#1482 closed defect (fixed)
ERROR: Negative SRID is not for GEOGRAPHY at character 69
Reported by: | strk | Owned by: | pramsey |
---|---|---|---|
Priority: | blocker | Milestone: | PostGIS 2.0.0 |
Component: | postgis | Version: | master |
Keywords: | Cc: |
Description
Try this:
select 'POINT(0 0)'::geography(point, 0);
Should behave the same as omitting the SRID instead.
Change History (6)
comment:1 by , 13 years ago
comment:2 by , 13 years ago
no it shouldn't. You can't have geography have an unknown. So geography(point) is the same as geography(point,4326)
geography(point,0) is totally illogical and no one should even be trying to doit and should get a nasty message if they do.
comment:3 by , 13 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
I guess the sanest thing is to consider UNKNOWN as omitted, which is using the default of 4326. Done so with r8862.
Remember: 0 == omitted == UNKNOWN
We discussed about 0 being the same as omitted with geometry as well, didn't we ? When I complained about being able to do : 'SRID=4326;POINT(0 0)'::geometry(point, 0);
comment:4 by , 13 years ago
Yes we did and we concluded its NOT the same.
In geometry omitted means you can have a mixed bag of geometries with various SRIDS - no SRID constraint checking is happening
where as 0 means you want it to be unknown.
comment:5 by , 13 years ago
Here's the discussion I'm talking about: ticket #1235, you convinced me of the contrary.
comment:6 by , 13 years ago
ah okay — I stand corrected. I forgot we decided its stupid to apply any spatial constraint to unknown.
I have some doubts about what needs to be done here, as the code has an explicit block added to check for the condition of an explicit UNKNOWN srid being given. See source:trunk/postgis/gserialized_typmod.c?rev=8615#L242