Document possible errors from ValidateTopology

[strk]

The manual page for ValidateTopology should contain a table with possible error messages and the meaning of "id1" and "id2" in those cases.

Here's the current set:

coincident nodes | node_id | node_id
edge crosses node | node_id | edge_id # NOTE: should this be swapped?
invalid edge | edge_id | null
edge not simple | edge_id | null
edge crosses edge | edge_id | edge_id
edge start node geometry mis-match | edge_id | node_id
edge end node geometry mis-match | edge_id | node_id
face without edges | face_id | null
face has no rings | face_id | null
face overlaps face | face_id | face_id
face within face | inner face_id | outer face_id

[robe]

Made the change to trunk at r13201. As seen here: ​http://postgis.net/docs/manual-dev/ValidateTopology.html

You have this slated for 2.2. Any reason I can't backport the doco changes to 2.1 (or did you add more error types? )

[strk]

It is ok to backport. As you're at it, the Availability field could be set to 1.0.0 (I checked, 1.0.0 did have this function in the topology.sql file, didn't bother checking older ones)

[robe]

Okay put in version 1.0.0 at r13202 and backported at r13203

I see what you mean about the edge crosses node. To be consistent with the others edge should by id1 and node should be id2.

Since we never documented the order before, I'm guessing people assumed the order followed the ordering of the description so okay to change and just treat the old as a bug. We can put a notice that order changed in 2.1.6 to be consistent with error description.

[robe]

r13203 is version change and r13204 is backport

[strk]

If you're strong about the "thread edge_id/node_id as a bug" argument, could you please file a ticket ? If it's a bug it'll need to be patched in 2.0, 2.1 and 2.2 …

[robe]

okay going to close this out and start a new one.