Daniel Lehman dlehman@esri.com writes:
This fixes a crash in PolyPolyline if points are in the same spot
There were timeouts in the Testbot but were unrelated to anything I changed
From 23edf2c2b89b98ef18849806e74b5d77f7889160 Mon Sep 17 00:00:00 2001 From: Daniel Lehman dlehman@esri.com Date: Wed, 25 Jan 2012 13:47:31 -0800 Subject: gdi32/dibdrv: don't add_join if points are coincident in wide_line_segments
if 2 points coincide (dx == dy == 0), wide_line_segments returns immediately in that case, face_1 is not changed. it is not initialized so junk values on the stack are used. when passed to add_join they create a region that causes a huge amount of memory to be allocated. depending on the values, all available/rlimit memory is exhausted and the program crashes. reverting to 1.3.26 works fine
the test case submitted is a drastically reduced, simplest possible form of what our internal test does, based on real map data
You should integrate the test with the other graphics tests, to reuse all the infrastructure and also verify the graphics results. Also there is probably the same problem with the first and last points.