There are several possible scenarios how you can extend the functionality of polymake:

• You want to implement (play with, tune, whatsoever) an algorithm in C++.

  The easiest way is to implement it as a client program. The polymake library release you from many tedious issues having nothing to do with the algorithm itself, such as data persistence, retrieval, parsing, consistency checks, etc. It will definitively decrease your development cycle time.

• The algorithm you have implemented can be seen as an improvement or replacement of algorithms delivered with polymake.

  Then you can integrate it into the rule set which would allow other components of polymake to take profit from it too. For this purpose you need to write a short rule declaring the input and output of your algorithm. The body of this rule should simply call a client function.

• You know about an interesting software package developed elsewhere and want to couple it with polymake.

  You need a new rule again. But as long as this external software isn't aware of the polymake communication method, some alternative data exchange mechanism should be deployed. In most cases it will be an exchange of ASCII data via files or pipes. The routines accomplishing the data format conversion are most conveniently coded in perl. They will comprise the body of your new rule. You will also surely need the perl interface to polymake for it.

• You want to introduce a new property of polyhedra.

  Then you must declare it. Obviously, you should also provide one or more algorithms calculating this property, packed in rules as explained above.