Talk:GNU Hurd/Archive 1

This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

Archive 1

Move to "HURD"?

The word "hurd" also means a part of a fibrous plant (the other part being called "bast"). I think therefore that the OS kernel should be put at HURD, since it's an acronym. -phma 16:06, July 24, 2002‎ (UTC)

More like a backronym -- like Perl, in all official documentation it is spelt "Hurd" not "HURD". The FAQ says that HURD is incorrect, but they won't take their tongue out of their cheek long enough to tell us why. In case you didn't know, "gnu" is another name for wildebeest, so you can probably guess the etymology. The biology usage should be dealt with in the standard way. -- Tim Starling 01:47 May 2, 2003 (UTC)

Update release date?

The article mentions that rms was planning on releasing GNU OS before the end of 2002. I guess this should be updated

Why? if he indeed was planning for a 2002 relase but failed why is that not good information to put there? or have i misunderstood something that you said? Did you perhaps mean that he also planned a 2003 and 2004 release but also failed there? --Ævar Arnfjörð Bjarmason 11:42, 2004 Mar 27 (UTC)

Alfred M. Szmidt, the maintainer of the GNU system, has announced GNU 0.3 for March. Moreover, note there are some 0.x "releases" on ftp.gnu.org. --sdschulze 12:09, 30 Jan 2005 (UTC)

somehow an anually repeated "it'll be released later this year" enters my mind, but I am not sure if this was really Hurd. If so, something like this should be put into the article, because this is some kind of running gag in geek circles. --Deelkar 07:20, 13 Sep 2004 (UTC)

Maybe a link to vaporware would be appropriate. I actually came to this page from that one.

Move page?

I suggest this page be moved, see Talk:GNU/Hurd --Ævar Arnfjörð Bjarmason 14:00, 2004 Mar 27 (UTC)

If it is moved, I think a disambiguation page should be placed here with "Hurd_(plant)" and "GNU Hurd". Markvs 01:57, 13 May 2004 (UTC)

Link?

Hi, I have written an interesting article about Hurd on Wikinerds. It is already linked from Wikinews. If you believe it should be also linked from the Wikipedia article, please feel free to add a link. If you think it shouldn't be linked from Wikipedia, raise your objections now because I will add the link myself after a week or so. NSK 06:04, 12 Feb 2005 (UTC)

I see no problem. Unfortunately, your link had been removed without any comment. I have re-added it. Please answer here before removing it again. This article has made numerous news site and has always been considered as very interesting. --ManuelMenal 00:00, 6 September 2005 (UTC)

Mac OS X's kernel (XNU) is not a microkernel (the article says it is) - it was based on Mach 2.5, not Mach 3.0

Actually the XNU has microkernel Mach. It is not a monolithic OS but microkernel based. 80.248.105.20 (talk) 18:24, 11 June 2009 (UTC)

Confusing statement

“Being built on top of a Microkernel the Hurd inherits the architecture of these systems. This implies that the drivers for the hardware will have to work in the Microkernel space, and probably they will have to be compiled inside.” …what? Sounds like a confusion of microkernels and monolithic kernels, but might be trying to say something sensible in an unclear way. -Ahruman 11:38, 18 October 2006 (UTC)

deleted. i see no sense in either sentence. whomever can come back and be clearer if they were trying to say something WP:notable and sensible. AND/OR hash it out here on the talk page. - Lentower 12:16, 18 October 2006 (UTC)

Vaporware?

No troll, please.

HURD is well known for it abysmally long development time. Should I add [[category:vaporware]] (real question)?
David Latapie (✒ | @) 20:42, 6 December 2006 (UTC)

« Vaporware (or vapourware) is a term applied to software or hardware which is announced by a developer well in advance of release, but which then fails to emerge ».

As the GNU Hurd has had three releases so far, it's not a vaporware. It may some common characteristics — like deception — but it's not the same thing. Manuel Menal 08:14, 7 December 2006 (UTC)

Deception is a strong word. The delay is probably due to overenthusiasm and not anticipating how hard the remaining work will be. Both are common problems in software development. Lentower 11:46, 7 December 2006 (UTC)

Even the latest version of HURD live-cds do not support sound, pcmcia, or USB. Compared to the various open source BSD flavours, and the Linux kernel, the gnu Hurd project might be considered vaporware to rival Duke Nukem Forever. Yes, we have seen releases. Incomplete pre-version-one-point-zero releases that threaten to never reach a point where it is possible to use the Hurd based live-cd as a real computing environment for average desktop users. Hurd seems to me be a toy or proof-of-concept, or educational project, like Minix, or Menuet. Obviously not a neutral point of view, mine. But if anything the article is rather missing out on the huge fact of Hurds complete lack of basic functionality that would render it useless to 95% of people who might be interested in Linux, or BSD and other comparable posix-like operating systems. What market-share among unix-like webservers running apache or another similar http daemon for unix, for instance, does Hurd enjoy? Zero percent? If not vaporware, then at least, "not-used-by-anybody-for-anything-ware". Wpostma —Preceding unsigned comment added by 208.49.111.140 (talk) 17:26, 29 October 2009 (UTC)

I'm not sure that the article as it stands implies that Hurd is a ready-for-prime-time fully-featured desktop OS. Why do you make that assumption? There are lots of other WP articles about obscure, experimental or abandoned OS projects. Letdorf (talk) 13:03, 30 October 2009 (UTC).

Take a look at Project Xanadu and Duke Nukem Forever, the other two contenders for the Academy Award for Longest Gestation Period in a Software Project. Both present their topics similarly to how this article presents HURD: seriously, not as a joke, and allowing the reader to decide for themself if 49 and 12 years are reasonable timelines for the results so far achieved. RossPatterson (talk) 21:11, 1 November 2009 (UTC)

Contradictory

At the end of "development history", it is claimed "there is no focus on L4 any more". However a little further up, and also in the next section it is also claimed that work "efforts are under way to port the Hurd to the more modern L4 microkernel". The Hurd/L4 article says something different again. --130.102.0.178 06:49, 6 February 2007 (UTC)

I think I've fixed this, so I removed the tag. Let us know if it's still contradictory. Gronky 11:58, 6 February 2007 (UTC)

commented discussion moved here

In the article, there was this:

These conversations should be had here on the Talk page. Gronky 11:56, 6 February 2007 (UTC)

We need to improve the information layout

The article was recently found to be contradictary, and I think I know how that happened. Some development decisions are discussed in "development history", and others are discussed in the "architecture" section. This happens because those topics overlap. It can be fixed by choosing two non-overlapping section topics. So all the Mach/L4/Coyotos stuff should be in one section and only one section. If it has to be mentioned in another section, that section should step back from the details, maybe by just noting that the underlying microkernel changed a few times. Gronky 13:34, 6 February 2007 (UTC)

Ok, I've moved stuff around so that each topic is in one place. I might have introduced other inconsistencies while doing so, but I think this puts the article in the right direction at least. Gronky 13:38, 6 February 2007 (UTC)

Relevent url

Don't have time to write up nicely right now :) http://www.coyotos.org/docs/ukernel/spec.html#frontmatter-2.2

—Hobart 17:32, 9 March 2007 (UTC)

Definition MAKES NO SENSE

How can the "GNU" have a kernel? It's like saying its a kernel for the GPL? What does it run on? Brain cells? Really. Is it a clustering kernel for multiple PC's so they run as one, or what? Please clarify! People come here looking for answers, not cryptic clues. —Preceding unsigned comment added by Tunasashimi (talk • contribs) 09:07 22 April, 2007 (UTC)

And it is not a cryptic clue: it is exactly what it means. What you call "GNU" is the GNU project, I guess. The aim of the GNU project is to create a free operating system for anyone: and this operating system is called by the FSF: GNU. Today almost all GNU operating systems use the Linux kernel, thus being GNU/Linux systems. But Linux is not the official GNU kernel, while Hurd/insert_microkernel_here should be, at least in theory. --Cyclopia 07:55, 23 April 2007 (UTC)

Wheras in practice its apparently even incompatible with the latest version of thier own bloody C library! ( http://lists.debian.org/debian-release/2007/05/msg00116.html ) Plugwash 18:23, 18 May 2007 (UTC)

citation of Linus Torvalds

I beleave that this citation has it's place somewhere in the article--80.201.79.196 10:06, 4 May 2007 (UTC)

“

In short, just say NO TO DRUGS and maybe you won't end up like the Hurd people.

”

Can you provide a reference where you found the quote? --Android Mouse 03:26, 29 May 2007 (UTC)

Speaking for myself, I first saw that quote in some Hurd docs I was reading. --Gwern (contribs) 04:06 7 June 2007 (GMT)

The quote can be found here. I don't think it should be included in the article, though. Manuel Menal 08:56, 9 June 2007 (UTC)

Given Linus's history and position as developer of a rival kernel, it would be a bit like quoting the President of Burger King on the quality of Macdonald's products. --Tony Sidaway 20:38, 24 January 2008 (UTC)

I'm against including this quote because it's irrelevant and stupid, but your argument is not valid, however it's a perfectly nice example of an ad hominem argument. Congrats! -- AdrianTM (talk) 21:44, 24 January 2008 (UTC)

