Hi Bjoern,

On Monday 03 of August 2009, Bjoern Michaelsen wrote:

> > it is probable that someone pushes between my pull/push, and I'll have to
> > do that once again
>
> Uhm, why do you think you would be required to "do it once again"? Even
> with a normal pull you cant be sure that the repo does not change while you
> are pulling.

Sure, but then you cannot push (push fails), and have to pull once again.  And 
that would be much more expensive => bigger probability somebody pushes there 
something too.

Of course, no problem if you work on the CWS just yourself.

> > Unfortunately not possible, how would you resolve conflicts when somebody
> > pushes to the Mercurial repo, and someone to the git repo?  No, I don't
> > want to bother the RelEng with such a thing.
>
> The same way as if those two changesets would have been pushed to the same
> repo. Each changeset has the old state as parent thus there are now two
> heads where once there was only one. There would be absolutely no
> difference to a single-repo scenario.
>
> Actually, I had that happening during the mercurial pilot already: I had my
> cws hosted at hg.services.openoffice.org. I started a build on
> wntmsci12.pro (which is not my primary dev platform). While it was building
> I did further work on linux and commited locally. Then a buildbreaker
> because of a compiler warning showed up on wntmsci12.pro. I fixed the
> stuff, commited and pushed the changeset. On linux I pull/merged the
> changeset with my ongoing work.

Oh sure, but you did that locally, and had no conflict.  If you had a conflict 
when you did the pull/merge on Linux, manual intervention would be needed 
from you.  On the server, I suppose just an admin would be allowed to solve 
that.

Regards,
Kendy

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to