I see CloudCompare uses the Triangle library in order to compute Delaunay triangulations. The CloudCompare license is the GPL v2, but the Triangke one says that
Indeed, for instance the integration of CC into Debian is blocked because of this particular library. The 'Debian masters' argue this is not compatible with GPL as a commercial software wouldn't be able to integrate it while a proper GPL license would allow for it... assuming the commercial software distributes its own code! This is not going to happen soon so it's purely theoretical. And by the way the integration is not strictly forbidden. It's negotiable with the author ;).
I don't really care about Debian, so as long as CC is free, at least Triangle's license is compatible with CC. Moreover we haven't found any replacement for it (as free/portable/simple/efficient).
My point is that if a commercial software asks me tomorrow how they could integrate CC in their own solution (?!) assuming they accept to distribute their own code, they'll help me find a replacement solution ;).
And for people interested only in CCLib (which is LGPL), we added an option in the CMake configuration script to compile it without Triangle.
Well, I understand the Debian reasons. The problem is the possible conflicts with the GPL. But, I agree, Triangle is the best Delaunay library ever. It's superfast. And it is not easy to find a replacement. A classical replacement could be QHULL, but it can't create constrained triangulations (I don't know if CC needs this capability). Another candidate to replace it could be the GNU Triangulated Surface (http://gts.sourceforge.net/). It is an easy to use C library and accepts constraints. The problem could be that it can be a bit slow working with lots of points, which is common with LIDAR or TLS data. But GTS could be a good replacement for Triangle, or at least an option in order to make a totally free version of CC
daniel wrote:I have already tried QHull but it was crashing too much.
And we tried GTS without much success (moreover this library doesn't seem to be supported anymore). So we are patiently waiting ;)
According to some people, GTS is mature for the needs of their developers: https://sourceforge.net/p/gts/mailman/message/31978908/ and the main goal of the project is the GFS solver. I think GTS could be a good option in spite of its relative slowness
I tried some time ago to put in contact with the Triangle's author, in order to ask if Triangle could be free in the future, but he didn't ansrer any of may mails