Comment 11 for bug 1100620

Could you let me know the SHA1 of the commit you are running from please.

I think this is as a bug, as the polygon should be broken down into triangles for rendering... and we don't need a huge buffer to render those, we just flush when it gets full.

I don't remember the exact behaviour of each version of the renderer, but I'd be surprised to see it genuinely fall-over, even for extremely complex boards.

I have a test-case here someone sent me with 8 layers, and 9x Spartan FPGAs on a PCI card! (And it renders just fine).

Can you get the bug to reproduce with a cut-down board? I'd like to confirm exactly which geometry is causing it.

I'll be busy for a bit, but if I can send you a patch to PCB to get some debug info out - I'd appreciate the info.

Perhaps you could run PCB inside gdb, and set a breakpoint on the exit function, and attach a backtrace from when it dies.

gdb --args pcb [designfile]
(gdb) break exit
(gdb) run

....

Breakpoint 1, __GI_exit (status=status@entry=0) at exit.c:99
99 exit.c: No such file or directory.
(gdb) bt

.... (Paste this into a file and attach here).