Opened 14 years ago
Closed 14 years ago
#364 closed defect (fixed)
Changes to BufferBuilder.cpp & operations\valid
Reported by: | tomatsafe | Owned by: | strk |
---|---|---|---|
Priority: | major | Milestone: | |
Component: | Core | Version: | main |
Severity: | Annoyance | Keywords: | isvalid, valid, buffer, bufferbuilder |
Cc: |
Description
Safe Software has made changes to GEOS code and would like to commit it back to trunk.
A summary of the changes is below, and a patchfile is attached. We'll wait a few days for any responses.
BufferBuilder.cpp
- Fixed a crash caused by BufferBuilder::bufferLineSingleSided(..)
IsValidOp.cpp, ConnectedInteriorTester.cpp & TestValid2.xml
- Fixed handling of some WKT Polygon EMPTY cases that would previously crash. (As spoken about with Martin Davis, Paul Ramsey, and Kevin Wiebe)
- Added those 2 WKT cases to an existing xml test.
Attachments (1)
Change History (4)
comment:1 by , 14 years ago
comment:2 by , 14 years ago
I've attached another revision where copyright headers now match the style of existing ones and tabbing is respected on a per file basis.
I'd like to commit these changes in tomorrow under Stephen Wong's account as per Paul Ramsey's suggestion.
RE: WKT/POLYGON's, I can't speak for a port to JTS, but Martin Davis was included in the discussion about such cases, and we reached a reasonable conclusion that OGC spec was unspecific in this case, and that this would be a good interpretation. Support for this is that an EMPTY LinearRing (*note, different from LineString) should be closed because EMPTY's should be valid. Thus a Polygon, which consists of LinearRings, should not be made invalid only because it has EMPTY inner-shells.
comment:3 by , 14 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
No more responses after a few days, the changes have now been submitted in revision #3087. Closing this ticket.
Few points: