Re: Check out doesn't set x-flag on CIFS

From: Erik Faye-Lund <kusmabite@gmail.com>
Date: Nov 20, 2013, 12:41 UTC
Message-ID: <CABPQNSaEs10U+3F8v3tBwMQ6cWa=YO=G2=p6t3VPcCz45OX02Q@mail.gmail.com>
In-Reply-To: <528CAD36.3030407@geneity.co.uk>

Sigh, it seems replying from Gmail on my phone culled the CC-list. Sorry about that, and here's the rest of the discussion in case someone else is interested:

On Wed, Nov 20, 2013 at 1:38 PM, Andre Esser <andre.esser@geneity.co.uk> wrote:

Show 24 quoted lines

> On 2013-11-20 12:23, Erik Faye-Lund wrote:
>> On Wed, Nov 20, 2013 at 12:59 PM, Andre Esser <andre.esser@geneity.co.uk> wrote:
>>>
>>> There is with POSIX extensions, and git recognises it correctly when an
>>> executable file is being committed, ie the file's exec flag is set in
>>> the repository.
>>>
>>> If that file is then however checked out to the same CIFS file system,
>>> the exec flag is not set, which seems a least inconsistent.
>>
>> OK, in that case, core.filemode probably leads down the wrong path. In
>> fact, shouldn't everything just work? I mean, normal POSIX calls
>> should behave as ususal, no? Perhaps this is a Samba issue?
>
> We've had a look at the GIT source code (after my original post) and it
> seems that after GIT identifies the file system as CIFS, it doesn't even
> try to set the x-flag. Given POSIX extensions have been around for years
> it would probably be a good idea to change GIT's behaviour. But I'm no
> developer and we don't have the spare manpower in the company to look
> into this any deeper.
>
> Thanks for your replies though!
>
> Andre