[postgis-users] trying to debug segmentation violation

[strk at refractions.net]

On Wed, Sep 21, 2005 at 10:03:34AM -0600, Jessica M Salmon wrote:
> A plpsql function I wrote is starting to occasionally kick me off with a
> signal 11 error code. It seems to occur while calculating a 300m buffer
> around the transformed point
> transform(geomfromewkt('SRID=4326;POINT(-118.335235595703
> 36.8628845214844)', 102008)  When I do this at the command line (from the
> same geometry in memory as the function was accessing), it gives a result
> just fine. The function calculates thousands upon thousands of these just
> fine, but objects to this one and one other.
> 
> I have two questions:
> 1) How can I obtain more information on where exactly the segV is
> occurring? I know where it happens in my code, but since it seems to happen
> in a geos function, I cannot use print statements to get any more detail.

You can run the postgres binary under gdb:
	pg_ctl stop
	gdb postgres
	gdb> run mydb < mytest.sql

BTW, that query does not involve GEOS.

> 2) What are the common causes of this sort of error?

Bugs :)

It would help reducing the test complexity (is it the geomfromewkt()
or the transform() function ?).

Also, if not using latest release it's worth reading the CHANGES file
(1.0.4 release contains a segfault fix in the transform() function)

--strk;