After an exception causes the debugger to break, continuing execution doesn't do more than unwind one frame of the callstack. That means if the exception is thrown 5 frame stacks below where it's caught, I have to press F8 5 times before execution actually continues. After talking to jimb on IRC, I assume that this is a deliberate decision to, in jimb's words "[..] make it more transparent what frames the exception is getting propagated past, and where it's getting caught." I think that this use case is fully covered by single stepping: when I do that, I'd fully expect each step to unwind a single frame until the exception is caught or the stack exhausted. Continue, OTOH, should do what it sounds like, and continue execution. Again in jimb's words: "once we've paused for a given exception, it seems like a 'continue' should sort of consider that exception to be "dealt with", in a sense, and just let it propagate freely."
It does seem to me that, once we've paused for an exception's propagation, that 'continue' should sort of 'let go' of that exception, rather than reporting it again for each frame it goes through.
Summary: Continuing execution after breaking on an exception should fully unwind the exception and actually continue execution → Continue after pause on exception should fully unwind exception
Bug 1092922 is a duplicate of this one, but since jryans is actively working on that one, I'm closing this one instead.
Status: NEW → RESOLVED
Last Resolved: 3 years ago
Resolution: --- → DUPLICATE
Duplicate of bug: 1092922
You need to log in before you can comment on or make changes to this bug.