... .1 microsecons / lookup. We're talking about a case here where ... something that's > quite clearly over-kill. Jenkins hash is fast ... the file as 'public domain' and pushed out. Btw, the reason ... Then I found out that Hsieh code is under totally werdo ...

git - Marko Kreen - Jan 24 2008 - 09:19

... as that's what we're likely to hash, no? >>> If ... something that's >> quite clearly over-kill. > > Jenkins hash is fast ... isn't such a good option. I'm not familiar with ... that Hsieh code is under totally > werdo license (http://www.azillionmonkeys ...

git - Andreas Ericsson - Jan 24 2008 - 12:00

... 't affect the hash value - I'm not writing out hash_utf8_char(), ... if it's a collision, it's often totally different. And ... of your load and - you're going to do literally *thousands ... against the *same* strings over and over (so that the cost ...

git - Linus Torvalds - Jan 24 2008 - 13:15

... that is very simple too. Ok, I pushed the squashed/fixed commit to ... a full patch, with a fixed up subject line etc. You tend to ... do etc), but it re-uses the existing hash.c files, ... name and the index itself totally case sensitive. Signed-off-by: ...

git - Linus Torvalds - Jan 22 2008 - 23:19

... 2008, Junio C Hamano wrote: > > I wonder if the issue Dave ... on-disk size also contains a header that we don't ... account, so it's already "over-estimated" to begin with for ... estimate_cache_size()" really does over-estimate its needs by a biggish amount ...

git - Linus Torvalds - Jan 25 2008 - 12:24

... but also an > identifier for the hash method used. For a reasonably implemented hash algorithm, computing the hash should be cheaper than ... : send the line "unsubscribe git" in the body of a message to majordomo@vger.kernel.org More majordomo info at ...

git - David Kastrup - Jan 23 2008 - 08:28

... also an > > identifier for the hash method used. > > For a reasonably implemented hash algorithm, computing the hash should > be cheaper ... : send the line "unsubscribe git" in the body of a message to majordomo@vger.kernel.org More majordomo info at ...

git - Theodore Tso - Jan 23 2008 - 08:56

... , because we're going to do it sanely. And by "sanely" I mean just ... ) character by expanded character. IOW, only a total *moron* does Unicode name comparisons with strcmp(convert_to_nfd(a), convert_to_nfd(b)); which is essentially what Apple ...

git - Linus Torvalds - Jan 23 2008 - 12:25

... because we're going to do it sanely. > > And by "sanely" I mean ... character by expanded character. > > IOW, only a total *moron* does Unicode name comparisons with > > strcmp(convert_to_nfd(a), convert_to_nfd(b)); > > which is essentially what ...

git - Johannes Schindelin - Jan 23 2008 - 12:34

... shorter, cleaner and > overall more elegant than Linus' code. I am not sure what you meant by that, and what exactly ... list: send the line "unsubscribe git" in the body of a message to majordomo@vger.kernel.org More majordomo info at http:// ...

git - Dmitry Potapov - Jan 27 2008 - 10:48

... -ups, which would help everybody. Two things I noticed (only two, not necessarily because you are good but mostly because I am still mired in day job and could ... line "unsubscribe git" in the body of a message to majordomo@vger.kernel.org More ...

git - Junio C Hamano - Jan 22 2008 - 22:23

... other things that are interesting from a practical PoV. Fowler/Noll/Vo cheap ... buckets, which (with my implementation) means a total hash-table size of 256KiB. ... *= 16777619; h ^= *k; } return (h); } I could provide figures for other table ...

git - Andreas Ericsson - Jan 23 2008 - 04:32

... would be better (pasted below), but I doubt > anyone will ever care, ... . I ran his program on my computer, advantage of his over ... . The new version -- lookup3, which I mentione above, has about the ... git" in the body of a message to majordomo@vger.kernel. ...

git - Dmitry Potapov - Jan 23 2008 - 05:15

... 32 bits as the lookup key, I don't worry too much. ... fairly fast, but obviously at a serious cost in hash goodness, to the point of being totally unusable for anybody who uses ... for now (exact match). Picking a better-researched constant might still ...

git - Linus Torvalds - Jan 23 2008 - 12:06

... better to use 'while' instead of 'if' here, i.e.: Yes, that looks like a good further micro-optimization. Linus - To unsubscribe from ... the line "unsubscribe git" in the body of a message to majordomo@vger.kernel.org More majordomo ...

git - Linus Torvalds - Jan 24 2008 - 15:08

... writes: > > > Basically, I dislike having two copies of the same ... other thing should be recomputed. > > Yes, I agree with that in principle. Storing ... not have > cache-tree for quite a long time until you noticed that ...

git - Johannes Schindelin - Jan 23 2008 - 08:25

... 1 microsecons / lookup. We're talking about a case here where ... more lookups than insertions (unless I'm much mistaken). > > If you ... something that's quite clearly over-kill. > It works always with ... wrote is "good enough". On a side-note, it was very ...

git - Andreas Ericsson - Jan 23 2008 - 10:39

... .1 microsecons / lookup. We're talking about a case here where ... more lookups than insertions (unless I'm much > mistaken). > >> If you ... or.cz/w/pgbouncer.git?a=blob;f=src/ >> hash.c ... something that's > quite clearly over-kill. > >> It works always with ...

git - Luke Lu - Jan 24 2008 - 02:51

... .1 microsecons / lookup. We're talking about a case here where ... more lookups than insertions (unless I'm much >> mistaken). >> >>> If you ... or.cz/w/pgbouncer.git?a=blob;f=src/hash.c ... something that's >> quite clearly over-kill. >> >>> It works always with ...

git - Andreas Ericsson - Jan 24 2008 - 06:24

... process the string in one go. > > > > I believe the ability to add unicode-juggling was a major point > with the patch, ... perhaps Jenkins' isn't such a good option. I don't think you ... git" in the body of a message to majordomo@vger.kernel.org ...

git - Dmitry Potapov - Jan 24 2008 - 12:28