Re: [python-win32] pyconfig.h: conflicting definitions for ssize_t

Mark Hammond wrote:
> On 30/03/2011 11:01 AM, Tim Roberts wrote:>> Wolfgang Rohdewald wrote:>> I agree with your conclusion.  However, this isn't a Python-Win32>> issue.  You need to file a bug report against Python itself.> it would be good to know exactly what vs2008 says about this - if it > uses 'int' we will need to take more care - but yeah - please report > this at bugs.python.org, add mhammond to the "nosy" list and I'll help > ensure it gets fixed...

Actually, on closer examination, it may be a bit difficult to sell
this.  The Microsoft compilers do not define this symbol at all.  The
SDK defines SSIZE_T (as a long).  Nothing defines ssize_t.  So, the
conflict only occurs because the KDE include files do this:
    typedef SSIZE_T ssize_t;

We (and KDE) are more or less assuming that Microsoft intended their
SSIZE_T to be the same as the relatively common ssize_t.  Python defined
their symbol independently of the SDK, which is also reasonable.

In the end, I'm not sure you can point to either KDE or Python as being
"at fault" here.  The warning can be fixed by changing one or the other,
but I don't know how one would establish which one was "right".

-- 
Tim Roberts, t...@probo.com
Providenza & Boekelheide, Inc.

_______________________________________________
python-win32 mailing list
pyth...@python.org
http://mail.python.org/mailman/listinfo/python-win32