[postgis-users] GEOS crashing PostgreSQL?

[strk at refractions.net]

On Wed, Oct 25, 2006 at 10:44:39AM +0200, strk at refractions.net wrote:
> On Tue, Oct 24, 2006 at 11:12:11PM -0600, Michael Fuhr wrote:
> > On Tue, Oct 24, 2006 at 09:53:09PM -0600, Michael Fuhr wrote:
> > > I saw that ptarray_from_GEOSCoordSeq() was passing a size_t *
> > > as the second argument to GEOSCoordSeq_getSize() and
> > > GEOSCoordSeq_getDimensions(), but those functions expect
> > > unsigned int *.
> > 
> > I see that this has changed between GEOS 2.2 and 3.0.  My 64-bit
> > build uses GEOS 2.2, where the abovementioned functions expect
> > unsigned int *.  In GEOS 3.0 they expect size_t *.
> 
> Argh! did I really break the CAPI ?! I want to die :'-(
> Luckly GEOS 3.0 is only at RC stage, so we might revert
> that change and consider this as a bug in postgis..
> Would that work ?

A quick look at differences in 2.2 branch and 3.0 branch
show that 'unsigned int' => 'size_t' occurred in a lot
of places... guess I'll have to revert them all.
Does anyone see an alternative ?

--strk;