Re: [PATCH] file-has-acl: new module, split from acl

[Pádraig Brady]

On 28/04/15 12:45, Andreas Grünbacher wrote:
> 2015-04-26 20:06 GMT+02:00 Tom G. Christensen <address@hidden>:
>> On 26/04/15 18:22, Paul Eggert wrote:
>>>
>>> Thanks for reporting that.  I installed the attached patch, which should
>>> fix it.
>>
>>
>> Confirmed, thanks.
> 
> I would rather have gnulib define XATTR_NAME_POSIX_ACL_ACCESS and
> XATTR_NAME_POSIX_ACL_DEFAULT when the kernel headers don't provide
> them: Checking for richacls also requires getxattr, and the combined
> test doesn't help there at all.
> 
> I have updated my changes to include that:
> 
>   https://github.com/andreas-gruenbacher/gnulib
>   https://github.com/andreas-gruenbacher/coreutils
> 
> 
> By the way, this part of commit "file-has-acl: new module, split from
> acl" in m4/acl.m4 is a bit strange:
> 
>>    dnl Set gl_need_lib_has_acl to a nonempty value, so that any
>>    dnl later gl_FUNC_ACL call will set LIB_HAS_ACL=$LIB_ACL.
>>    gl_need_lib_has_acl=1
>>    LIB_HAS_ACL=$LIB_ACL
> 
> Function gl_FILE_HAS_ACL depends on gl_FUNC_ACL to be called
> afterwards, and gl_FUNC_ACL does get called through the gnulib module
> dependencies. Now why is LIB_HAS_ACL being set in both functions?

I'm going to revert this for the moment as it breaks the build
on Mac OS X >= 10.4 which has an incompatible 6 parameter getxattr() call.
Revert commit is attached.

thanks,
Pádraig.

gnulib-file-has-acl-osx.patch
Description: Text Data