X-Authentication-Warning:	delorie.com: mail set sender to geda-user-bounces using -f
X-Recipient:	geda-user AT delorie DOT com
X-Cam-AntiVirus:	no malware found
X-Cam-ScannerInfo:	http://www.cam.ac.uk/cs/email/scanner/
Message-ID:	<1423411472.6649.7.camel@cam.ac.uk>
Subject:	Re: [geda-user] FOSDEM
From:	Peter Clifton <pcjc2 AT cam DOT ac DOT uk>
To:	geda-user AT delorie DOT com
Date:	Sun, 08 Feb 2015 16:04:32 +0000
In-Reply-To:	<54D67B5B.6060205@neurotica.com>
References:	<1420499386 DOT 3521 DOT 3 DOT camel AT cam DOT ac DOT uk>
	<20150202152654 DOT GA13336 AT cuci DOT nl> <54CFD589 DOT 9040702 AT xs4all DOT nl>
	<CAHBYzfRkn-nJb4JfrDYyaD0WwPrpZvAgi0QdHCusgz185iNoHA AT mail DOT gmail DOT com>
	<CAGde_xN-iNZUvHh-E47kx1EyoPRt1018wWiDwHhYQ9+od+cJwA AT mail DOT gmail DOT com>
	<20150203112631 DOT 3507a0c1 AT Parasomnia DOT thuis DOT lan>
	<20150204054256 DOT Horde DOT Pm1JV8RJbICk9SHvIGwZ7A3 AT webmail DOT in-berlin DOT de>
	<CAOP4iL2stWVCy3WK0=SNu2zAbs8t6B0uyAgFuOnzG8v_MrYNfw AT mail DOT gmail DOT com>
	<CAGde_xN5gs5r_on=HP2RN7cy6E=2EL9eK3cp+sd9BfBaWNLVew AT mail DOT gmail DOT com>
	<20150204193720 DOT Horde DOT 42xUN-NzhCJRWZne-M5eCQ1 AT webmail DOT in-berlin DOT de>
	<90236728-E79D-47C7-BFB1-34140DB85ACB AT sbcglobal DOT net>
	<CAOFvGD4M48Ap=UQzL_T3yzas2rJrNFfxXRUOkOe8gA8J3bQCHg AT mail DOT gmail DOT com>
	<1423323918 DOT 1592 DOT 10 DOT camel AT cam DOT ac DOT uk>
	<CAOFvGD6rwqf79obnSRQ4gyM0h8_5PQ-C0G2guRp8YU80oOAkig AT mail DOT gmail DOT com>
	<1423329222 DOT 1592 DOT 12 DOT camel AT cam DOT ac DOT uk> <54D67B5B DOT 6060205 AT neurotica DOT com>
X-Mailer:	Evolution 3.12.7-0ubuntu1
Mime-Version:	1.0
Reply-To:	geda-user AT delorie DOT com
Errors-To:	nobody AT delorie DOT com
X-Mailing-List:	geda-user AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On Sat, 2015-02-07 at 15:53 -0500, Dave McGuire wrote:
> On 02/07/2015 12:13 PM, Peter Clifton wrote:
> > I'm currently taking time out from the PCB+GL and 3D stuff I want to get
> > merged in order to fix some polygon bugs that are raising their ugly
> > heads. (Long-standing bugs, but perhaps more commonly seen now due to
> > the nm core coordinate switch).
> 
>   ...and thanks for this, by the way!


I'm definitely down the rabbit hole on this one now.

Just to cheer myself up, I read a book chapter yesterday on algorithms
to do boolean operations on 3D BREP solid models. The 3D cases (even for
simple polyhedra, no curves...) makes computing with 2D polygons look so
very simple its almost laughable.

And yet... we hit bugs, because it really isn't that simple.


I think I know the root-cause for the test-case you sent (bad logic in
the contour walk gathering result polygon), and am currently chasing
down the one or two other bugs that its fix exposes. (How to correctly
deal with single vertex touching contours).

The remaining nicety will be fixing the clearance polygon generation to
avoid the rounding errors that created the geometry in my test-cases
that trigger these bugs.

Sometimes I think my efforts here would be better spent writing a solid
test-suite for the polygon algebra, rather than playing whack-a-mole
with these bugs on a case-by-case basis.


Peter

-- 
Peter Clifton <peter DOT clifton AT clifton-electronics DOT co DOT uk>

Clifton Electronics

- Raw text -