Conflict of interest is a perfectly valid argument against inclusion. The statement that Linus is developer of a kernel in no way reflects badly on him, but does prejudice his opinion of rival kernels. --Tony Sidaway 21:46, 24 January 2008 (UTC)

I cook good steak does that impede me to judge a steak that I eat into a restaurant? Does that prejudice my opinion of other cooks? And what's the competition about, if Hurd gets more users Linus is not going to "sell" Linux anymore, it's laughable. -- AdrianTM (talk) 21:53, 24 January 2008 (UTC)

I agree that this discussion is getting ridiculous. Let's leave it there. --Tony Sidaway 22:01, 24 January 2008 (UTC)

Yeah, I agree, but have to apologize for the tone (sorry, I've been cranky today) -- AdrianTM (talk) 22:21, 24 January 2008 (UTC)

You can call it conflict of interest, if you want to characterize it that way, but it could be equally characterized as a witty quote from the fellow, who, with his Usenet buddies, got so tired of waiting for GNU Hurd (and FreeBSD) to get to a released/usable state that he went and wrote a little thing called the Linux kernel. I think that's a very relevant and interesting quote; where else in the article do you get a sense of the low esteem GNU Hurd is held in by a portion of the FLOSS world? --Gwern (contribs) 05:29 25 January 2008 (GMT)

Linus avowedly wrote the Linux kernel as a hobby[1]--he already had Andrew Tannenbaum's minix kernel. I agree that HURD is regarded as, at best, an embarrassing irrelevancy due to its high promises and poor delivery, but there are probably other ways of illustrating this than by quoting a flippant remark by the developer of a rival kernel. --Tony Sidaway 11:52, 25 January 2008 (UTC)

I've redone Slrubenstein's removal of a large section on Torvalds. It isn't appropriate to dwell at length on the opinion of a single person (who moreover is not, in fact, associated with the GNU project in any way, though that isn't the point here in my opinion.) --Tony Sidaway 20:05, 16 February 2008 (UTC)

Linus Torvalds is "a single person" in the same way that Richard Stallman and Andrew Tanenbaum are each "a single person". They're all individuals whose high profile and respected opinions on the topics of their expertise are worthy of consideration. Stallman's picture is the first image in the Linux article and Tanenbaum's critical comments are given appropriate treatment there - there's even a MINIX section that discusses his work. Given Torvalds' stature and his occasional statements that he wouldn't have written Linux if the GNU kernel had existed at the time, I think his comments are appropriate for this article. Factwhen did a nice job (over the course of 11 edits, occasionally reconsidering the text) of summarizing Torvalds' comments, managing to get them down to exactly one sentence and giving Stallman the last word (in fact, more words than Torvalds). Factwhen's Motivation section is entirely appropriate and should be restored ASAP. RossPatterson (talk) 20:33, 16 February 2008 (UTC)

I don't think that "not 100% free" makes much sense, or I guess it depends on how you define "freedom", Linux is covered by GPL v2 license and that's a "free software license" as it was promoted by Free Software Foundation itself. Firmware and binary only drivers are not part of the kernel (and BTW, Linus talked many times against binary only drivers). -- AdrianTM (talk) 20:09, 16 February 2008 (UTC)

I agree there. With the Linux kernel being free this is a non-issue. There are good arguments for a HURD-style kernel but they're primarily technical and (increasingly as Linux thrives and BSD is now free of the infamous advertising clause and thus free too) academic. --Tony Sidaway 20:29, 16 February 2008 (UTC)

RfC: Motivation section

Ross Patterson says:

Factwhen's Motivation section is entirely appropriate and should be restored ASAP.[2]

I've looked at that section. In its last version, it reads as follows:

Linus Torvalds has been at times a prominent deprecator of Hurd, citing the lack of progress on the project. In response, Stallman has asserted that the lack of intellectual property issues and legal liability offered by Hurd will attract a sufficient audience to justify the investment of effort involved.[3]

Trouble is, I've read that cited article and Stallman says nothing of the sort. In fact he only mentions HURD in passing ("In 1992, the GNU system was complete except for the kernel. (Our own kernel project, started in 1990, was going slowly.) In February 1992, Linus Torvalds changed the license of Linux, making it free software.")

Notice here that Stallman actually says that the Linux kernel is free software. He elaborates: "The kernel Linux filled the last major gap in GNU; the combination, GNU/Linux, was the first free operating system that could run on a PC" though he also laments that "But this achievement is precarious. There are hundreds of GNU/Linux distros, and nearly all include some non-free software."

