[postgis-devel] Parser/Emitter Changeover

[Paul Ramsey]

The changeover has now happened to the new parser and emitter for WKB/WKT.

All occurrences in ./loader and ./postgis have been flipped to the new versions.

Occurrences in cunit have been left in place for now, so that
backwards compatibility tests can still run for a while.

My only concern while fitting them in was that the to_wkt and to_wkb
functions lack an UNPARSER object to drop more verbose error messaging
into. That said, the verbose errors were there mostly for things like
validity checks on the output, which were all disabled. So...? Anyhow,
it's a point of potential improvement.

I think there is a little bit of re-naming to do over the code base,
to change lwgeom_from_wkt to lwgeom_parse_wkt and create a wrapper
lwgeom_from_wkt with simpler parameters. I also need to read the
float_extra_digits GUC in WKT emitters called inside the backend.

Anyhow, I expect the number of broken things has just taken a sharp
leap upwards.

P.

On Mon, Dec 6, 2010 at 2:28 PM, Paul Ramsey <pramsey at opengeo.org> wrote:
> That's an interesting idea, a GUC to move from ISO AsBinary/AsText
> back to the old OGC AsBinary/AsText.
>
> Yes, the parser will happily consume the old style mixed collections.
>
> P.
>
> On Mon, Dec 6, 2010 at 2:23 PM, strk <strk at keybit.net> wrote:
>> On Mon, Dec 06, 2010 at 09:44:45AM -0800, Paul Ramsey wrote:
>>> Yeah, there we go, back in business.
>>>
>>> postgis20=# select asewkt('GEOMETRYCOLLECTIONM(POINTM(0 0 0))'::geometry);
>>>               asewkt
>>> -----------------------------------
>>>  GEOMETRYCOLLECTIONM(POINT(0 0 0))
>>
>> Does it read GEOMETRYCOLLECTIONM(POINT(0 0 0)) back too ?
>>
>> I'm also happy with getting to emitting ISO wkt, just maybe
>> get some session-level settings to switch between one and another ?
>>
>> --strk;
>>
>>  ()   Free GIS & Flash consultant/developer
>>  /\   http://strk.keybit.net/services.html
>> _______________________________________________
>> postgis-devel mailing list
>> postgis-devel at postgis.refractions.net
>> http://postgis.refractions.net/mailman/listinfo/postgis-devel
>>
>