Argument Clinic: add unsigned integers converters

[larry]

This patch is just about perfect.  I have only minor complaints.  If you accept
both my suggested changes, you have my blessing to check it in.  Thanks, this
helps!

http://bugs.python.org/review/20260/diff/10568/Doc/howto/clinic.rst
File Doc/howto/clinic.rst (right):

http://bugs.python.org/review/20260/diff/10568/Doc/howto/clinic.rst#newcode732
Doc/howto/clinic.rst:732: \           ``unsigned_short``
Please remove these from this table.  This table shows the reader how to change
legacy converters into real converters; these new converters have no legacy
equivalent.

Instead, please add a whole new paragraph, just above line 752 here ("Argument
Clinic will show you [...]").  The paragraph should read:

    One advantage of real converters is that
    they're more flexible than legacy converters.  For
    example, the ``unsigned_int`` converter (and all the
    ``unsigned_`` converters) can be specified without
    ``bitwise=True``.  Their default behavior performs range
    checking on the value, and they won't accept negative
    numbers.  You just can't do that with a legacy converter!

http://bugs.python.org/review/20260/diff/10568/Include/longobject.h
File Include/longobject.h (right):

http://bugs.python.org/review/20260/diff/10568/Include/longobject.h#newcode69
Include/longobject.h:69: PyAPI_FUNC(int) _Py_UnsignedShort_Converter(PyObject *,
void *);
This should be named "_PyLong_UnsignedShortConverter", and the same approach
applies to the other four new functions.

Sorry for the bikeshedding but I do think this is more in keeping with Python's
style.  Python C functions with this camel-case style name don't get additional
underscores as word breaks (e.g. "PyUnicode_FSConverter").

[larry]

Quick feedback on one aspect of the patch, not a full review.

http://bugs.python.org/review/20260/diff/10585/Objects/longobject.c
File Objects/longobject.c (right):

http://bugs.python.org/review/20260/diff/10585/Objects/longobject.c#newcode1426
Objects/longobject.c:1426: if (PyLong_Check(obj) && _PyLong_Sign(obj) < 0) {
Maybe instead of doing these checks by hand, it should just call
PyLong_AsUnsignedLong()?  I'm sure that already throws an exception if the
object isn't a Long, or if it's negative.

[storchaka]

https://bugs.python.org/review/20260/diff/10585/Objects/longobject.c
File Objects/longobject.c (right):

https://bugs.python.org/review/20260/diff/10585/Objects/longobject.c#newcode1426
Objects/longobject.c:1426: if (PyLong_Check(obj) && _PyLong_Sign(obj) < 0) {
On 2014/01/24 16:38:56, larry wrote:
> Maybe instead of doing these checks by hand, it should just call
> PyLong_AsUnsignedLong()?  I'm sure that already throws an exception if the
> object isn't a Long, or if it's negative.

It throws an OverflowError for negative Long, but we want a ValueError.