Jump to content

Talk:Linear hashing

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

"This is discussed later in this chapter."

[edit]

Definitely a book (or chaptered-website) rip. It's certainly not something Google knows about. Should prob. be deleted unless it's from a PD/GFDL source—such text as "This is discussed later in this chapter" (right at the end of the article) scares me. --AnOddName 19:06, 29 October 2005 (UTC)[reply]

I agree. - Bevo 15:48, 23 December 2005 (UTC)[reply]

Not only that, but the material is very strange. Superficially it is correct, but at the detail level it does not correspond to the linear hashing algorithm that I know about. There is no discussion about the incremental table expansion feature, nor the bucket selection algorithm. These are what in my mind an entry should contain about linear hashing. --MegaHasher

Replaced the ripped material with new content. Litwin's original article is on the net, but I just put in the paper reference for simplicity. Do I need to mention the assumption that hash array slot indexing start at 0? All modern programming languages already do that. Pascal is the only exception... --06:04, 2 January 2006 (UTC)

Absolutely no reason to use a power of 2

[edit]

The general approach used in the algorithm relies on the follwing relation: if A = B (mod C), then either A = B (mod 2C) or A = B + C (mod 2C). There's is absolutely no theoretical requrement for the modulo value to a power of two. I.e. the initial table size can be arbitrary. Yes, it will grow by power-of-two factors, but theres no requirement for it to be a power of two itself. —Preceding unsigned comment added by 198.182.56.5 (talk) 23:21, 12 September 2007 (UTC)[reply]

Confusion between dynamic and linear hashing

[edit]

It appears to me that the article (as read 1/4/08) confuses dynamic and linear hashing. The main difference is that dynamic hashing splits an overflowing bucket while LH splits buckets in a fixed order. (Same for merges)

Also notice that the original paper by Witold allowed other moduli than 2.


Tschwarzsj (talk) 23:26, 4 January 2008 (UTC)thomas schwarz, s.j. Santa Clara University[reply]