Subject: Re: [boost] GGL issues - Access Violation
From: Barend Gehrels (barend_at_[hidden])
Date: 2009-11-18 08:34:59
Brandon Kohn wrote:
> I'm trying to run a test that will take the boolean difference of two
> polygons, but haven't been able to find an explicit difference
> function. So I've tried instead to get the difference by taking:
> A & !B (the intersection of A and the negation of B, where negation is
> done by reversing the winding.) Anyway, when I tried this I got a
> crash when the winding was reversed on B. I've attached a text with
> the test.
Thanks for stressing our library to its limits. The exception you
reported does happen indeed if you input a reversed polygon, invalid
input. On the normal case it didn't throw (but results are, also then,
not correct). I know that it should be handled correctly, also in
reverse cases. The exception is solved now. That will not say that the
numerical FP behaviour (probably why you test this) is solved, I'm
investigating this further today. As there are also various other
messages to be answered, I come back to this, probably tomorrow.
Anyway, I'm glad with this case.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk