Integration and Unit Testing
Testing would be a valuable investment after the Python 3 port (90% done as of this writing)
-
Unit tests - must be robust in order to provide the foundation for the arguably more complex work of shoring up the integration tests
-
Integration tests - will require much courage, from the French "with heart"
Integration testing key points:
- Amount of code needs to be reduced to make maintenance easier
- Helper library was originally designed to be run by a person, so there is no programmatic interface
- Failures are hard to debug due to no messages, particularly in the case of timeouts
- Tracebacks are difficult to understand
- Tests fail for spurious reasons (e.g., text doesn't match)
- Better organization and implementation of ad hoc, single-use tools
The removal of Foolscap might make things easier, because its statefulness contributes to tests failing.
No due date
0% Completed
