[postgis-devel] ST_OrderingEquals and memcmp

[Sandro Santilli]

On Mon, Nov 17, 2014 at 09:03:49AM -0800, Paul Ramsey wrote:
> There’s only one pad hole, it's for polygons with an odd number of rings. I don’t know why it wouldn’t be consistently zero’ed out, since the pad is set with 
> 
> 		memset(loc, 0, sizeof(uint32_t));

The failure I saw was with point, some test under sfcgal.
But maybe it was my own implementation failing.
Shall I re-try and push the change as soon as it succeeds here ?

Would you like such a dropping-code change ?
In git terms:
 1 file changed, 5 insertions(+), 26 deletions(-)

--strk;

> -- 
> Paul Ramsey
> http://cleverelephant.ca
> http://postgis.net
> 
> On November 17, 2014 at 7:12:49 AM, Sandro Santilli (strk at keybit.net) wrote:
> 
> I was reading the implementation of ST_OrderingEquals and wondered  
> why we're not using memcmp on the serialized memory.  
> 
> Saint testsuite revealed it can sometime fail.  
> I guess the reason for that is some "holes" in the serialized format  
> left for alignment purpose, does it sound correct ?  
> 
> Could we ensure serializers fill any hole to speedup identity checking ?  
> 
> --strk;  
> 
> () Free GIS & Flash consultant/developer  
> /\ http://strk.keybit.net/services.html  

> _______________________________________________
> postgis-devel mailing list
> postgis-devel at lists.osgeo.org
> http://lists.osgeo.org/cgi-bin/mailman/listinfo/postgis-devel


-- 

  ()   Free GIS & Flash consultant/developer
  /\   http://strk.keybit.net/services.html