On 07/13/2011 02:16 PM, Stephen Gallagher wrote:
On Wed, 2011-07-13 at 12:17 +0200, Jakub Hrozek wrote:
> On 07/13/2011 11:57 AM, Alexander Bokovoy wrote:
>> On 13.07.2011 12:56, Jakub Hrozek wrote:
>>> On 07/13/2011 11:09 AM, Alexander Bokovoy wrote:
>>>> Now, there is one concern -- admittedly, my fault as I haven't
mention it:
>>>> str(true) == 'True'
>>>> str(false) == 'False'
>>>>
>>>> so they should be acceptable as well. You have them explicitly denied in
>>>> the unit tests. Do we have any particular reason not to allow them?
>>>>
>>>> This is minor comment, other than that patches look good!
>>>
>>> No reason, my previous incarnation of the patch actually had
>>> strcasecmp(str, "true") -- would that be preferable? Or would it
be
>>> better to only explicitly allow for instance true/True/TRUE?
>>
>> I'd vote for strcasecmp() way (as in UTF-8 that should be the same as
>> for ASCII). I guess this is the case where some tolerance is good. ;)
>>
>
> Agreed. New patches attached.
Nack. Makefile.am is missing reference to sss_python.h, so 'make dist'
doesn't include it (and RPM generation fails).
Thanks for catching this. Attached are new patches that built in koji
just fine:
http://koji.fedoraproject.org/koji/taskinfo?taskID=3196550