[PATCH v4] vfs_glusterfs: Implement proper mashalling/unmarshalling of ACLs

[Anand Avati]

On 08/12/2013 05:11 PM, Jeremy Allison wrote:
> On Mon, Aug 12, 2013 at 04:48:45PM -0700, Anand Avati wrote:
>> On 08/11/2013 10:23 PM, Andrew Bartlett wrote:
>>> On Sun, 2013-08-11 at 15:59 -0400, Anand Avati wrote:
>>>> Use the primitives available in Samba byteorder.h for implementing
>>>> proper (un)marshalling of ACL xattrs.
>>>
>>> I'll preface this by saying I know how painful it is when every man and
>>> his dog has a different view on how something should be implemented, but
>>> anyway:
>>>
>>> I think this would be better done in IDL.  In general, Samba has moved
>> >from hand-marshalling to IDL for everything that isn't totally silly to
>>> do that way (and even then, we fit those into IDL most of the time).
>>>
>>> Could you try adding this to smb_acl.idl instead, and then using the
>>> generated the pull/push functions?
>>>
>>> This is much less error prone than hand-marshaling, and comes with free
>>> pretty-printers and the ability to trivially manipulate the objects in
>>> python, if you ever need that.
>>>
>>> Thanks,
>>>
>>> Andrew Bartlett
>>>
>>
>> Andrew,
>>   Thanks for your comments. I am new to the IDL syntax used in Samba,
>> and still trying to understand how it works. While that is
>> definitely a better approach and for long term maintainability, this
>> patch is fixing a bug. Can we view them as two separate tasks? I
>> will post the IDL conversion patch as soon as I wrap my head around
>> how that works. In the mean time fixing this bug will be good
>> progress, independently. Please let me know if you agree with the
>> approach, and if so, any comments on the patch itself?
>
> Anand,
>
> 	Whilst Andrew is right (we shouldn't be adding more
> hand-marshalled code into Samba) I appreciate the immediacy
> of the bug fix, and yes we should treat them as two different
> things :-).
>
> At first glance the code looks like it does exactly what
> I'd hoped for - thanks very much !
>
> I should be able to do a more formal review tomorrow.
>

Thanks Jeremy. In the mean time, are there some pointers / docs on how 
the IDL system works in samba.git?

Avati