Most of the exceptions raised in our code are transient (I/O) but in cases where we identify a condition where a retry does not make sense, we would like to raise a special exception and immediately abort. This used to be possible with the special

Failure

exception, but it got "fixed" recently to abide by the retry policy.

Right now, the only way to achieve this is to not set a retry policy at all, catch all exceptions in the

op

and raise a

RetryRequested

except in the few specific cases that I mentioned. That's a bit clunky.

It would be cool if the reverse were possible: retry on everything except if one special exception is raised.