Right now, the mechanism for reporting an error from server to host is
to use T_EXCEPTION messages. However, there don't appear to be any
defined other than the "unknown RPC" or equivalent.
Is there any proposal to systematically define a set of errors? I can
think of:
- out-of-memory
- invalid argument
- out-of-resources
just for starters.
Exceptions appear to return a string and an integer. Is this always
true, or just an artifact of the current implementation(s)? Do the
string/integers have any semantic content?
Thanks
Mayan
