[geos-devel] Re: MakeValid in GEOS/JTS ?

[Martin Davis]

Sandro,

Yes,some sort of MakeValid function would be nice to have in JTS/GEOS.  I
actually have some code already for fixing polygons, but it's not
production-ready.  I don't have any timeline for doing this work - no
funding, for one thing!  And lots of other stuff lined up before it.

As for functionality, that's a good point about not being able to fix
1-point linestrings and non-closed rings.  So this would have to remain on
the boundary of JTS/GEOS - such as in the PostGIS MakeValid routine.

The main focus of the work for me would be to make an efficient, robust way
of creating valid polygons from bad data.

Martin



On Wed, Sep 28, 2011 at 8:15 AM, Sandro Santilli <strk at keybit.net> wrote:

> Martin, I've heard you've been considering adding a MakeValid-like
> routine to JTS. When asked to port it from postgis to GEOS I've been
> recluctant to avoid diverging too much from JTS, so I'm interested
> in knowing if you really plan to do that and how.
>
> In particular the PostGIS MakeValid attempts to also fix 1-point
> lines and non-closed polygon rings, which would be impossible in
> GEOS/JTS due to early catch of such malformities. I'm concerned
> that allowing those constructions would open up a can of worms
> all over the codebase being full of assumptions about that level
> of validity.
>
> So, questions in order are:
>  1) Are you planning to provide a function capable to fix all the
>    cases that the PostGIS MakeValid fixes ?
>  2) What is the estimated time by which it would be released ?
>
> TIA
>
> --strk;
>
>  ()   Free GIS & Flash consultant/developer
>  /\   http://strk.keybit.net/services.html
>