[postgis-devel] PostGIS release

[Mark Leslie]

Well, that's what I get for trying to be fancy.  I'll track down a
better solution for this.  Until this is fixed, the release won't be
made.  I also have a fix for bug 156 (deprecated quote escaping in the
loader), that I need to test further before committing.
I'll continue to try and get the release out and will post a new RC as
soon as I have something for this issue.
Thanks for your debugging Michael, this would have been very difficult
for me to track down without it.
Mark Leslie

Michael Fuhr wrote:
> On Thu, Jul 12, 2007 at 07:49:31PM -0600, Michael Fuhr wrote:
>> ./pgis-test 16384.0001 16384.0001 16385
>> FP_CONTAINS_INCL(16384.000100, 16384.000100, 16385.000000) = 0
>>
>> When A gets large enough, A - PGIS_EPSILON requires more precision
>> than available to be distinguishable from A, so A - PGIS_EPSILON
>> equals A and is no longer less than B.
> 
> I get the above results on the various platforms that I've already
> mentioned.  Here's what I get on a different system (Linux i686):
> 
> ./pgis-test 16384.0001 16384.0001 16385
> FP_CONTAINS_INCL(16384.000100, 16384.000100, 16385.000000) = 1
> 
> ./pgis-test 33554432 33554432 40000000
> FP_CONTAINS_INCL(33554432.000000, 33554432.000000, 40000000.000000) = 1
> 
> ./pgis-test 33554433 33554433 40000000
> FP_CONTAINS_INCL(33554433.000000, 33554433.000000, 40000000.000000) = 0
>