemovalBot problems

Please see Wikipedia:Administrators' noticeboard/Incidents#Bot_being_disruptive.2C_owner_not_responding_to_the_criticism for details - I was pointed to here after being pointed there by WP:BOT which says to add to Wikipedia talk:Bots and if it can't be resolved go to WP:AN/I. I believe it should be blocked until fixed. -81.178.126.124 20:54, 29 August 2007 (UTC)

I would block, but I don't really understand what the bug is. —METS501 (talk) 21:03, 29 August 2007 (UTC)

Logged In Checks?

Does anyone have a good magic method for a bot to use to check if it is logged in? I'm looking at fixing the "bot gets logged out and fails" problem that the HBC AIV helperbots had the other day, but the only relatively simple way I've found, using MediaWiki.pm, is to edit a page, then check if the last edit to that page came from the expected username. This method works, but means the bots have to make periodic edits to a check page to see if they are logged in or not, and that presumably isn't very nice to the servers. Looking at the source code of the edit page when logged in vs. not logged in, I see several places where one could look for the username, but they all seem somewhat non-portable, and likely to break with any changes to the page formatting (and would require local modifications to MediaWiki.pm, since it doesn't do this currently). Any thoughts or suggestions? Thanks! —Krellis (Talk) 20:02, 25 August 2007 (UTC)

We AWB devs check for JS variables, e.g.

var wgUserName = "MaxSem";

MaxSem 22:55, 25 August 2007 (UTC)

Hmm, yeah, that seemed to be the best bet when I was looking at the source. In your experience w/ MediaWiki, is that something that has been changed often, or has it stayed pretty constant? —Krellis (Talk) 23:26, 25 August 2007 (UTC)

From my experience with MediaWiki, nothing can be considered constant enough, but you can always change your code when Brion & Co change their. MaxSem 23:59, 25 August 2007 (UTC)

Right, thanks :) —Krellis (Talk) 00:17, 26 August 2007 (UTC)

wgUserName and similar variables were as far as I remember added for the benefit of user scripts, which would mean that they're likely to be relatively stable. --ais523 16:50, 30 August 2007 (UTC)

OrphanBot checks for a link with the text "My talk" (regex />My talk<\/a>/), which has only broken once in the years that the bot's been running, and has never let the bot edit while logged out (but requires the English interface and the Monobook skin). You could also check for the presence of a hyperlink ending in a colon followed by the bot's name ("http://en.wikipedia.org/wiki/User talk:OrphanBot"), which should be robust against all changes in language, skin, and interface, but can be fooled if someone's talking about the bot. --Carnildo 18:31, 30 August 2007 (UTC)

Couldn't that be fooled if someone piped a link to the text 'my talk', or does it only check outside the bodyContent? And just out of interest, what caused that one break? --ais523 18:40, 30 August 2007 (UTC)

No, it won't fool the bot, since it doesn't do the check except when it's getting a page with "&action=edit". I don't remember what caused it to break, only that a year or so ago, I checked OrphanBot's logs, and found that it had stopped running because it thought it was about to edit while logged out. --Carnildo 23:21, 30 August 2007 (UTC)

Geo-reference

Template:GR is currently being used by thousands of articles. It is linking to a Wikipedia page thereby failing WP:ASR. I am trying to add a fixed reference to the geo-reference template in my sandbox. But the <references/> tag is not picking up the reference added by the template. Can anyone please check and advise how this can be done? Thanks, Ganeshk (talk) 01:17, 31 August 2007 (UTC)

I'm not sure exactly what you mean. I can't see a reference in the template; when I added one, the references/ tag picked it up fine. See User:Ganeshk/sandbox/Verisimilus Could you elucidate a little? Thanks, Verisimilus T 13:11, 2 September 2007 (UTC)

The template should not have a references/ tag. I am trying to get the reference inside the template to show up in the article that the template is transcluded in. My guess is references/ tag only picks up ref tags that are on the actual article and not templates that are included in it. So, I have asked for bot approval for a task that would replace all ocurrences of {{GR|India}} with <ref>{{GR|India}}</ref>. Thanks for checking and responding. Regards, Ganeshk (talk) 18:44, 2 September 2007 (UTC)

Ah, I see what you mean. You're right - that appears to be the only solution. Verisimilus T 13:46, 3 September 2007 (UTC)

User groups / protection levels

I've been taking a look at API.php and converting my PHP bot over to using that. One thing I've run across that I have a question about is the protection levels of articles. For instance, API.php can return the edit protection level, which may be "sysop" or "autoconfirmed". In searching around a bit, I noticed that there's a javascript statement on every page I view that says:

var wgUserGroups = ["*", "user", "autoconfirmed", "emailconfirmed"];

and a little more searching around shows other possibilities:

"boardvote", "bureaucrat", "checkuser", "oversight", "sysop", "*", "user", "autoconfirmed", "emailconfirmed"

So my question: Is there a level system here? Is "oversight" able to do more things than "autoconfirmed"? Is "bot" a level in that system? Any info is appreciated! -- SatyrTN (talk | contribs) 04:41, 3 September 2007 (UTC)

