Re: distributed version control

[John W. Eaton]

On 16-Jan-2008, Thomas Weber wrote:

| I'd say that there should be more than one way to forward patches (e.g.,
| people without a webspace should be able to send patches by mail simply)

Yes, we could still accept simple diffs, but I'm hoping that they
would become the exception.

| Problem here might be Windows.

Mercurial at least has binaries for Windows and OS X systems.  Also
Solaris.

| It should be possible to have a trust model: e.g., if someone suggests a
| patch that's only for Windows, it should be possible to point him to
| Michael. If Michael accepts the patch in his repo, you can simply pull
| from there. This is definitely possible with Git (Linus pulls only from
| a few people).

Yes, it would be nice to move in this direction.

| Also, an easy interaction with a future bugtracker might be nice.

That was my next message (or soon, anyway).

| > Possibilities to consider are mercurial, git, bzr (and possibly
| > others).  Does anyone on the list have experience with these tools?
| 
| I currently use Mercurial, but have also use Git occasionally. One thing
| about Git: it's fast, even on large files (> 100 MB). However, I find
| its interface with the staging area ("index") unintuitive. Other people
| might disagree. Git is written in pure C, this might be a problem on
| Windows.

On my system, a number of the git-* commands seem to be shell scripts,
so it seems that would cause more trouble for porting to Windows than
straight C programs.

| The conversion tools from CVS to other system are quite mature. It might
| be possible to convert the entire history!

I tried "hg convert" on the Octave CVS archive and it did not seem to
work properly.  Now I'm trying to do a cvs -> svn -> hg conversion to
see if that works any better.  But I'm a novice here, so I'm not sure
I'm doing all the right things.

jwe