Dag Sverre Seljebotn wrote: > when modifying the testcase (r_doctests.pyx), I encountered a strange > thing: If I made the doctest a unicode string, then Python 2.5 still > wouldn't print it out with a u prefix in the doctest. Is this supposed to > work differently?
No, that's the way docstrings work in Py2 for classes and modules, at least at the C level (maybe also functions, don't remember). They are passed as plain encoded byte strings (char*). Only Py3 decodes them from UTF-8, Py2 takes them as they are. This is definitely sub-optimal (thus the fix in Py3), and IIRC the current implementation in Cython even encodes docstrings to UTF-8 to make them work with Py3, even if they were originally byte strings encoded as, say, latin-1. Might be worth another work-around for Py2 here... Stefan _______________________________________________ Cython-dev mailing list [email protected] http://codespeak.net/mailman/listinfo/cython-dev
