Hi David,
Thanks for all those follow-up updates, and fixing the glaring error I
left in the initial man page. *cough*
----- Original Message -----
> [...]
> I'm all for qa tests, but writing a qa test for this small bug seems
> like overkill, especially since there really isn't a good way to test
> it. The effort required to develop a test for this small bug vastly
> outweighs the bugs importance IMHO.
I understand where Marks coming from though - back in the day, when we
were working on developing PCP originally (on IRIX), it was mandatory
that any bug fix would be accompanied by a test.
If it is truly difficult to hit this branch (really? can't tell, not
enough info) and since its newly developed code, then it may indeed
be OK to skip a test case in this case - IMO (not my call though).
Personally, I would prefer to see an automated test case, since bugs
have an annoying ability to un-squash themselves.
Re "effort required to develop a test..." - there should be little/no
effort to writing automated test cases (the fix was tested somehow,
right?) - and with practice it just gets easier. If there are ways
we can make writing tests easier, we should do that. The ./qa/new
script is pretty handy if thats not been discovered yet - but please
share any pain points you've come across.
cheers.
--
Nathan
|