Open m4lvin opened 9 years ago
Malvin - I don�t think the approach taken by hBDD scales up to large computations. At some point you want to be a lot more careful about resource use. I think a monadic interface with an explicit free operation would be the way to go. It may still be possible to support at least some of the convenience of hBDD though, through the use of regions or something.
Also it�s been a long time since I�ve used hBDD, and I�m not surprised there appear to be bugs in there. I�m no longer in a position to sort them out unfortunately.
Sorry, no. I suggest you take it up with the CuDD developers if possible.
For large numbers of variables (>4000) I run into the following error, happening somewhat randomly:
Using null finalizers is a quick and ugly fix, see https://github.com/m4lvin/hBDD/commit/104acf2bae66a7200ee69b4af64781c6a9c7cf4e, but how can we do this properly?