On Mon, Dec 21, 2020 at 10:18:53AM +0300, Darafei "Komяpa" Praliaskouski wrote:
> Hi,
>
> ST_MakeValid('MULTIPOLYGON EMPTY') crashes (any) GEOS in PostGIS 3.1.0
> after the removal of EMPTY was dropped.
Please check both 3.1 and master branches now. I pushed a fix for this
crash (fully inside
Hi,
ST_MakeValid('MULTIPOLYGON EMPTY') crashes (any) GEOS in PostGIS 3.1.0
after the removal of EMPTY was dropped.
In PostGIS input functions and for Wagyu clipping library there's a set of
fuzzers that make sure the function never crashes.
MakeValid in GEOS may be a good candidate for such a
On Thu, Dec 10, 2020 at 12:16:40PM +0100, Sandro Santilli wrote:
> On Thu, Dec 10, 2020 at 12:08:58PM +0100, Sandro Santilli wrote:
>
> > Odd, for a MULTIPOINT with empty components it seems to strip it.
> > But I cannot add the test because it looks like EMPTY is not accepted
> > in WKT:
> >
>
On Thu, Dec 10, 2020 at 12:08:58PM +0100, Sandro Santilli wrote:
> Odd, for a MULTIPOINT with empty components it seems to strip it.
> But I cannot add the test because it looks like EMPTY is not accepted
> in WKT:
>
> MULTIPOINT(EMPTY,1 1,EMPTY,0 0)
> ParseException: Expected number but
On Thu, Dec 10, 2020 at 01:35:17AM +0100, Even Rouault wrote:
> On mercredi 9 décembre 2020 12:13:17 CET Sandro Santilli wrote:
> > I found out that GEOSMakeValid will remove EMPTY components from
> > collections. The PostGIS implementation of it does not do this.
> >
> > The rationale was that a
On mercredi 9 décembre 2020 12:13:17 CET Sandro Santilli wrote:
> I found out that GEOSMakeValid will remove EMPTY components from
> collections. The PostGIS implementation of it does not do this.
>
> The rationale was that a collection with EMPTY component is NOT
> invalid as per OGC
Hi,
I believe there are two modes functions can operate in: geometry-preserving
or structure-preserving.
MakeValid will try to recover the geometry, but makes no effort to preserve
structure, sometimes it is impossible.
If you're using MakeValid, you expect the structure (the number of
Sandro,
> I found out that GEOSMakeValid will remove EMPTY components from
> collections. The PostGIS implementation of it does not do this.
>
> The rationale was that a collection with EMPTY component is NOT
> invalid as per OGC specification, so why removing them ? Isn't
> that a job for
I found out that GEOSMakeValid will remove EMPTY components from
collections. The PostGIS implementation of it does not do this.
The rationale was that a collection with EMPTY component is NOT
invalid as per OGC specification, so why removing them ? Isn't
that a job for CollectionHomogenize or