From: Jeff King <peff@peff.net>
To: Andreas Ericsson <ae@op5.se>
Cc: Mikaël <mikael.donini@gmail.com>, git@vger.kernel.org
Subject: Re: upgrading GIT from version 1.7.0.4 to 1.7.6.1.
Date: Mon, 19 Sep 2011 15:00:23 -0400 [thread overview]
Message-ID: <20110919190023.GC26115@sigill.intra.peff.net> (raw)
In-Reply-To: <4E773A71.6020905@op5.se>
On Mon, Sep 19, 2011 at 02:49:53PM +0200, Andreas Ericsson wrote:
> On 09/19/2011 02:30 PM, Mikaël wrote:
> >
> > 1- Is it possible to have two GIT installations pointing on the same
> > repositories?
> >
>
> The core structure and layout of a git repository hasn't changed
> since may 2005, so it should work just fine provided you use any
> git version that actually has a version number.
>
> We've upgraded in a hodge-podge fashion at $dayjob. One of our
> servers is still running 1.4.something. We've never even come close
> to anything resembling a problem. It's actually a bit weird, since
> we started using git in late 2005 and it's so far been the most
> reliable and backwards-compatible piece of software we have in the
> company pretty much ever since.
This is not completely true. Any two versions should be able to
interoperate over the network using the git:// protocol. However, the
disk format has changed slightly over time.
Since v1.6.0 (August 2008), git defaults to using delta base offsets in
packfiles and version 2 of the pack index format. These features are not
understood by versions before v1.5.2 (May 2007) or v1.4.4.5 (July 2008).
A very old git accessing those repositories directly on disk (or by a
dumb protocol like rsync or non-smart http) would have problems[1].
So it has happened[2], but it's important to note that:
1. We waited a year to flip the default for the code supporting it to
become more pervasive.
2. The switch happened on a major version boundary (1.5 -> 1.6),
was already supported by versions in the prior major series (1.5),
and we released a maintenance version for the series before that
(1.4.4.x).
3. The change, along with the affected versions, was mentioned
prominently in the 1.6.0 release notes.
So no, I wouldn't expect any disk format issues moving from v1.7.0.4 to
v1.7.6.1. But it never hurts to read the release notes if you're unsure.
-Peff
[1] I do very occasionally run into this while bisecting some ancient
code on a modern repository.
[2] I suspect a similar thing happened with turning on packed refs
(around the v1.4.4 era?), but I didn't dig around for details.
next prev parent reply other threads:[~2011-09-19 19:00 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-09-19 12:30 upgrading GIT from version 1.7.0.4 to 1.7.6.1 Mikaël
2011-09-19 12:43 ` Sverre Rabbelier
2011-09-19 12:49 ` Andreas Ericsson
2011-09-19 19:00 ` Jeff King [this message]
2011-09-19 21:26 ` Junio C Hamano
2011-09-19 22:44 ` Jeff King
2011-09-19 12:55 ` Jakub Narebski
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20110919190023.GC26115@sigill.intra.peff.net \
--to=peff@peff.net \
--cc=ae@op5.se \
--cc=git@vger.kernel.org \
--cc=mikael.donini@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.