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

[Michael Fuhr]

On Fri, Nov 18, 2005 at 09:37:07AM -0000, Mark Cave-Ayland wrote:
> Thanks for getting to this so quickly. You are in fact right - I've been
> developing the transform() patch on a different box than normal and of
> course, it hadn't been configured with --enable-cassert. I've rebuilt and
> applied a fix to CVS for the two SPI warnings and the assertion failure, and
> a CVS checkout now works for me with --enabled-cassert under Linux. Please
> let me know how you get with the latest CVS.

Looks good on Solaris 9/sparc with PostgreSQL 8.0.4.  Regression
tests pass, aside from the trivial differences I noted earlier;
transform() no longer crashes, and it gives the same results as
the 1.0 branch on a sample data set but about 35 times faster.

-- 
Michael Fuhr