[PATCH] subrepo: don't push mercurial subrepositories unless necessary (issue3867)

[Angel Ezquerra]

On Mar 26, 2013 4:58 AM, "Nathan Hoad" <nathan at getoffmalawn.com> wrote:
>
> For me it is a usability issue, not so much an issue of how long it
> takes. Some of the subrepos require auth to push, and are not mine, so
> I can't actually push to them. Because of this, I get prompted for
> auth, and can't continue with pushing any of the other subrepos, or
> the parent repository.

My patch series should address your problem as well. It keeps track of
which subrepos have been "modified" versus each of its push/pull sources
and if you have not modified a subrepo it won't be pushed.

Another thing we discussed during the sprint is adding a way to configure
subrepo properties. For example you could tell mercurial that a subrepo is
"read only" and then it would never be pushed.

I have not done anything along those lines yet because I'd rather be done
with the two subrepo related patch series that I have pending first.

Cheers,

Angel

> I should have been more descriptive of the issue in the commit
> message, apologies.
>
> Thanks,
>
> Nathan.
> --
> Nathan Hoad
> Software Developer
> www.getoffmalawn.com
>
>
> On Tue, Mar 26, 2013 at 2:50 PM, Angel Ezquerra <ezquerra at gmail.com>
wrote:
> >
> > On Mar 26, 2013 4:12 AM, "Sean Farley" <sean.michael.farley at gmail.com>
> > wrote:
> >>
> >>
> >> Bryan O'Sullivan writes:
> >>
> >> > On Mon, Mar 25, 2013 at 3:45 PM, Nathan Hoad
> >> > <nathan at getoffmalawn.com>wrote:
> >> >
> >> >> subrepo: don't push mercurial subrepositories unless necessary
> >> >> (issue3867)
> >> >>
> >> >
> >> > This patch makes Mercurial perform a whole lot of extra work if some
> >> > changes must be pushed (it has to discover the set of changes to push
> >> > twice), but it does not reduce the amount of work done if nothing
needs
> >> > to
> >> > be pushed. I believe you are trying to fix a non-bug.
> >>
> >> Well, I don't know about "non-bug" but Angel has a patch series (that
> >> needs some love) that tackles it another way:
> >>
> >> http://markmail.org/message/zt76am36n2yjiio6
> >>
> >
> > Yes, the approach I took was proposed by Matt during the London sprint.
> >
> > My series has already been reviewed by Augie, but I think he wanted
someone
> > else to give his opinion as well. I'm waiting for comments on the latest
> > version that I sent a while ago.
> >
> > BTW, this is a real issue. Push times when you have a fair number of
> > subrepos increase dramatically (proportionally to the number of
subrepos).
> >
> > Cheers,
> >
> > Angel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://selenic.com/pipermail/mercurial-devel/attachments/20130326/343d4e20/attachment.html>