make test TESTS=acl.python fails

Matthias Dieter Wallnöfer mdw at
Fri Feb 11 04:20:57 MST 2011


so are you still fully against my changes?


Matthias Dieter Wallnöfer wrote:
> Andrew,
> the wrong integer types prick my eyes when I'm looking over a code 
> part - and I just find it more easy to have this right. But also here 
> the primary reason was to analyse if the mentioned problem (failing 
> ACL tests) could derive from these files - but I do not think so.
> And at the same time I've done these corrections. And please don't 
> forget - I'm not really the only one which does such changes - I have 
> seen a similar patch from obnox applied on s3. But if you are really 
> unhappy then we should forget about this patch.
> The second patch however fixes format specifiers on outputs - which I 
> appreciate to have applied.
> Cheers,
> Matthias
> Andrew Bartlett wrote:
>> Honestly, my answer on the remaining changes is no.  There is just way
>> to much churn with this stuff, and changing from unsigned int to
>> uint32_t seems quite pointless, as is the worry that when walking over
>> an array of at most 64, that we might somehow end up at 2^31.
>> Matthias,
>> I understand your passion for ensuring this is is 'right', but I think
>> we have other things that would be better to spend the analysis on.
>> Andrew Bartlett

More information about the samba-technical mailing list