We cannot include a section in the article that misrepresents its source so grossly. --Tony Sidaway 21:02, 16 February 2008 (UTC)

Also, Factwhen is a banned user. banned means they cannot edit Wikipedia. period. If this article can be improved on I am sure plenty of other editors (like Tony) have good ideas about how to do it. Slrubenstein | Talk 21:53, 16 February 2008 (UTC)

As Ross has reinserted the "Motivation" section, I've called a RFC to get some other views on this. --Tony Sidaway 22:35, 16 February 2008 (UTC)

Ross has apparently decided that it doesn't hold up, and removed it. I've closed the RfC. --Tony 07:10, 18 February 2008 (UTC)

1Gb limit

At one time Hurd had a limit of 1Gb per disk partition. I understand that this was due to an early design decision which connected the partition size to the address size of a 32 bit processor (yes, I know that would be 4Gb). This limitation seems to have been removed. It would be interesting to include some information on this story. —Preceding unsigned comment added by 81.187.201.253 (talk) 22:49, 6 October 2007 (UTC)

Hurd as a Kernel?

Kernels don't need kernels. They are kernels. Damncrackmonkey (talk) 20:23, 24 January 2008 (UTC)

The most parts of Hurd run in Userspace not Kernelspace. The part of it in kernelspace supplying the most basic access to hardware is GNU Mach —Preceding unsigned comment added by 84.149.120.137 (talk) 11:25, 9 November 2008 (UTC)

Kernel does not run in userspace. The OS is that part of the software system what runs in kernel space or in supervisor mode http://www.amazon.com/gp/reader/0130313580/ref=sib_dp_pt/103-7158569-1619062#reader-link (page 3. says: On top of the operating system is the rest of the system software. Here we find the command interpreter (shell), compilers, editors and similar application independent programs. It is important to realize that these programs are definitely not part of the operating system, even though they are typically supplied by the computer manufacturer. This is crucial, but subtle, point. The operating system is that portion of the software that runs in kernel mode or supervisor mode.. The monolith kernel is the old OS structure http://www.usenix.org/publications/login/2006-04/openpdfs/herder.pdf what has be used even on Linux. While microkernel is the newer one what's idea is to have a more modular OS than monolithic. The operating system has then a microkernel and needed OS servers (modules) what are moved to user space (or kernel-space). And those servers has a protected process to microkernel itself. Together microkernel and these servers build up a operating system. The microkernel alone is not a OS, but needs those servers (drivers, filesystem, networking etc) and all those together ran in supervisor mode. Monolithic kernel alone is a complete OS. Linux kernel is the operating system because it is a monolithic kernel and it runs alone in supervisor mode. That is the thing what GNU fans miss totally about kernels and operating systems, that kernel can be the OS. And that is the technical reason why there is not such OS than GNU/Linux but just Linux, because none of GNU project code is on Linux OS and their own OS Hurd is totally under development. http://www.topology.org/human/?a=/linux/lingl.html . It is just such big propaganda by GNU project and then there comes conclusions that kernel does have a kernel and operating system does always build up from kernel and other software. All because need for political ideology and it has nothing to do witht he computer science or technology. And all the GNU/Linux propaganda because one guy managed to develop a own open source operating system with GNU tools. You do not call software as GNU/software if you compile it with gcc, that would be just stupid. 80.248.105.20 (talk) 18:43, 11 June 2009 (UTC)

Hurd joke

« In short, just say NO TO DRUGS and maybe you won't end up like the Hurd people. » said Linus Torwald. You can also say YES TO HURD and maybe you will end like clean people. --Eurobas (talk) 09:55, 4 March 2009 (UTC)

Ha ha. Hilarious. 94.193.240.216 (talk) 12:09, 1 November 2009 (UTC)

Recent edit

I felt I had to do something about Golftheman's recent edits:

Hurd is intended to perform the same functions as a traditional Unix kernel, therefore it is more useful to describe Hurd as a kernel, than an operating system (of which the kernel is merely one component). In GNU Project terminology, the entire OS is/was called "GNU".
The D in "Hurd" stands for "Daemons", so using the term "daemon" is appropriate in the intro
Hurd servers/daemons are not the same as kernel modules in the usual sense
the appropriate article to include references for the definition of the term microkernel is microkernel, especially when the reference describes a different microkernel-based OS
the distinction between the Linux kernel, and Linux as an OS has been confused
the term monolithic kernel is well-known and defined; "monolithic OS" is not
Not to put too fine a point on it, some of the English was not to encyclopedic standard; several perfectly good sentences were edited to no benefit at all.
QNX predates Hurd, so we have to qualify Hurd appears to be the first operating system to be implemented in this way with "Mach-based".

