On Tue, Apr 9, 2013 at 1:44 PM, Nils Bruin <nbr...@sfu.ca> wrote: > On Apr 9, 1:18 pm, Tom Boothby <tomas.boot...@gmail.com> wrote: >> This shouldn't really come as a surprise. From the Cython documentation, >> >> "This is about 20 times slower, but still about 10 times faster than >> the original Python-only integration code. This shows how large the >> speed-ups can easily be when whole loops are moved from Python code >> into a Cython module." > > Since it is entirely unclear from that tutorial what the factor 20 (or > the factor 10) refers to, I would not have understood that code to > mean "even the fastest path of a cpdef function is slower than a > cdef". There is a lot of documentation around that states "a cpdef > function is essentially a cdef function that is also available and > override-able as a python function" without stating that the > overriding can also take effect at cdef level and (naturally) causes > overhead.
To let x.foo() call different functions depending on whether your in Cython or Python, or even worse whether x is fully typed, would be even more surprising. Eventually, it would be interesting to explicitly disallow overrides for performance reasons, but that's hard to enforce. I bet we could leverage tp_version_tag though, I'm going to give that a try. > I think this definitely deserves to be clarified in the > cython documentation. Yes, please, send us a pull request :). - Robert -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To post to this group, send email to sage-devel@googlegroups.com. Visit this group at http://groups.google.com/group/sage-devel?hl=en. For more options, visit https://groups.google.com/groups/opt_out.
