[postgis-devel] current_parser_check_flags

[Mark Cave-Ayland]

Paul Ramsey wrote:

> - In 2.0 (and in 1.5) the lwgeom_in functions are called with
> PARSER_CHECK_ALL however,
> - In 2.0 (and in 1.5) the checks are not actually run.

...due to an error in Mark's parser branch merge. I see this as being a 
mistake rather than a conscious design decision.

> - In 2.0 there is now a clean.sql regression tests that loads in
> polygons with unclosed rings (and then presumably cleans them).
> Because my new WBK reader both declares PARSER_CHECK_ALL and actually
> checks, I cannot pass this regression.
> 
> (This is the last regression I have to pass / alter before I can commit back.)

I'm still of the mind that we should *never* accept polygons with 
unclosed rings, so I'd remove these parts of the regression tests. I'm 
fairly sure I posted about this when I wrote the code originally, but 
here's what worked well for me:

- WKT and WKB input should be checked with all flags
- WKT and WKB output should have no output checking (to allow people to 
dump databases with invalid geoms)
- Ring closure should only check in 2 dims, as GEOS 2.5D calculations 
can produce unclosed rings in the Z plane


ATB,

Mark.

-- 
Mark Cave-Ayland - Senior Technical Architect
PostgreSQL - PostGIS
Sirius Corporation plc - control through freedom
http://www.siriusit.co.uk
t: +44 870 608 0063

Sirius Labs: http://www.siriusit.co.uk/labs