[postgis-devel] Gserialized Enabled vs Disabled

[Paragon Corporation]

> The new types is more of a mystery, but rather than puzzling over the
differences, is there a correct behavior we want to enforce consistently?

Well our standard has always been to kick out when the answer is obvious.  I
don't see any reason to change that behavior.  True it does give a somewhat
false notion that things like ST_Disjoint understand 
TINs and TRIANGLEs.  Well they sort of do to some degree.  They know if
bounding boxes don't intersect the geometries must be disjoint.

I'm just puzzled why the order of operation has changed is all, since I
assumed it was the same code base except for how they are validating
bounding boxes and so forth.

So I wasn't sure  if the gserialized bounding box check is failing in some
way so it really doesn't know that their bounding boxes don't intersect and
doesn't know "there is nothing to see here"

Thanks,
Regina