[Top][All Lists]

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

Re: [lmi] Sharing git repositories

From: Vadim Zeitlin
Subject: Re: [lmi] Sharing git repositories
Date: Tue, 17 Mar 2020 23:41:47 +0100

On Tue, 17 Mar 2020 22:28:59 +0000 Greg Chicares <address@hidden> wrote:

GC> I've also pushed commit 0eafcb048a, which adds a script comparing
GC> two different ways of sharing a non-bare repository. Maybe that
GC> shouldn't be done, but we're doing it today, and it kind of makes
GC> sense in a use case where we keep an lmi repository in a server's
GC>   /opt/lmi/src/lmi
GC> and use it as a source of updated scripts to generate
GC>   /srv/chroot/lmi-chroot-name/the-entire-lmi-universe
GC> because we frequently pull to that repo but never push from it.
GC> Anyway, for that case, I'm leaning toward the viewpoint that it's
GC> better to
GC>  - 'git clone' and then fix up the permissions, than to
GC>  - 'git init --shared' and then generate a worktree, in which
GC>      case we still have to fix up the permissions
GC> but I'd be interested to know if you disagree.

 To be honest, I don't think it's important enough to disagree, and the
main argument in favour of using git-init I see here is for consistency: if
other, bare, repositories will be initialized using it rather than
git-clone, it would seem to make sense to use it here too.

 Also, I believe things could be slightly simplified in git-init case by
using git-pull instead of separate git-fetch and git-checkout.

 But, again, I think using git-clone is perfectly fine for non-bare
repositories too.


Attachment: pgprRGF3AQnkt.pgp
Description: PGP signature

reply via email to

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