Re: Polygon with holes
- To: mathgroup at smc.vnet.net
- Subject: [mg97598] Re: Polygon with holes
- From: Harrie Kraai <hakraai at xs4all.nl>
- Date: Mon, 16 Mar 2009 06:27:46 -0500 (EST)
- References: <gp5fn9$9mc$1@smc.vnet.net> <gp7vsh$1cv$1@smc.vnet.net> <gpda77$jmv$1@smc.vnet.net>
Hi ragfield, I did some tests and you are right. Thanks a lot. HK ragfield wrote: > On Mar 12, 5:42 am, Harrie Kraai <hakr... at xs4all.nl> wrote: >> Thanks for this suggestion. However, (as you are porbably aware) it does >> not work so easily for arbitrary polygons. The connection between inner >> and outer contours is made by a straight line. A general algorithm would >> have to select the points to be connected and make sure that the >> connection itself does not intersect the contours. There should also be >> a check on the orientations. Making two or more holes would complicate >> things further. > > It shouldn't matter whether the lines cross the contours or not, so > long as the exact same line is followed in reverse order at the end > (which my earlier example does). It's possible the implementation of > this algorithm in Mathematica code I provided earlier has a bug (I > didn't test it thoroughly, if you have an example where it fails I'd > be interested in seeing it), but the algorithm works and is completely > general. > > Mathematica's built-in PDF import support appears to use the same > algorithm for converting arbitrary multi-contour PDF paths into > Mathematica Polygon primitives. Take a look at: > > Import["ExampleData/mathematica.pdf"] > > Notice the output for the "e", "a", and "i" characters. The only > caveat (of which I'm aware) of using this algorithm is that if you try > to print a graphic containing one of these polygons to a PostScript > printer, some PostScript printers will interpret the 0-width segment > to be 1 device pixel wide (so lines do appear even though they cover > zero area). > >> As I expected there seems to be no simple facility available. This could >> be a feature request for wolfram. How do I submit one? > > Try this: > > http://support.wolfram.com/submitabug.cgi > > -Rob >