Wikipedia:Administrators#Other access types and Wikipedia:User access levels should explain it. Quick and probably not quite correct explanation: It is not quite a level system, but almost. From most to least power, we have steward - bureaucrat - sysop (administrator) - autoconfirmed (account is more than 4 days old) - user (anybody who is logged in) - * (everybody). The other groups are independent of this hierarchical systems: "bot" means that the edits are hidden by default in "recent changes", "oversight" that you can do Wikipedia:Oversight, "checkuser" that you can do Help:CheckUser, "boardvote" has something to do with when there are elections. I don't know what "emailconfirmed" does. -- Jitse Niesen (talk) 05:15, 3 September 2007 (UTC)

"emailconfirmed" means that your account has associated itself with an email address, and that the email address has confirmed the association. If Special:Emailuser doesn't return an error, that is. Gracenotes^T § 05:47, 3 September 2007 (UTC)

The various groups each allow a user to do various things; this can be customized on various wikis. Here's an example of what sort of things a user in each group can do (this is just one example for each group, many groups can do other things as well): '*' (all users) can edit, 'user' can create a page, 'autoconfirmed' can move a page, 'sysop' can delete a page, 'bureaucrat' can rename users, 'emailconfirmed' can send and receive emails, 'bot' edits don't show up by default in Recent Changes, 'checkuser' can match usernames to IPs, 'oversight' can delete edits in such a way that even 'sysop's can't see them, and 'boardvote' can make changes to a running Board election. There's also 'steward', which can change any user's rights to anything, even on a different Wikimedia wiki (they're mostly at Meta, though, rather than here, because the right works from one wiki to another). It's not a hierarchy as such (although all bureaucrats are also admins here, because a 'crat could just make themselves into an admin if they weren't); each right is independent of the other. Your wgUserGroups shows that you can do anything that everyone can do (not very surprising), that you've become 'autoconfirmed' (by having a username for 4 days) and so can move pages and edit semiprotected pages, that you've confirmed an email address and so can send and receive emails, and that you have a username (which allows you to do all sorts of things; see WP:WHY for details of what a user can do). See Wikipedia:User access levels for a complete guide (which might have got slightly out of date by now). --ais523 10:36, 3 September 2007 (UTC)

Gotcha - thanks for such a complete answer! -- SatyrTN (talk | contribs) 13:26, 3 September 2007 (UTC)

Help - edit problems

Hi,

I've not been able to get the PHP class to work, so have had to reinvent the wheel myself. I've got as far as placing the edit I want to, but face two problems:

1. My bot account isn't allowed to include external links in an edit, without decrypting an image. I assume this will be most easily fixed by waiting?
2. When I try to use the bot to edit using my own credentials (as Verisimilus T) I get an edit conflict every time.

I'm getting the hidden form values successfully, by sending a first request to ?title=PAGENAME&action=edit. Do I need to change any of them? Any tips?

Thanks for your help,

Verisimilus T 23:53, 26 August 2007 (UTC) P.S. Source code is posted at User:Botodo/Source. P.P.S. You can run the bot (1 edit limit) here.

Is your bot flagged? If so, CAPTCHAs are bypassed. If that's not the case, then we need to contact a developer as soon as possible. You can also get the edit token through api.php (prop=info, intoken=edit) — madman bum and angel 05:56, 27 August 2007 (UTC)

$submit_vars["wpStarttime"]--;
$submit_vars["wpEdittime"]--;

I suspect that these lines are the problem, as those parameters are directly related to edit conflicts. Why are you doing this? You either need to set them to time() or pass them on unmodified. — madman bum and angel 06:02, 27 August 2007 (UTC)

api.php has edit token support now? Excellent! Gracenotes^T § 07:20, 27 August 2007 (UTC)

Thank you all for taking a look, and so quickly! I'll investigate the bot flagging and api.php (the help page for this was a bit overwhelming!).

Regarding the Starttime--; I've tried leaving them unchanged, and doing ++, to try and beat the edit conflict. I should really have removed those lines before posting the code to avoid confusion!

Also, CAPTCHA's are not bypassed by bot flags, at least on other Wikimedia wikis where I have run bots within a few hours of registering the accounts. Matt/TheFearow (Talk) (Contribs) (Bot) 07:46, 27 August 2007 (UTC)

(Undent) Hey -- What PHP class are you guys talking about? :) Sounds like what I've been looking for... --SXT4^{$\color {Red}\oplus$} 03:32, 28 August 2007 (UTC)

I couldn't make this class (BasicBot) work, possibly because of issues with PHP5 on my server; there are also a few cases where /wiki/ has to be replaced by /w/.

