[postgis-users] isvalid - fixing for self-intersections "fixes" ring self-intersection too

[fiorenontaglio]

hello.
i'm using postgis 1.3

i had some problematic polygons in the same feature class:
   some with ... NOTICE:  Self-intersection at or near point ...
   some with ... NOTICE:  Ring Self-intersection at or near point ...

I ran poor man's fixing for self-intersections, as you know:

 update ... set the_geom = st_buffer(the_geom,0)

The question is:
if i check geometry validation with isvalid(the_geom) after updating, no records are found.
Why? The ring self-intersection is still there but does not raise notice now.

Second question:
Why postgis have sometimes problems running an intersection and sometimes not?
For example I can normally execute st_intersects and st_intersection with these invalid features (before updating). But maybe with ohter feature classes it would be in trobules.
Maybe the borders? (no borders are intersecting the "ring-self intersection" problematic zone)

Thanks
toni