4. Open Problems

As we have seen, this work is in progress. The functions for conversion among domains between implementations need to be defined . Since there are more than two formats, an universal format is suggested so that it can be converted to other format and vice versa, to reduce the number of needed conversions. Nonetheless, this approach influences the performance in some cases.

Concerning the signatures, there are decisions to make about function names, return codes, exception handlers or argument types. These decisions are dealt with in the prototype developed by Bertrand JEANNET. This prototype will be used to adapt New POLKA's interface, and then experimentations with the new version of New POLKA are planned.

A lot of work remains to be done to construct a common interface. Not treated in APRON are products of domains, dynamic adaptation for higher precision, or switching domain. In fact, a great deal of problems mentioned in our HQ interface also are problems for APRON. Since this is a collective work, the main question will be how to persuade the others, not only APRON members but also the static analyses community, to adopt our technology?