So I'm basing my bot on Snoopy. I wish I'd discovered this earlier - it makes life so much simpler!! Verisimilus T 08:51, 28 August 2007 (UTC)

I would recommend HTTP_Client, SXT40. It's part of the PEAR repository.[18]. An (outdated) example of its use: User:MadmanBot/Source/Task2. Contact me if you have any problems with it.  :) — madman bum and angel 14:45, 28 August 2007 (UTC)

I'm having this exact problem with similar code as Verisimilus; if anyone can figure this out it would be greatly appreciated!! —METS501 (talk) 21:49, 29 August 2007 (UTC)

Thanks, Madman! However, I'm already about 50% on my own PHP MediaWiki bot framework :P (Tryin to make it simple... things like GetArticle("FooBar", [$revision]); and PostArticle("FooBar", $content); etc etc etc) --SXT4^{$\color {Red}\oplus$} 00:02, 30 August 2007 (UTC)

Hurrah! I've fixed it. Turns out I was scraping the data incorrectly. Thanks for all your help! Verisimilus T 14:21, 2 September 2007 (UTC)

Just for future reference, there is another PHP bot framework, quite extensive and well tested. wikibot.classes.php, part of ClueBot. -- Cobi^(t|c|b|cn) 08:25, 6 September 2007 (UTC)

Beginner's question

Sorry if this is the wrong place to ask this, but just a very basic point: what do I submit if I want my program to submit an article? I guess it's something like http://xx.wikipedia.org/w/index.php?title='yyy'&action=submit, plus some data (edit token and so on). What is the format exactly?--Kotniski 14:47, 1 September 2007 (UTC)

My approach to this problem, whilst not very fruitful, was to look at the source of the "edit" page: the required data can be found in the form. Remember to include the hidden fields. User:Botodo/Source is my attempt at implementation, and almost works. That might be enough to get you started — but an experienced coder will be able to tell you more! Good luck, Verisimilus T 13:08, 2 September 2007 (UTC)

P.S. Yes, that's the right URL.

Thanks a lot for your help. I'm working on it...--Kotniski 22:08, 2 September 2007 (UTC)

Well, based on the form and values in the html source of the edit page, I constructed a request that looks like this:

h t t p://en.wikipedia.org/w/index.php?

title=User:Kotniski/Exper&action=submit&wpSection=&wpStarttime=20070903143844&

wpEdittime=20070902123410&wpScrolltop=&wpTextbox1=newword&wpSummary=exp&wpMinoredit=1

&wpWatchthis=1&wpSave=Save+page&wpEditToken=87b268c73f39defb3b835a1264116a0b%2B%

5C&wpAutoSummary=d41d8cd98f00b204e9800998ecf8427e

and when I tested it (by pasting it into the IE address bar) I expected it to respond as if I'd saved the page with the content "newword". However, it doesn't work (it just presents me with the edit page again, with no changes made). Am I doing anything obviously wrong?--Kotniski 14:56, 3 September 2007 (UTC)

Two things. First, the request has to be sent using an HTTP POST request, not a GET request (which correspond to submitting a form and to visiting a website respectively). Second, the edit token can't sensibly be hardcoded; it's there as a method of preventing people constructing links that would cause a page to be edited when they were clicked on (so that people can't be tricked into editing pages in ways they wouldn't want); you have to get an edit token quite close to the edit itself, and have to be logged in with the same session. Also, you would need to send the login cookies, but the browser's likely to have done that automatically. --ais523 15:02, 3 September 2007 (UTC)

Great, thanks for your help! I'm slowly starting to understand this stuff...--Kotniski 18:00, 3 September 2007 (UTC)

It may be programmatically easier for you to request an edit token of api.php. — madman bum and angel 13:39, 4 September 2007 (UTC)

That won't work. There is no edit functionality in the api.php on enwiki at the moment, since the api is still not completed. There is experimental code being developed to implement editing through api.php, but it's incomplete (I can easily get it to crash) and not part of the code that is running Wikipedia at the moment. — Carl (CBM · talk) 13:47, 4 September 2007 (UTC)

You can get an edit token from api.php (whether it's valid or not I'm not sure) - my problem was getting a starttime and edittime. Verisimilus T 09:49, 6 September 2007 (UTC)

You're right, you can get an edit token from the API. But you can't feed that edit token back to the api, because the edit action isn't implemented yet. Getting an edit token from api.php and feeding it back to index.php is scary; it would require reading part of the mediawiki source to make sure it will always work. That seems beyond the scope of this question. — Carl (CBM · talk) 13:17, 6 September 2007 (UTC)

Using whatlinkshere rather than categories for maintenance lists

I've suggested we should convert most of the maintenance templates to use a "whatlinkshere" based mechanism rather than categories to keep track of articles needing maintenance, and have implemented an example using template:copyedit/test. If you have an interest in this, please comment at Wikipedia talk:Maintenance#Using whatlinkshere rather than categories for maintenance lists. -- Rick Block (talk) 14:14, 7 September 2007 (UTC)

Beginner's question II - User Agent

This time I'm trying to get (read) access to en.wikipedia urls programmatically (using Python), and they all seem to be blocked (except perhaps ...index.php...&action=raw). Having read some posts in Python newsgroups, I understand that the problem (or at least one of them) is getting the right User Agent. The solutions suggested in the newsgroups revolve around setting the User Agent to be Mozilla/5.0 or some such. I don't doubt that this works, but is it good practice? Might it not be considered dishonest to pretend to be a browser you're not? Is there a better way?--Kotniski 20:04, 7 September 2007 (UTC)

