Colloborating via NFS vs. umask

[Matt Mackall]

On Fri, Dec 21, 2007 at 02:16:12PM +0000, Georg Sauthoff wrote:
> Matt Mackall <mpm at selenic.com> wrote:
> >> > Georg Sauthoff wrote:
> [..]
> >> Btw, yesterday I hacked a umask extension for Mercurial ;).
> >> Basically you can specify via
> >> [umask]
> >> umask = 007
> >> a specific umask for mercurial, or if this is not present,
> >> mercurial chooses a umask depending on the permissions of the
> >> parent directory.
> 
> > There was a patch floating around (from Alexis?) that instead picked
> > up a umask from the permissions on .hg/. I think this was probably the
> > right approach (and something we could enable everywhere), the patch
> > just never got finalized.
> 
> Hm, I don't see why picking up .hg helps. If you have a umask
> like 0077, .hg is 0700 ...

chmod g+rw .hg
 
> Anyway, I would appericiate some movement in this umask area. And
> I could imagine, that I am not the only one, who is working in
> multiuser environments, where this matters. Is there is a ticket-ID
> for this umask patch? I couldn't find one with a quick search.

Not that I know of.

> Btw, it seems just calling os.umask inside the mercurial process
> is not enough for all cases. With my extensions enabled, a few
> files (like .hg/{requires,dirstate,...}) are still created with
> 0600 permissions ...

We don't message with create permissions that I know of. I suspect
experimental error.

-- 
Mathematics is the supreme nostalgia of our time.