Possible change to the "git push" behavior

Na'Tosha Bard natosha at unity3d.com
Tue Mar 20 16:52:24 UTC 2012


2012/3/20 Augie Fackler <lists at durin42.com>

> On Tue, Mar 20, 2012 at 11:50 AM, Na'Tosha Bard <natosha at unity3d.com>
> wrote:
> > 2012/3/20 Augie Fackler <lists at durin42.com>
> >>
> >> On Tue, Mar 20, 2012 at 9:42 AM, Martin Geisler <mg at aragost.com> wrote:
> >> > Angel Ezquerra <angel.ezquerra at gmail.com> writes:
> >> >
> >> >> On Tue, Mar 20, 2012 at 2:28 PM, Martin Geisler <mg at aragost.com>
> wrote:
> >> >>> Greg Ward <greg at gerg.ca> writes:
> >> >>>
> >> >>>> On 20 March 2012, Angel Ezquerra Moreu said:
> >> >>>>> It seems that the Git project is considering changing the default
> >> >>>>> behavior for the "git push" command
> >> >>>>> (http://lwn.net/Articles/487131).
> >> >>>>>
> >> >>>>> Since mercurial bookmarks are meant (AFAIK) to behave similarly to
> >> >>>>> git
> >> >>>>> branches, would mercurial change the way bookmarks work if git
> >> >>>>> changes
> >> >>>>> the way git branches work? How would that work given mercurial's
> >> >>>>> backwards compatibility policy?
> >> >>>>
> >> >>>> I don't speak for Matt or for anyone else, but I suspect the answer
> >> >>>> is
> >> >>>> "not bloody likely". "hg update" does not behave the same way as
> "cvs
> >> >>>> update" does, so why should "hg push" behave the same way as "git
> >> >>>> push"? Mercurial is its own project with its own aims and its own
> >> >>>> policies.
> >> >>>>
> >> >>>> That said, the "push all changesets by default" policy could be a
> >> >>>> minor annoyance when using bookmarks.
> >> >>>
> >> >>> I know this is about bookmarks, but as a related data point, I can
> >> >>> mention that Henrik and Sune's repoman wrapper always operates with
> a
> >> >>> -b
> >> >>> flag in order to push/pull the right named branch instead of
> operating
> >> >>> on all of them.
> >> >>>
> >> >>> Last time I spoke with them about it, they said that it would be
> >> >>> impossible (very impractical) for them to push/pull everything by
> >> >>> default.
> >> >>>
> >> >>> --
> >> >>> Martin Geisler
> >> >>
> >> >> Yes, I was referring to the way bookmarks are pushed specifically. I
> >> >> am not suggesting that mercurial's push should be the same as git's
> >> >> push. I just wonder whether the idea is for mercurial's bookmarks to
> >> >> closely follow git's branch behavior. If that is the case and git
> >> >> changes the way it pushes branches, what will mercurial do?
> >> >
> >> > Mercurial's bookmarks were inspired by Git branches in the sense that
> >> > they're pointers to changesets that move on commit. But it's not a
> goal
> >> > to replicate Git branches 100% and users should not expect changes in
> >> > Git to propagate directly to Mercurial bookmarks.
> >>
> >> I'm actually stunned they're even considering changing this behavior.
> >
> >
> > Really?  "Push only the branch I'm working on" is the behavior most users
> > expect, in my experience, and pushing everything at once by default is
> > actually an inconvenience.
>
> Think of all the deployed software out there that depends on the
> existing behavior. This has been the way git has worked for 4+ years,
> it's going to be pretty entrenched.
>

OK, right, from that perspective, I can understand the shock.


-- 
*Na'Tosha Bard*
Build & Infrastructure Developer | Unity Technologies - Copenhagen

*E-Mail:* natosha at unity3d.com
*Skype:* natosha.bard
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mercurial-scm.org/pipermail/mercurial-devel/attachments/20120320/4e24778e/attachment-0002.html>


More information about the Mercurial-devel mailing list