Mercurial development workflow

[Benoit Boissinot]

On 7/29/07, Martin Geisler <mgeisler at mgeisler.net> wrote:
> [snip]
>
> From the mercurial-devel mailinglist I see four
> repositories in play:
>
>   mercurial: http://www.selenic.com/hg/
>   hg-stable: http://www.selenic.com/repo/hg-stable/
>
>   mercurial/crew:        http://hg.intevation.org/mercurial/crew/
>   mercurial/crew-stable: http://hg.intevation.org/mercurial/crew-stable/
>
> By downloading the repositories and looking at them with 'hg view' hg
> and crew seem identical, and likewise for hg-stable and crew-stable.
> The hg-stable repository seems to be a subset of the hg repository
> (likewise for crew-stable and crew).
>
> Is this done by having people push stable and safe changes to the
> {hg,crew}-stable repositories and then pulling these changes into
> {hg,crew} whereas more unsafe changes are made in the {hg,crew}
> repositories?
>
Bugfixes only changes go to -stable, crew is maintained by various dev
(see http://www.selenic.com/mercurial/wiki/index.cgi/CrewRepository)
and Matt regularly pulls from it (resp. crew-stable to hg-stable and
crew to hg).

> What if a change is made in hg and needs to be applied to hg-stable as
> well? Do you then manually apply the patch to hg-stable?
>

Yes, but usually the fix goes to -stable directly and is then merged
back into the dev branch.

regards,

Benoit