glob2-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [glob2-devel] Hg cleanup status


From: Michael [Plouj] Ploujnikov
Subject: Re: [glob2-devel] Hg cleanup status
Date: Tue, 25 Nov 2008 18:52:45 -0500

On Tue, Nov 25, 2008 at 6:33 PM, Stéphane Magnenat
<address@hidden> wrote:
> On Wednesday 26 November 2008 00:25:56 Michael [Plouj] Ploujnikov wrote:
>> On Tue, Nov 25, 2008 at 6:17 PM, Stéphane Magnenat
>>
>> <address@hidden> wrote:
>> > On Wednesday 26 November 2008 00:12:49 Kai Antweiler wrote:
>> >> You are absolutely right. We ignore the undering concept - as we often
>> >> do. We have only a hand full of active developers. Therefore it isn't
>> >> that bad.
>> >>
>> >> If some users already have those revisions locally (which is unlikely
>> >> because they aren't in the default branch), it won't hurt.
>> >
>> > As far as I know, updating or cloning the repository does clone
>> > everything. So this is not clean :-(
>>
>> You are right that cloning a repository essentially creates copies of
>> all of the files that ever existed in the history of a repository. The
>> older files are stored in this directory, I think: .hg/store/data/
>> (http://hgbook.red-bean.com/hgbookch4.html)
>>
>> Can you explain what is wrong with keeping the object files in the
>> history and spreading them on other people's computers even if they
>> are hidden somewhere in .hg/store/data/?
>
> It just consumes useless bandwidth and storage space forever, and it is not
> clean to do so.

This is what you get for using a distributed Software Configuration
Management (dSCM) system like Mercurial. It's a property of the system
that all old files will be present in every clone of the repository.
It is also a fact of life that people are not perfect and might
occasionally commit unwanted/unnecessary/wrong files. The dscm will
simply continue to keep those files in the history. Otherwise, what
good is a revision history? To conserve bandwidth and space it is best
to avoid dscms completely and use a centralized system like SVN or try
something that allows lazy pulls/clones
(http://lists.osuosl.org/pipermail/darcs-devel/2007-April/005540.html
http://marc.info/?t=120249209700001&r=1&w=2). Also, forcing people to
re-clone large repositories goes against the idea of conserving
bandwidth. That is why a lot of Git gurus scream at novices for
deleting their local copies!



-- 
Michael Ploujnikov
http://plouj.com/

reply via email to

[Prev in Thread] Current Thread [Next in Thread]