Bad handling of Inf values in GEOS

[pramsey]

See  http://postgis.refractions.net/pipermail/postgis-users/2011-February/028999.html

Older versions of GEOS (pre 3.3) will loop infinitely when handed some coordinates with Inf values in them. Proj4 sometimes generates those values. The solution (?) in this case is to patch PostGIS to not pass the values through to GEOS.

[pramsey]

Inf testing in the GEOS handler

[strk]

That check would make ST_IsValid throw an exception rather than returning false. Note that ST_IsValidReason will tell you it's due to NaN / Inf ...

[pramsey]

Add regression tests and isvalid shortcuts

[pramsey]

Committed to 1.5 branch at r7344, should be committed to 2.0 also.

[pramsey]

This seems to have been applied to trunk also. Closing out.