[postgis-devel] in_gml

[Paul Ramsey]

On Sun, Nov 21, 2010 at 11:33 AM, Olivier Courtin
<olivier.courtin at oslandia.com> wrote:

> But when i reproduce it, i also got 2 others regress:
> measures and tickets, with for instance:
>
> ERROR:  getPoint_internal called outside of ptarray range (n=0,
> pa.npoints=5, pa.maxpoints=-1)
> On 3dDistancetest2
>
> Do you think this could be linked ?

Maybe, maybe not... the error is handy, it's pointing out places where
POINTARRAYs are being created by hand, so that the maxpoints variable
is not being set. This is another place were things have to be tracked
and forced back into the ptarray API. Hopefully that will mostly fix
things, though if they are doing append/insert points against
POINTARRAYs backed by serialized_form generated by the old WKT or WKB
handlers, we'll have other issues.

>> I think the final solution will be pulling the in_* code into
>> liblwgeom where we can properly valgrind it in isolation.
>
> Moving to liblwgeom should be anyway a good idea for
> the  import functions.
> But for GML one i met architecture issue before, see #444 comments.
> Thoughts welcome !

I'll have a looksee.

>> Are you changing over the bbox (from BOX2DFLOAT4 to GBOX) in this pass
>> too? If not, that should be your next goal, IMO.
>
> Yeap that was done,
> Now we have GBOX in each LWGEOM.
>
> Next step on this, will be to serialize GBOX
> And then to remove all box2dfloat stuff.

Note that there are already lwgeom->gserialized converters, though
they at the moment two-step around the fact that the gbox available in
gserialized is not there in lwgeom. So those can be upgraded to
properly fill-in/read the GBOX in lwgeom now. At that point we can
consider flipping that actual serialization over from serialized_fom
to GSERIALIZED.

P.

> --
> Olivier
> _______________________________________________
> postgis-devel mailing list
> postgis-devel at postgis.refractions.net
> http://postgis.refractions.net/mailman/listinfo/postgis-devel
>