I think I'd like to do option (b), leave in the deep check for Inf and put upper level checks for Inf in the IsValid* calls. I've put all the above code in checks for GEOS_VERSION < 33 so that it'll go dark as we move forward. Sound OK?
p. On Thu, May 26, 2011 at 10:42 AM, Sandro Santilli <s...@keybit.net> wrote: > On Thu, May 26, 2011 at 07:26:25PM +0200, Sandro Santilli wrote: >> On Thu, May 26, 2011 at 09:51:12AM -0700, Paul Ramsey wrote: >> > On Thu, May 26, 2011 at 9:44 AM, Sandro Santilli <s...@keybit.net> wrote: >> > > On Thu, May 26, 2011 at 09:26:15AM -0700, Paul Ramsey wrote: >> > > > Erm, do old versions of IsValid still catch Inf geometries safely? >> > > >> > > If you have an HEXWKB I can give it a quick try against 3.1.0 >> > >> > 0103000020E61000000100000005000000737979F3DDCC2CC0F92154F9E7534540000000000000F07F000000000000F07F8F806E993F7E55C0304B29FFEA8554400634E8D1DD424540B5FEE6A37FCD4540737979F3DDCC2CC0F92154F9E7534540 >> >> 3.1.0 was unsupported. >> 3.1.2 entered the infinite loop on ST_IsValid. > > 3.2.3 is good: > > POSTGIS="2.0.0SVN" GEOS="3.2.3-CAPI-1.6.3" PROJ="Rel. 4.7.1, 23 September > 2009" LIBXML="2.7.6" USE_STATS > strk=# select > st_isvalid('0103000020E61000000100000005000000737979F3DDCC2CC0F92154F9E7534540000000000000F07F000000000000F07F8F806E993F7E55C0304B29FFEA8554400634E8D1DD424540B5FEE6A37FCD4540737979F3DDCC2CC0F92154F9E7534540'::geometry); > NOTICE: Invalid Coordinate at or near point inf inf > f > > --strk; > > () Free GIS & Flash consultant/developer > /\ http://strk.keybit.net/services.html > _______________________________________________ > postgis-users mailing list > postgis-users@postgis.refractions.net > http://postgis.refractions.net/mailman/listinfo/postgis-users > _______________________________________________ postgis-users mailing list postgis-users@postgis.refractions.net http://postgis.refractions.net/mailman/listinfo/postgis-users