Apparently there are many people complaining about the Mercurial
inability to deal with empty directories (and with moving
directories). I didn't really understood exactly the deep woodoo of
this problem, but from the emails exchanges it doesn't look like there
will be a fix for this any time soon.
My point was the people who use Mercurial, can create this file in
their repository once and for all, and they will stop having problems
with this empty directory. And this without propagating fixes for
Mercurial "features" into the main subversion repository.
Anyway, the file is back so I guess this topic is closed :)
On Feb 28, 2008, at 3:03 PM, Jeff Squyres wrote:
> On Feb 28, 2008, at 2:32 PM, George Bosilca wrote:
>> I do not caution this kind of support. Subversion is the official
>> version control system for Open MPI. If one decide to use another
>> version control system, then he/she should add this file in their
>> own repository.
> Not to start a religious war, but... :-)
> I agree that SVN is the official repo of OMPI. But some of us are
> using other SCM's in order to do distributed development (e.g.,
> Mercurial). It allows us to test on a variety of platforms without a)
> lots of commits back to SVN, and b) committing potentially untested /
> broken code back to SVN before it has been tested / fixed.
> Mercurial (and others) are different from SVN in that they are
> distributed by nature (whereas SVN has a singe, common repo). Hence,
> it's not a matter of just committing a single helper file back to your
> top-level Mercurial repo because you have multiple top-level Mercurial
> repo's. The OMPI SVN is the only repo that they all have in common.
>> If the community need this file, then someone should devote and
>> commit it back.
> Er... isn't that what Ralph did? :-)
It came bundled with the ORTE import, while it is obvious it doesn't
have anything in common with orte.
> If it hadn't been named .hg<something> you might not have
> noticed. :-D
> Perhaps we could call it .not-empty or .placeholder or something
> else? I don't see the harm in having this file there.
>> Eventually, the same person should consider submitting a patch to
>> ROMIO developers ...
> Sounds more like a Mercurial bug to me, actually...
> Jeff Squyres
> Cisco Systems
> devel mailing list
- application/pkcs7-signature attachment: smime.p7s