ST_AsMVTGeom changes line direction

[nextstopsun]

While this may be the expected behaviour for polygons (forcing right-hand-rule), this causes issues when styling or analysing line objects. I.e. road directions get messed.

[robe]

ST_AsMVTGeom is a 2.4 feature not a 2.3.

Bjorn,

Can you see if this is quick to address in a week or 2, if not we'll have to table it till 2.4.1 or 2.5

[Björn Harrtell]

I'm pretty sure I can easily fix this, will report back ASAP and within the timeframe.

[Björn Harrtell]

I think ​https://git.osgeo.org/gogs/postgis/postgis/pulls/16 is the solution but would like CI testing before pushing to svn. Well, would like to not have to push to svn at all I've forgot how to.. :)

[Björn Harrtell]

Ack seems I needed to get it into svn to get proper CI. But indeed having problem doing the git svn dcommit dance (no problem with git svn fetch though), getting "Unable to determine upstream SVN information from HEAD history.".

[Björn Harrtell]

In 15590:

ST_AsMVTGeom only force CW for polys and do not normalize
Closes #3819

[Björn Harrtell]

In 15591:

ST_AsMVTGeom keep the last step normalize for platform consistency
References #3819

[Björn Harrtell]

Managed to get it into svn.. by using svn.

[Björn Harrtell]

In 15592:

ST_AsMVTGeom fix regression tests
References #3819

[Björn Harrtell]

Still appears to be an issue. I kept ST_Normalize as the last step which I didn't think would change orientation but it probably does. Will move that logic to tests where it belongs (was only added to get consistent output for the test cases).

[Björn Harrtell]

In 15605:

ST_AsMVTGeom move normalize to test logic
Closes #3819

[Björn Harrtell]

As a side-effect this should improve performance as the normalization did more work than I expected.