Comment 4 for bug 405732

Hi Eric,

thanks for the feedback. Seems as we've completely missed the fact that a WORK_FAIL is expected after a WORK_EXCEPTION and just implemented it like a fail carrying an exception message (string!) for the ruby lib [*]. Your proposal seems more robust and compatible to me now, so I'll change the ruby lib to send the sequence WARNING (with exception message) + FAIL in case of an exception in the worker.

It would, however, be nice to see this implemented for Perl and Python as well (where Ruby and Python can't even handle a WARNING atm). Would the mailing list be the right place to discuss this? I know Dennis, but not the guy maintaining the Python API on github...

:kim

[*] Actually, this would've led to odd behavior with real "legacy" workers as we've copied the re-submit functionality from the original on_fail handler -- with the two packets in sequence, we would've rescheduled the job twice :/