I call my agent "pywikibot", though I don't know if that's an approved practice. Staecker 20:12, 7 September 2007 (UTC)

I call mine MetsBot/en.wikipedia/EMAIL_ADDRESS@DOMAIN.com or something like that. —METS501 (talk) 21:26, 7 September 2007 (UTC)

Thanks. So in fact any string will do for the User Agent, it doesn't have to meet any specific conditions?--Kotniski 09:37, 8 September 2007 (UTC)

Indeed. — madman bum and angel 19:08, 8 September 2007 (UTC)

Product tokens are often used, but anything will do. Gracenotes^T § 20:31, 8 September 2007 (UTC)

Bot Code Wiki

After a bit of a discussion on irc I've started up the this sourceforge project, thew basic idea is it is an index of the source code for wikipedia bots and their libraries, each page has a description of the code on it, links etc and a copy of the code on a sub page(for bots with multi pal pages just zip them up and I'll upload them to sf) so can bot owners please upload their bots' and and bot's libraries code to the wiki, thanks --Chris  G  10:36, 8 September 2007 (UTC)

The problem with that is that the code is not as easily accessible as it is on an SVN server, which is where I have mine. You can't check it out with one command, you can't diff between different versions (or, you can, but not to a patch file), you can't revert the real code to a given version, etc... I really think an SVN repository is the best option for most bot operators. — madman bum and angel 19:10, 8 September 2007 (UTC)

You store a cashed version of the code on the wiki in case your svn sever goes down or something and then you also have a link to your svn sever so the user can get the latest source, the basic idea of this is that It makes the source easy to find, for example say you where looking on wikipedia for the source of an archive you would first have to search through all the bots to find one and even then there is no guarantee that the bots source will be on a sub page or an svn sever, I just think it make it easier to find code if you use proper categories it makes it much easier. --Chris  G  02:28, 9 September 2007 (UTC)

Perhaps. You'll excuse me if I stick with the Toolserver SVN repository, however. I don't see it going down in the near future, and I find the redundancy unnecessary. — madman bum and angel 03:06, 9 September 2007 (UTC)

Its not ment to be a copy of the toolsever's SVN repostory more like an idex that covers code on the tool sever and other severs. --Chris  G  08:23, 10 September 2007 (UTC)

Well, if it's just meant to be an index, then you wouldn't store a "cashed [sic] version of the code on the wiki", would you? I honestly don't mean to be combative, I'm just not clear on this project's purpose. — madman bum and angel 15:26, 10 September 2007 (UTC)

Reason being for cashed version is in the unlikely event there is a fire at the tool sever or what ever and the svn sever hosting your code goes down there is a cashed version to fall back to, it could happen! --Chris  G  10:12, 13 September 2007 (UTC)

For the record, coding via an SSH connection to Toolserver is so slow that I keep a local copy of my Toolserv SVN repo to code with. I'm 99% sure that most other devs with toolserv accounts do the same thing, making the idea of a third backup site redundant. Oh, and it's cache. Shadow1 (talk) 21:00, 13 September 2007 (UTC)

Yeah, I actually don't log in to the Toolserver. I do everything through an SSH tunnel, and have a local backup of everything... — madman bum and angel 21:57, 13 September 2007 (UTC)

iw for Natalie Glebova

Bots keep adding an iw for ru:Глебова, Наталья Николаевна to this article, which keeps getting reverted with comments that the name is improper, however, there does exist what appears to me as the proper iw article at that location. Obviously, I don't speak this language, so can anyone provide some analysis, please? --After Midnight ⁰⁰⁰¹ 02:14, 15 September 2007 (UTC)

The issue seems to be her patronymic (middle) name. The bots are linking to an article with middle name "Nikolayevna", while it seems that the editors think her middle name is "Vladimirovna". Right now the good article is at "Nikolayevna", with no article at "vladimirovna", even though the article text (at "Nikolayevna") says that her middle name is "Vladimirovna". Somebody at ru needs to move the article to "Vladimirovna" and delete "Nikolayevna" (assuming that this is the correct name), and the bots should get the clue. Staecker 12:08, 15 September 2007 (UTC)

Alex Bakharev has in fact done that on Russian wiki. But I am not sure that the change was right. Here we find the name of her father: Vladimir Slezin. This is how they got the "Vladimirovna", daughter of Vladimir. But if the father's family name is Slezin, how did his daughter get to be called Glebova? I suppose if Nikolay Glebov was her real father and is now dead, calling her Vladimirovna would be OK in Russia, but she does not live there anymore. I will have to check Yandex on this one.--Pan Gerwazy 23:10, 17 September 2007 (UTC)

He's been reversed, using my argument. There seems to be a little edit war going on there. Yandex: I've seen a page stating Vladimirovna, and a page with comment by her grandparents, not mentioning any father switch at all. And a lot of pages about Glebova Natalya Nikolayevna, a Rusian banker... Under these circumstances, I think that Vladimirovna is probably correct: this "Slezin" must be a mistake, and they found Nikolayevna by yandexing without checking the articles. But to be on the safe side, perhaps dropping the father's name altogether would be better - but I suppose that may be asking too much of people accustomed to this naming style. Now trying to re-state the main problem: since January 2007, there has been an ru article Глебова, Наталья with a photo and an ru article Глебова, Наталья Николаевна without photo. Both try to link to the en Natalie Glebova and both authors introduced a number of silly re-directs: eg there is actually a second article entitled Глебова, Наталья which re-directs to Глебова, Наталья Николаевна. How they did it, I do not know, but someone in ru needs to clean up the mess. --Pan Gerwazy 23:51, 17 September 2007 (UTC)

