[Python-checkins] r42382 - in python/trunk: Include/abstract.h Include/bufferobject.h Include/cStringIO.h Include/ceval.h Include/dictobject.h Include/intobject.h Include/listobject.h Include/longintrepr.h Include/longobject.h Include/marshal.h I

[Neal Norwitz]

On 2/15/06, martin.v.loewis <python-checkins at python.org> wrote:
> Author: martin.v.loewis
> Date: Wed Feb 15 18:27:45 2006
> New Revision: 42382
>
> Log:
> Merge ssize_t branch.

Very nice.  I actually reviewed this patch.  Here are my cryptic
comments.  Let me know if they aren't understandable.  Hopefully I can
decipher them tomorrow.  I mean today.

n
--

Need to update doc wrt Py_BuildValue (format codes):  n and z.  I can
take a stab at this if you want.  Let me know.

modsupport.h: Don't we want to bump the version?

 #define PYTHON_API_VERSION 1012
 #define PYTHON_API_STRING "1012"

cPickle.c, but same for cStringIO.c:

           Why do {read,write}_{file,cStringIO,none,other}() and
           load_counted_long() return an int rather than ssize_t?

rgbimgmodule.c: Why casting (int)fwrite()?

zipimport.c: Why casting result of strlen(name)?  I realize the name
             length is not going to be larger than an int, but it
             seems easier to not worry about this and return ssize_t.

fileobject.c:  Why? assert(nfree < INT_MAX); /* Should be atmost MAXBUFSIZE */
unicodeobject.c: Why can't unicode->length be ssize_t?

Parser/parsetok.c: Why can't tok->buf be longer than INT_MAX?

exceptions.c: *value = (size_t)PyLong_AsLongLong(attr); // should be Py_ssize_t?

unicodeobject.c: PyUnicodeDecodeError_Create() why asserts < INT_MAX and (int)?
                 UnicodeTranslateError__str__() also

getargs.c:  Should PyArg_*_SizeT use SsizeT to be consistent?