ST_Normalize

[strk]

It may be useful to have a function reordering items inside a GEOMETRY in a predictable way so to be immune to ordering changes that do not affect point-set composition.

Such a function is available in GEOS though the C-API.

[robe]

strk - are you going to do this or punt?

[strk]

punt

[nw]

There's a function in GEOS

extern int GEOS_DLL GEOSNormalize(GEOSGeometry* g1);

But no documentation on what it actually does, except for the following from the C++ api reference:

Converts this Geometry to normal form (or canonical form).

Which is not really helpful. However, if this function is what is wanted, it would be pretty easy to expose it to the SQL layer.

[strk]

It's useful to compare expected/obtained result in tests, in case the implementation changes order of vertices. Not urgent at all.

[nw]

What I meant was I can't tell from the documentation if the GEOSNormalize() function actually does what this ticket is requesting. I don't know if "normal form (or canonical form)" is the same as "reordering items inside a GEOMETRY in a predictable way so to be immune to ordering changes that do not affect point-set composition".

[strk]

That's what GEOSNormalize does, yes.

[strk]

lwgeom_normalize is also present in liblwgeom. I'm on this, will be in 2.3.0

[strk]

In 14949:

Add ST_Normalize function

Includes tests and documentation

Closes #1768

[robe]

The docs have ST_Normalized in the function def which is breaking the builds. Since the function should be ST_Normalize.

So getting:

psql:/var/lib/jenkins/workspace/postgis/branches/2.3/regress/00-regress-install/share/contrib/postgis/postgis_comments.sql:314: ERROR: function st_normalized(geometry) does not exist

[strk]

In 14951:

Fix typo in documentation

Closes #1768 again