Solved, it seems. Someone found an explanation for the daughter of Slezin being called Glebova (her mother kept her family name and so could pass it on to her daughter). "Nikokaevna" seems to be the result of the mistake I described. That article has been deleted, and the silly redirects with it. The bots should have no problem now...--Pan Gerwazy 14:55, 18 September 2007 (UTC)

Thanks --After Midnight ⁰⁰⁰¹ 00:21, 19 September 2007 (UTC)

CFD discussion that bot owners may be able to answer a question about

There is a discussion at CFD about deletion of Category:IP addresses used for vandalism. The question has been raised whether any bot (presumably an anti-vandal bot) uses this category, apart from the HBC AIV Helperbots when commenting on AIV reports. Does anybody here know the answer? Bencherlite^Talk 21:45, 26 September 2007 (UTC)

Reverting a bot's edits

Is there a way to "undo" all the edits a bot has done in a certain time period? I recently had a situation where User:SatyrBot went a bit amok and managed to make improper edits to the talk pages of >500 articles. The bot was blocked and has since had all it's automated tasks taken off line until I can get it recoded properly, but I'm looking for a way to undo the damage. Suggestions? -- SatyrTN (talk | contribs) 23:28, 3 October 2007 (UTC)

Question are you an admin? β^_command 23:31, 3 October 2007 (UTC)

I have a script that can do any top edits, if you give me a time frame? --uǝʌǝsʎʇɹnoɟʇs 23:59, 3 October 2007 (UTC)

Nope - not an admin. And luckily enough, they are all "top" edits. They're the ones done by User:SatyrBot starting at 03:43, 2007 October 1 (I'm pretty sure that's UTC) until 07:33, 2007 October 1. They're all to talk pages, and some of them have already been reverted - will that be a problem? That would be *MUCH* appreciated! -- SatyrTN (talk | contribs) 02:36, 4 October 2007 (UTC)

Used admin rollback on the ones it could be used on. Edits that could not be rolled back were either the first edit to the page (fine, no duplicates) or edits that were not (top) (for example, because MadmanBot rolled around shortly thereafter, having just been fixed, and consolidated banners...) — madman bum and angel 03:36, 4 October 2007 (UTC)

Merci beaucoup!!! -- SatyrTN (talk | contribs) 21:59, 5 October 2007 (UTC)

Critical - intro and sandbox bot missing

Howdy. The current bot in charge of resetting the Wikipedia:Sandbox and Wikipedia:Introduction (User:OverlordQBot) hasn't been working for 3 days.

We seem to have gone through a number of bots at the Introduction (Tawkerbot2, Sandbot, EssjayBot, AntiVandalBot, and MartinBotIV (!)), and so I'm here to request/beg for a more stable/permanent/efficient solution.

I also asked OverlordQ 2 weeks ago to make the bot a bit more aggressive at resetting the Introduction (because the Introduction was recently changed to direct users to the Sandbox for test edits), but haven't gotten a reply yet (So I presume he is busy IRL).

Please help! --Quiddity 20:09, 30 September 2007 (UTC)

How often should it be reset? This is a mindless task that MadmanBot can pick up. — madman bum and angel 21:21, 30 September 2007 (UTC)

They should ideally be reset -

1. immediately if the 2 permanent lines at the top (eg intro, eg sandbox) are changed.
2. immediately if profanity is added underneath. (I think there is a "bad word" list somewhere? Am not sure about this criteria)
3. every n minutes just to clear away the previous tests. - the sandbox says it gets cleaned every 12 hours. but essjaybot was at one point cleaning the intro every 15 minutes. Personally I don't see a problem with every 15 minutes, and would suggest that.

I'm more familiar with the Intro than the Sandbox(es), but I'd guess that they can be treated the same (and that all the pages using the various sandbox templates, should be treated the same? (eg Template:Please leave this line alone (tutorial sandbox heading)).

Looking through some of the historys, I see that User:Uncle G's 'bot#SANDBOT is doing some reset work (twice a day only) too. I'll leave a note on his talkpage pointing here, in case he has more info. (Strike that, Uncle G hasn't edited since June.)

Hope that helps. --Quiddity 00:54, 1 October 2007 (UTC)

OverlordQbot is was editing again, and has stopped again. I've notified OverlordQ of this thread now too. --Quiddity 02:40, 4 October 2007 (UTC)

I'm not sure what bad words list you're thinking about, but it might be this: User:Lupin/badwords.--Sunny910910 ^{(talk|Contributions)} 21:21, 5 October 2007 (UTC)

Its working except for some very esoteric and rare bugs which I cant fix till they happen so I know exactly what went wrong. I'm fixing them as they show up, and I'll look into automatically restarting the bot when it encounters problems so we wont have outages like this. Q ^{T C} 18:10, 11 October 2007 (UTC)

There is a bot problem on a Pokémon list.

