Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 13 Sep 2011 16:19:21 +0400
From:      Lev Serebryakov <lev@FreeBSD.org>
To:        Andriy Gapon <avg@FreeBSD.org>
Cc:        hackers@FreeBSD.org
Subject:   Re: Which way for developing multi-commit feature for FreeBSD is proper at these days -- ask for perosnal space on SVN or use git-svn?
Message-ID:  <1483956987.20110913161921@serebryakov.spb.ru>
In-Reply-To: <4E6F432E.2000600@FreeBSD.org>
References:  <1093822073.20110912175151@serebryakov.spb.ru> <4E6E49C0.1030609@FreeBSD.org> <1638258448.20110913135932@serebryakov.spb.ru> <4E6F432E.2000600@FreeBSD.org>

next in thread | previous in thread | raw e-mail | index | archive | help
Hello, Andriy.
You wrote 13 =F1=E5=ED=F2=FF=E1=F0=FF 2011 =E3., 15:49:02:

>>  But I can not figure out how to have git-svn repo which could be
>> cloned (and used as point-of-rendezvous) as git repository -- it
>> doesn't work "out of box," as I demonstrated in other message :(
> There is a certain contradiction in what you tried to do.
> git development model is more suitable for pulling than pushing and what =
you
> wanted to do is more naturally done with pulling than with pushing. After=
 all,
> there must be a person who decides what should be committed to svn, that =
person
> decides what changes to pull from other developers and how to commit them=
 to svn.
   I don't want to commit to svn often (tens of commits), and could do
 this with svn itself later, without git-svn at all. It is why I don't
 want to work with FreeBSD svn repo directly -- I will need many
 commits which are "bad", partial, without meaningful message, etc. :)

> You tried to implement the workflow in the opposite direction - each of t=
he
> developers may decide what he wants to push in the svn-facing repository.=
  This
> doesn't look like a correct workflow to me.
  I'm the only developer (yet), but with two workstations and each
workstation could not see other one (and they both could not be seen
from server with "central" repo -- NAT and corporate firewalls with
application-level proxies) -- so they could "push" and "pull" to
central repository, but not from each other and central repository
could not initiate "pull" from them.

> If you really want to have some central repository for pushing, then you =
can
> create another repository.  That repository should be bare and should be =
used only
> for pushing of local changes from the developers.  Then, the git-svn meis=
ter would
> pull changes from that repository.  But I haven't thought through this ki=
nd of
> workflow, especially rebasing the central repository on top of svn change=
s.
  It seems, I need more deep understanding of git model and
 limitations. It is much more unobvious compared to Mercurial (hg),
 which seems very natural for me :)

> P.S. Is this really a topic for developers@?  This is a normal technical
> discussion with no secrets from public, so I think that hackers@ would be=
 quite
> appropriate for it.
  Ok, I've redirected it.

--=20
// Black Lion AKA Lev Serebryakov <lev@FreeBSD.org>




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1483956987.20110913161921>