*** Specific goal: improve transparency, stability,
and confidence in the kernel by building a consistent
loggable exception model.
Background:
The exception handling in the current AI Kernel is
horrid. As the project has evolved I've been putting
in custom exception handling at the package level. And
its been inconsistent at that. A major task that I'd
like to undertake is to create a base object for all
AI Kernel exceptions that all our other, custom,
exceptions can extend. It is my hope that that this
customized exception handler can buy us something. I'd
also like to investigate how other large projects do
exception handling and reporting. Of special concern
is that this software is a "library" type so reporting
to users can be tricky...
Logged In: YES
user_id=52713
Look, exception handling always needs to be better. I think
the best way to deal with it is to start letting certain ones
bubble up, treat them as normal events and have a cell that
can tell the user what is going on. I.e., the system is
supposed to be natural intelligent and fun... shouldn't this be
part of it??