We are having serious issues here, please discuss this on the appropiate place, thanks. Just bringing this up, but I want to discuss it on the WikiProject, since I'm not really watching the noticeboard. TheBlazikenMaster 15:06, 7 October 2007 (UTC)

In affect, they are right... They are probably going to the Nidoran♂ page, and being redirected to that list. Therefore, the interwiki links are being added on that page. (From target page) Reedy Boy 15:29, 7 October 2007 (UTC)

STBotI

User:STBotI operated by User:ST47 made 5 "no source" tagging errors in 15 of my uploads (e.g. [19]), and its approval does not seem to include adding "no source" tags at all.

To be fair, all 5 cases were of similar images with {attribution} tags and so probably represents a single error in logic, but if it is acting outside its mandate then this should probably be stopped while the bugs are worked out. Dragons flight 19:02, 9 October 2007 (UTC)

Support for X0 sandbox

Please add to Wikipedia:Sandbox and its template links to sub-sandbox http://en.wikipedia.org/wiki/Template:X0 and add robot maintenance support to this sub-sandbox to make all one-digit sandboxes present. Wikinger 15:17, 23 October 2007 (UTC)

The "one-digit" must be a reference to the ten Template:X0 through Template:X9 sandboxes. Several sandboxes are provided because template tests often take a while. (SEWilco 15:36, 23 October 2007 (UTC))

I think the point is that {{X0}}, as opposed to the other 9, currently isn't bot-maintained. --ais523 11:00, 24 October 2007 (UTC)

New database information/statistics request area

Hey! The Toolserver administration folks have set up a system [20] where you can request information to be harvested from the database. You do need to make an account on that site, but once you do, select 'create new issue', set the first box to 'database queries' and ignore the second box, click next, fill in a summary and description and if possible a url, then hit submit. We'll see your request immediately and respond as soon as possible. --uǝʌǝsʎʇɹnoɟʇs 10:50, 24 October 2007 (UTC)

Whatever happened to m:Requests for queries? --ais523 10:59, 24 October 2007 (UTC)

Some people like me still watchlist Requests for queries, but some toolserver users prefer tickets because it's easier to assess the status of a request. — madman bum and angel 14:39, 24 October 2007 (UTC)

I prefer jira, since it's easier to keep track of - however I will watchlist that meta page. --uǝʌǝsʎʇɹnoɟʇs 19:25, 24 October 2007 (UTC)

The crux of the problem is...

That bot-owners think that the stuff their bots do is more important than the feelings of the editors whose contributions are tagged, deleted, reverted or otherwise wrecked. Here's a heads up. Wikipedia is written by HUMAN editors, not bots. Without us your bots wouldn't have anything to work on. It's time these things were shut down, permanently. Exxolon 21:53, 6 November 2007 (UTC)

What about the bots that don't do deletion tagging? WikiProject tagging, etc. - what's wrong with those? — H₂O —  22:12, 6 November 2007 (UTC)

Is there a specific bot you are having an issue with? — xaosflux ^Talk 00:32, 7 November 2007 (UTC)

$20 says it's BetacommandBot. ^demon^{[omg plz]} 19:06, 8 November 2007 (UTC)

You would get your money, this user has been anti-bot for a while, because they dont like to follow policy. β^_command 19:08, 8 November 2007 (UTC)

Indeed; he's been happily anti-bot since September. — madman bum and angel 19:10, 8 November 2007 (UTC)

If a bot makes a mistake that is one thing, but if you just don't want a valid tag added then I don't think it is an issue. I don't really know because you gave no specific examples. 1 != 2 19:16, 8 November 2007 (UTC)

The user does not like the fact that a bot is finding and pointing out mistakes that they have made. β^_command 19:18, 8 November 2007 (UTC)

Well, here are some great examples. [21][22]. — madman bum and angel 20:59, 8 November 2007 (UTC)

Looks productive, if I a bunch of my images did not meet standards and were being considered for deletion I sure would appreciate such a notification. 1 != 2 00:30, 9 November 2007 (UTC)

Many may be interested in this graphic by User:Dragons flight which shows how reliable bots are overall:

Article edits divided into categories of editor and types of edits.

Rich Farmbrough, 16:29 13 November 2007 (GMT).

