2.9.6 Known Problems

1. The user input is not checked very rigorously, and some kinds of undetected errors can cause core dumps while other kinds of errors are fatal. This problem will gradually go away as error handling and recovery is improved.
2. The interpreter can be very stack hungry when working on deep terms. This is an unfortunate consequence of the highly modular design of the rewrite engine. Stack overflow usually manifests itself as a segmentation fault with a corrupted core dump; try the UNIX command unlimit stack size before running the interpreter. Stack overflow is also characteristic of nonterminating computations.
3. Heap usage for storing and processing large modules can be quite large. In particular, the memory required for parsing can be quadratic in the size of the (flattened) signature. This problem is magnified because top level modules are not yet garbage collected.
4. Response to control-C can be delayed a very long time when the interpreter is reading in modules or printing results.