Mailing List Archive
tlug.jp Mailing List tlug archive tlug Mailing List Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]Re: [tlug] Open-source repository question
- Date: Thu, 16 Jul 2009 11:39:03 +0900
- From: John Fremlin <john@example.com>
- Subject: Re: [tlug] Open-source repository question
- References: <5634e9210907141717r18873031s7dfc4dd216c708a5@example.com> <877hyak75b.fsf@example.com> <4A5D556F.2050605@example.com> <873a8yjvhd.fsf@example.com> <4A5D9487.7010604@example.com>
- User-agent: Thunderbird 2.0.0.21 (X11/20090409)
Edward Middleton wrote:That wasn't what I said, and it wasn't my point. Git is better thendarcs because more people know git then darcs.No.In my opinion, the reason git achieved so much adoption despite its insistence that you learn the concepts behind it before you meddle with the user interface is because of its good design.The reason more people use git is because it's better, it's not better because more people use it.[...]soon. Git even has a few defectors among the Haskell crowd. If you areinterested in selling a better DVCS then use darcs,Why do you think darcs is superior. Could you describe any reasons?I would say that the architecture of git is more advanced. Three examples where it is more advanced: compressed database format, the ability to track function moves across files (not sure if darcs has this?), very lightweight branches and stashing.Check out git rebase --interactiveDarcs came out with some talk about how it has a "theory of patches". Unfortunately, the original darcs suffered issues where the repository would become massive confused (!) -- which apparently are only partially fixed in darcs2. This so-called theory (apparently invented by a physicist so inherently suspect ;-) is in its application impractical (i.e. the exponential blowup on conflicts). Secondly, the implementation was massively screwed up (my impression and if you have evidence to the contrary I would be pleased to hear it).http://wiki.darcs.net/ConflictsFAQDarcs1#what-the-big-conflicts-bug Check out this message for a list of the problems with the old version http://www.mail-archive.com/darcs-users@example.com/msg03440.htmlThe approach in darcs2 is extraordinary. They seem to keep adding code to recover from corrupted indexes and corrupted repositories. This is IMNSHO an extremely dangerous approach. Instead of papering over the problems (and overemphasizing the few rare cases where the repository was corrupted by external tools -- there is no way to fix this in general so don't bother), they should carefully make sure that their repository (database) cannot get into a confused state by fixing the bugs in their design which make this possible. This is just my impression from their release notes and I would be very happy to be corrected.I use both darcs and git. I haven't had any problems with either, but my impression is that git has a better theory of patches than darcs, a better database structure and a better design in general.
- Follow-Ups:
- Re: [tlug] Open-source repository question
- From: Stephen J. Turnbull
- References:
- [tlug] Open-source repository question
- From: Jim Breen
- [tlug] Open-source repository question
- From: Stephen J. Turnbull
- Re: [tlug] Open-source repository question
- From: Edward Middleton
- Re: [tlug] Open-source repository question
- From: Stephen J. Turnbull
- Re: [tlug] Open-source repository question
- From: Edward Middleton
Home | Main Index | Thread Index
- Prev by Date: Re: [tlug] See USB drive over network
- Next by Date: Re: [tlug] See USB drive over network
- Previous by thread: Re: [tlug] Open-source repository question
- Next by thread: Re: [tlug] Open-source repository question
- Index(es):
Home Page Mailing List Linux and Japan TLUG Members Links