Obviously bots tend to be more reliable as they're under more scrutiny, but I'm not sure these pie charts show that. I mean: It may be true that abotu 1-2% of bot edits get reverted, but that may just be because it's hard to keep up with their edit rate. I'd prefer to see this illustrated with hard numbers rather than percentages. (MacGyverMagic)- Mgm|^(talk) 12:59, 15 November 2007 (UTC)

Damn, those numbers would be big though... Dihydrogen Monoxide 22:25, 15 November 2007 (UTC)

TempDeletionBot revisited

After TempDeletionBot proposal was turned down here and there, I propose another approach: a bot that will create a list of pages from CAT:TEMP for sysops to delete manually, using scoring system. For example:

• Either page is tagged with {{indef}} or derivatives: +10
• No edits for 6 months +5
• Either user page contains word "sock", but no sock tag: -5
• Page is protected: -3
• User has less than 50 edits: +5
• User has less than 20 edits: +7

Thoughts? MaxSem^{(Han shot first!)} 10:54, 9 November 2007 (UTC)

On the one hand, I don't see how this bot could do any harm. On the other hand (speaking as a non-admin), it wouldn't take an admin that long to work this out themselves and make the appropriate deletion, would it? I don't see the harm, but I'm all for only doing something with a clear benefit. Dihydrogen Monoxide 11:22, 9 November 2007 (UTC)

Try visiting CAT:TEMP right now. Choose a page, then: a) check if the user is really indef'd 2) check page history to learn when it was last edited 3) take some time to analyze if it's a sock whose pages shouldn't be deleted. It takes some time (not much, but quite enough if multiplied by the number of temporary pages). MaxSem^{(Han shot first!)} 19:11, 12 November 2007 (UTC)

Does it matter what pages get deleted first? --Erwin85 11:56, 9 November 2007 (UTC)

It doesn't. But splitting lists to pieces allows admins to be more sure about deleting pages with higher scores and therefore it will ultimately save some time. MaxSem^{(Han shot first!)} 19:11, 12 November 2007 (UTC)

Sounds like a good idea. Also give +10 (?) for no edits ever. Rich Farmbrough, 17:18 13 November 2007 (GMT).

As well as giving a score, why not actually list the data: tagged, last edit, sock/sock tag, protected page, number of edits. That would save an admin from having to do the research (except sampling a few pages to verify the accuracy of the bot info), but without an admin necessarily agreeing with the scoring system. -- John Broughton (♫♫) 00:56, 17 November 2007 (UTC)

Antivandal bots

Just as a note to anyone who might be interested, I have posted a rather long comment about antivandal bots here that I would appreciate feedback about. Regards. Will (aka Wimt) 21:10, 12 November 2007 (UTC)

Conversions

Is there a bot that can do conversions, for example form km² to mi²Ctjf83 03:00, 17 November 2007 (UTC)

No. but consider the {{convert}} template. :: maelgwn - talk 01:16, 18 November 2007 (UTC)

You could have a bot apply the convert template, right? Dihydrogen Monoxide ♫ 02:01, 18 November 2007 (UTC)

Hmm - see here and here. There seem to be a few semi-automated tools around that can do it ... :: maelgwn - talk 02:24, 18 November 2007 (UTC)

AssertEdit

The assertedit extension has been installed. This extension allows a bot to set URL parameters which check certain settings. To use it, add &assert=assertion on the end of the edit url. Assertion can be any of user, bot, exists. I'm currently preparing a patch which would allow assertions such as userANDexists, but that is not yet live. All bots should consider using this. --uǝʌǝsʎʇɹnoɟʇs(st47) 18:31, 17 November 2007 (UTC)

But what does this do? --Russ (talk) 19:26, 17 November 2007 (UTC)

See mw:Extension:Assert Edit. MaxSem^{(Han shot first!)} 20:00, 17 November 2007 (UTC)

Nice thingie, but bot API doesn't support it, so it's less useful. MaxSem^{(Han shot first!)} 20:00, 17 November 2007 (UTC)

I wasn't aware that it was possible to edit via API. If that's possible, then yes, it isn't useful, but to screen-scraping bots this is possibly the easiest say to check these things. --uǝʌǝsʎʇɹnoɟʇs(st47) 03:08, 18 November 2007 (UTC)

It isn't possible yet. Once it is possible, I'm planning to do some testing to see how to ensure a bot is logged in. API also doesn't support maxlag right now. — Carl (CBM · talk) 04:11, 18 November 2007 (UTC)

Breaking news - the maxlag issue is fixed in r27596. — Carl (CBM · talk) 13:23, 18 November 2007 (UTC)

Half the code for my bot is in its monobook.js anyway, so it won't damage the wiki if the bot gets logged out. (It's happened once or twice, causing edit windows to come up with no edit commands being given.) --ais523 10:58, 4 December 2007 (UTC)

Bots in C++

Are there any bots written in C++ with published source code? Happy‑melon 15:13, 30 November 2007 (UTC)

I don't know, but I like that idea. What I would like is for something to be added to the {{Infobox Bot}} page that says "Source Code Published", and if yes, then it adds a category. In fact, I'll propose that to the template talk page. Soxred93 has a boring sig 17:58, 2 December 2007 (UTC)

There is a bot written in C++, but AFAIK, there's no common API. For more information, contact this user. --Filip (§) 14:09, 8 December 2007 (UTC)

Coordination between fair use bots

Note: feel free to copy/paste this elsewhere if there is a more appropriate place for this request.