Letdorf (talk) 13:31, 4 September 2009 (UTC).

I agree with most of the points, but I am very unsure of Hurd being called a kernel. Hurd servers can easily run in user space, AFAIK. --Cyclopia (talk) 16:28, 4 September 2009 (UTC)

Micro-kernel desicion was not a mistake

I changed the wording of the statement since it says that Stallman has admitted that choosing Mach as the micro-kernel was a mistake but all that Stallman admits is that he was wrong about the time he thought would be saved by choosing Mach. (Tyomero (talk) 00:07, 16 April 2010 (UTC))

hurd 0.401 a real release?

the announcement is at http://www.gnu.org/software/hurd/news/2011-04-01.html , but http://lists.debian.org/debian-hurd/2011/04/msg00008.html would imply that it is only an april fools joke. kgoetz (talk) 09:35, 1 May 2011 (UTC)

It is an April Fool of course: see http://people.debian.org/~sthibault/hurd-0.401/README. Letdorf (talk) 12:04, 1 May 2011 (UTC).

Magic Server

The article includes a list of servers. It describes their function, except for "magic", which is a "(magic server)". What does it do? GNU's official website is baffling. -Ashley Pomeroy (talk) 14:29, 13 June 2015 (UTC)

Unix-like systems expose system and process information as entries in the root filesystem. Some of these entries appear different to each process as they contain information specific to that process. For example if a process opens a file or a pipe (to talk to another process) then these are given file descriptors which show up as /proc/[pid]/fd/[#] for every process (with permission) to see and as /dev/fd/[#] which only that process can see.

My understanding is that since the filesystem is managed by user space servers instead of the kernel, the servers cannot directly examine the state of the process asking for a lookup of /dev/fd/ so instead it is told to generate the result itself. To do this the magic server handles directory lookups for /dev/fd/ and returns a value which tells the process to generate the result itself. The server likely takes it's name from Magic_(programming). [4][5]

-Icosian (talk) 17:32, 31 August 2015 (UTC)

I've taken a stab at improving the definition in the article. -Icosian (talk) 17:55, 31 August 2015 (UTC)

Is The Hurd Dead?

It should be mentioned somewhere in the article that the Hurd is an absolutely dead project, with no use other than as a vehicle for Stallman's politics, which is the only reason its death isn't officially acknowledged.176.83.205.28 (talk) 04:47, 14 January 2018 (UTC)

Sorry, but you will have to wait longer before making funeral arrangements. The last release was on 18 December 2016. --Guy Macon (talk) 13:32, 14 January 2018 (UTC)

Recent GNUmach development/porting

We would need some external sources to add content to the article but at https://tracker.debian.org/pkg/gnumach it's clear that Debian GNU/Hurd development and porting is continuing, e.g. [2019-01-14] gnumach 2:1.8+git20190109-1 MIGRATED to testing (Debian testing watch). The upstream git commit list https://git.savannah.gnu.org/cgit/hurd/web.git/log/ also shows that even if the development team is much smaller than that of Linux, work is continuing as of January 2019. Boud (talk) 16:04, 23 January 2019 (UTC)

Boud, do we need to update Debian GNU/Hurd, which currently says "The current version of Debian GNU/Hurd is 2017, published in January 2017"? --Guy Macon (talk) 17:01, 23 January 2019 (UTC)

See Talk:Debian GNU/Hurd. I saw that using the existing source was enough to update to June 2017. Boud (talk) 20:15, 24 January 2019 (UTC)

Free vs. Open

User:Eman235, regarding this edit:[6] I agree with Template:Infobox OS/doc when it is applied to most software, but this software is from the GNU Project of the Free Software Foundation, an organization that actively hates the term "open source" and insists that what they create is "free software", not open source or free and open source. I say we make an exception on this page. --Guy Macon (talk) 06:18, 19 September 2019 (UTC)

Perhaps. Actually, is there a reason {{Infobox OS}} has a "source model" parameter, instead of just the "license" parameter, like {{Infobox software}}? "License" is explanatory enough, I think—I'd be in favor of getting rid of that parameter entirely. Eman 235/talk 18:52, 19 September 2019 (UTC)

That's a really good solution. I agree that Infobox OS should be made to match Infobox software on this. So how do we make that happen? --Guy Macon (talk) 00:01, 20 September 2019 (UTC)

Testing it in preview on GNU Hurd, it seems that removing the parameter from the infobox code is enough to get the parameter not to display. Removing the parameter entirely from Infobox OS seems fairly simple too, but probably warrants discussion at Template talk:Infobox OS first. Eman 235/talk 17:02, 20 September 2019 (UTC)