[postgis-devel] Memory alignment issues (was: crash (somehow) resolved)

[Michael Fuhr]

On Mon, Nov 14, 2005 at 05:21:49PM +0100, strk at refractions.net wrote:
> If all regression tests pass successfully on the 1.0 branch
> you might try the HEAD branch - it uses autoconf-based
> configuration which would be another thing to test.

Regression tests passed for both the 1.0 branch and HEAD with the
exception of the differences below.  The first is due to the output
being "Infinity" instead of "inf"; the second appears to be due to
slight differences in floating-point calculations; the third is a
NOTICE message that isn't in the expected output.

 Running Normal Regression Test.
 [...]
 62,63c62,63
 < 62|POINT(inf 0)
 < 63|POINT(-inf 0)
 ---
 > 62|POINT(Infinity 0)
 > 63|POINT(-Infinity 0)
 
 Running OGC Regression Test...
 [...]
 1c1
 < buffer|POLYGON((1 0,0.707106781202421 -0.707106781170674,4.48965921677393e-11 -1,-0.707106781138927 -0.707106781234168,-1 -8.97931843354786e-11,-0.707106781265914 0.707106781107181,-1.34689776503218e-10 1,0.707106781075434 0.707106781297661,1 0))
 ---
 > buffer|POLYGON((1 0,0.707106781202421 -0.707106781170674,4.48965921697616e-11 -1,-0.707106781138928 -0.707106781234168,-1 -8.97931843395232e-11,-0.707106781265914 0.707106781107181,-1.34689776509285e-10 1,0.707106781075434 0.707106781297661,1 0))
 19a20
 > NOTICE:  Self-intersection

The systems I tested were:

PostgreSQL 8.1.0 on sparc-sun-solaris2.9, compiled by GCC gcc (GCC) 3.4.2
POSTGIS="1.0.5CVS" GEOS="2.1.4" PROJ="Rel. 4.4.9, 29 Oct 2004" USE_STATS DBPROC="0.3.0" RELPROC="0.3.0"

PostgreSQL 8.2devel on sparc-sun-solaris2.9, compiled by GCC gcc (GCC) 3.4.2
POSTGIS="1.1.0CVS" GEOS="2.1.4" PROJ="Rel. 4.4.9, 29 Oct 2004" USE_STATS DBPROC="0.3.1" RELPROC="0.3.1"

-- 
Michael Fuhr