In my efforts of tagging new articles with {{vgproj}}, I frequently encounter articles that contain(ed) a lot of (invalid) fair use images, and are written by very new editors. The talk pages of these editors are often filled with two or more (I've seen twenty) template messages from one of the image license control bots. To the best of my knowledge, three of such bots currently operate: User:STBotI, User:BetacommandBot and User:OrphanBot. While they might do different things, I see notices from the same bot often on different pages. This leads to the following examples I recently encountered myself:

• User talk:Shin Kairi
• User talk:Ewfr Gomulee

And some particularly horrible examples I fished from recent bot contributions:

• User talk:Deesquared
• User talk:N.Flen

This is a bad situation, for three reasons. Firstly, it is biting newcomers. Most of the users suffering from the "copyright notice filled talkpage" syndrome are new. They are not encouraged to edit Wikipedia, and may just have a genuinly bad experience. Secondly, it's just clutter. Thirdly and finally, it is not delivering the right message about the use of talk pages to users. Wikipedia:Bots/Frequently denied bots#Welcome bot is there for a reason: it would be impersonal. Talk pages are for discussion too, not only notification. A user whose talk page has been flooded by bots, perhaps several times, may be less responsive to editors seeking discussion, or offering help.

I propose bots are stopped from adding a second section to a talk page if an image related section already exists. There are marginal differences between the warning templates, but each of them tells the same message anyway. Messages after the first should be simple, and in the same section. Something like:

• Image:Image.jpg has a problem too. Its source is missing/Its fair use rationale is incomplete/It's not used anywhere. -Bot

User:Krator (t c) 19:12, 19 December 2007 (UTC)

Yes I agree it is biting newcomers. I think you idea is a good one. Maybe the bots should make a list of users who have recieved say 5 warnings in a fortnight, to be assisted by another editor. Maybe there should also be a list of users who haven't been welcomed but have recieved warnings as well? -- maelgwn - talk 23:59, 19 December 2007 (UTC)

Actually this is a good reason for a welcome bot. It will ensure that at least a friendly notice is placed at the top of the talk page before any potential notices come up - and the user can read the links on the message. Redrocketboy 00:06, 20 December 2007 (UTC)

How do bots answer questions about editing wikipedia? -- maelgwn - talk 00:12, 20 December 2007 (UTC)

They shouldn't need to ask. If they do, they can use the helpme tag. Redrocketboy 11:14, 20 December 2007 (UTC)

FWIW, there's a welcome bot on Wikiquote and Commons, and maybe other projects too. If you'd prefer users to get a warning as their first message, that's fine with me. I'd rather not though. Redrocketboy 11:15, 20 December 2007 (UTC)

I don't think OrphanBot is leaving warnings on user talk pages. BetacommandBot is about fair use rationale and STBotI is about missing source information, I don't really think there messages are combatible. Saying that, only one of the same message should be given. -- maelgwn - talk 00:12, 20 December 2007 (UTC)

From a technical and Wikipedia-policy point of view, the messages are not the same. But, for the recipient of the message, the message is the same: "You didn't properly follow the guidelines for this image." That's why my suggestion above calls for a generic image warning template, under which the bots can make a nice list of images with the specific problems. User:Krator (t c) 13:22, 20 December 2007 (UTC)

User:Polbot

It appears that User:Polbot has departed from its approved functions. It is curently adding / editing fair use rationale of varrious types of images. I can find no indication of this function on the user page. Additional many of its edits are being reverted as vandalism. It is just not doing an acceptable job in this area. Dbiel ^(Talk) 03:20, 20 December 2007 (UTC)

See User talk:Quadell it has stopped for the moment and he has been asked to revert the changes. -- maelgwn - talk 03:31, 20 December 2007 (UTC)

As documented on my talk page, this drama sprung up while I was out. By the time I got back Polbot had been blocked and her changes reverted. I'm still thinking about how best to proceed, but I won't run her again without getting bot approval (which I should have done in the first place). More later. . . – Quadell ^{(talk) (random)} 13:13, 20 December 2007 (UTC)

Why posting here too? There is a discussion at the talk page, you then posted to WP:ANI. I think the issue is resolved, isn't it? Snowolf ^{How can I help?} 13:29, 20 December 2007 (UTC)

I agree. There will probably be a Bot Approval coming along later. I suggest we all wait until then to discuss this in more detail. Carcharoth (talk) 14:19, 20 December 2007 (UTC)

"Why posting here too?" - Simply to answer this question, it was due to the seriousness of the issue. A bot being given new powers, being run unattended, making errors; no request for such use being made nor any statement of the functions being done posted on the bot's user page. How many more ways can one break the rules regarding the use of a bot account? BUT, this has all been taken care of, apologies have been made and things are back to the way they should be; so hopefully everyone is happy, I know that I am more than satisfied that the situation has been resolved. Dbiel ^(Talk) 17:59, 21 December 2007 (UTC)

Questions before I create a bot

I was thinking about creating a semiautomatic bot using automate 5. To help me fix typo errors. Has anyone done this? Do I need permission?

So I do not ever check a possible error twice after I have deemed it not a error. I will create a database to track these. Has any anyone already done this?

Thanks —Preceding unsigned comment added by Swhite70 (talk • contribs)

No you do not need approval because it is semi automatic and so all edits need to be checked manually. Yes you can do it already with WP:AWB -- maelgwn - talk 07:37, 21 December 2007 (UTC)