By default a cdef (or cpdef) function returning C type suppresses all
exceptions, which is quite surprising to new (and old) users, and
makes debugging difficult. I would propose that we make exception
propagation the default.
The primary motivation to not do this would be for fear of performance
regression, especially for external functions that can't possibly
raise exceptions. While the the catch-all "except *" does seem to have
non-trivial overhead, the "except? value" version looks
indistinguishable from no exception throwing at all (in tight loops,
yay branch prediction, would be interested in what other people see
[1]).
Does this sound like a reasonable proposal?
There are still some open questions, e.g.
- What value should we pick. We want something unlikely, but
recognizable to external code (e.g. 0xBadBad).
- What about non-numeric/pointer/reference types; are there other
sentinel values we could use?
- Should we default to except * in that case? Presumably these are
likely to be more expensive functions anyways (making the relative
cost lower).
- Or could we use, essentially, reinterpret_cast to get a possible
error value? This could work for PODs (including C structs) but
destructors could wreck havoc here.
- Should we add syntax (e.g. "except -") to manually obtain the old
behavior? Or should such be done by surrounding the entire block with
an try/bare except statement? This wouldn't catch argument-parsing
errors. (Another crazy idea, though I'm not sold, is lifting an
initial try and with block that spans the full function body to above
the argument parsing--this could simply the gil syntax as well.)
- Robert
[1] https://gist.github.com/robertwb/0ef6bfaa3c9b2ea11ed64d1c1a0bda43
time_with_nothing 0.0322368144989 3.22368144989e-08
time_with_star 0.097442150116 9.7442150116e-08
time_with_value 0.0940091609955 9.40091609955e-08
time_with_maybe_value 0.10076379776 1.0076379776e-07
time_with_nothing 0.0767869949341 3.8393497467e-08
time_with_star 0.189253091812 9.46265459061e-08
time_with_value 0.176230192184 8.81150960922e-08
time_with_maybe_value 0.179567098618 8.97835493088e-08
time_with_nothing 0.110851049423 3.69503498077e-08
time_with_star 0.268408060074 8.9469353358e-08
time_with_value 0.254024982452 8.46749941508e-08
time_with_maybe_value 0.244876861572 8.16256205241e-08
time_with_nothing 0.138132095337 3.45330238342e-08
time_with_star 0.345059156418 8.62647891045e-08
time_with_value 0.329736948013 8.24342370033e-08
time_with_maybe_value 0.341785907745 8.54464769363e-08
time_with_nothing 0.0680961608887 6.80961608887e-10
time_with_star 0.285541057587 2.85541057587e-09
time_with_value 0.0707340240479 7.07340240479e-10
time_with_maybe_value 0.0762150287628 7.62150287628e-10
time_with_nothing 0.148243904114 7.41219520569e-10
time_with_star 0.560220956802 2.80110478401e-09
time_with_value 0.136430978775 6.82154893875e-10
time_with_maybe_value 0.143126964569 7.15634822845e-10
time_with_nothing 0.211837053299 7.06123510997e-10
time_with_star 0.812189102173 2.70729700724e-09
time_with_value 0.197682142258 6.58940474192e-10
time_with_maybe_value 0.203264951706 6.7754983902e-10
time_with_nothing 0.283420085907 7.08550214767e-10
time_with_star 1.09991884232 2.74979710579e-09
time_with_value 0.281104803085 7.02762007713e-10
time_with_maybe_value 0.288016080856 7.20040202141e-10
DONE
_______________________________________________
cython-devel mailing list
cython-devel@python.org
https://mail.python.org/mailman/listinfo/cython-devel
