Re: Rietveld review

[David Kastrup]

Werner LEMBERG <address@hidden> writes:

>>> This is a serious question I'd like to ask you.  If you were the
>>> king of LilyPond, what would you establish as the workflow?  I'd
>>> really like to hear your opinion.
>> 
>> I'd not prohibit any work flow that can be maintained with
>> single-line git commands.  That includes posting patch series (using
>> git-format-patch and git-mail) and private and public branches.
>> There are graphical tools for dealing with git branches and commits,
>> there is Emacs.  The established development environment for Lilypond
>> already is a GNU/Linux machine (even if it is a virtual one), so the
>> usual caveats about git workflows on Windows machines don't apply.
>
> This basically boils down to give much more interested people write
> access to the git repository under the premise that noone should apply
> non-trivial changes to the trunk but create branches instead which can
> then be merged after review and approval.

I don't think that this is the case.  If you take a look at Linux kernel
development, the canonic repository is maintained just by Linus Torvalds
himself.  After all, the point of git is that it is a distributed
development system without central repository.

The main problem for outside contributors is how to publish their
private branches in case others want to work together with them.  Not
everybody has access to a git server of his own, so a standard hosting
solution would certainly be welcome.  It need not be the same
repository, and I think that per-branch permissions are a possibility
with some git servers that would also reduce possible clashes.

> What do others think?

Maintaining reasonably tight control over the "master" and common public
branches of the central repository should not preclude distributed work
from happening.

-- 
David Kastrup