Default modules to load ... [using git question]

[Sean]

On Tue, 24 Oct 2006 18:43:42 +0200
Matthias Hopf <mhopf at suse.de> wrote:

[...]
> And how do I 100% follow some upstream development? Including new
> branches?
> That is what I would typically want to do in a standard centralized
> working repository environment (which we still have at X.org). I know
> that this isn't git's preferred working environment.

As far as I know, the only way is to manually add branches to your
repo to indicate you care about them.  It may be that xorg is the first
project using Git where more automatic support for following new
branches is needed.  For any projects i've been involved in, new public
branches are rare enough that it hasn't been an issue.

But this is really a good question for the Git list; I may be missing
something obvious.  Please don't take my answers as authoritative.

> > You shouldn't really ever need "git-pull -t"; just use "git pull".  It will
> > automatically grab all the tags you need.
> 
> In that case it didn't. Well, -t didn't help too much either.

Right, it won't help for the case where a commit is made on a branch you're
not following.

> So you're telling me git doesn't support this yet?
> OMFG.

Well of course it supports it; it's just that maybe it could be made a bit
easier.  This would be the type of thing that should be submitted as a feature
request to the git mailing list.

> 
> So basically I have to check 'git-ls-remote -h origin'
> and add missing lines to .git/remotes/origin? Seems like it...
> 
> Now I'm using
> 
>    git-ls-remote -h origin | cut -f 2 | perl -e 'while (<>) { chomp; $h{$_}=$_; } open F, ".git/remotes/origin" or die; while (<F>) { print; if (/^Pull:\s+([^: ]*):/) { delete $h{$1}; } } foreach $f (keys %h) { print "Pull:   ${f}:$f\n"; }' >$tmpgit
> 
> to create a new .git/remotes/origin. Unfortunately, this slows down the
> update process even more, which already is pretty damn slow. The only
> thing of git that is slow, so it's even more recognizable.

Hmmm, strange that you should seeing a slow update process.  For me it's very fast.
I don't read perl so i'm not exactly sure what your script is trying to do, it
looks like you're trying to automatically follow all remote branches?   While
it might have performance issues, it should have prevented the problem you seem
to have hit with not seeing a commit because of an untracked branch.

> I'm on that mailing list for several months now.
> I'm following the development closely.

Okay.  Git questions are much more likely to get better answers there and may
even get new features and documentation added to Git.

> Still I happen to stumble upon seemingly trivial issues every now and
> then. When we made the change from CVS to git I was under the impression
> that git was stable and actually pretty feature complete. Guess I was
> wrong, at least when it comes to centralized development.

It is feature complete, although there are a few rough edges.  The Git
developers are very responsive to feature requests or improvements.
Please don't be afraid to mention even trivial issues on that list.

> Please don't get me wrong - I don't want to work with CVS any more.
> Never ever.

lol.. agreed.

Sean