[pypy-dev] lltype Questions

Amaury Forgeot d'Arc amauryfa at gmail.com
Fri Jun 18 10:19:07 CEST 2010


2010/6/18 Dan Roberts <ademan555 at gmail.com>:
> Hey Everybody,
>        It seems no one's in #pypy at the moment, so I figured I'd post to the
> mailing list for non-realtime questioning.  I wrote my implementation of
> PyUnicode_DecodeUTF16() which is correct as best I can tell (and it
> passes its test, but I wrote the test too :-) ).
>        The code is here: http://paste.pocoo.org/show/226753/ and its test is
> here: http://paste.pocoo.org/show/226752/ . I have a few questions, like
> I said, the test passes, but I'm unsure of some of the code I've
> written.  There are XXX and FIXME and other comments throughout the
> code, those represent places where I'm unsure.
>   In the test code, line 6 I'm not sure if that malloc() is correct,
> and even if it is, would it have been better (and also valid) to
> lltype.malloc(lltype.Signed, ...) ?  Then from line 7 to 12 I assign
> "native" integers to the malloc-ed memory, is this done correctly? ex.
> pendian[0] = -1 or should it be pendian[0] = rffi.cast(lltype.Signed,
> -1) ?
>   In the implementation, I'm fairly certain I've marked everything with
> FIXME or XXX, so there's not much more to say about that...  I'd really
> appreciate anyone who's willing to take a look.

All this looks quite good, except for the second cast - it should be
    pbyteorder[0] = rffi.cast(rffi.INT, byteorder)

And I suggest to add another test: automatic detection of byte order
when a Byte Order Mark is present.

Amaury Forgeot d'Arc

More information about the Pypy-dev mailing list