Python + Borland - msvcrt = grr.

[jj]

On Wed, 25 Apr 2001 06:50:06 GMT, "Stephen Hansen" <news at myNOSPAM.org>
wrote:

>    Scanning over your patch, what is obvious from it being a plain-diff, it
>seems we were indeed fixing the same things. Did you run a full test-suite
>afterwards? I just did, and an 'import types' under a Borland-compiled
>python21.dll + python.exe gives up a Memory Error.
>
I got the same errors.
>    I'm also hitting some problems with porting _winreg -- Borland doesn't
>have a '_mbstrlen', apparently, dispite hte fact that its documented. :) And
>i'm not sure if it'd be safe to replace it with 'strlen', I'm sure they did
>a multibyte version for a reason.
Ops, I excluded it from build. Mark H.

>    Any ideas? Did you fix any of the above? And, this -D_DEBUG problem you
>mentioned, I hanv't run into that I have found, yet.
>
when compiling marshal.c with _DEBUG and _ASSERTE switch, an error is
encountered.

Error E2188 c:\jj\cvspython\python\dist\src\Python\marshal.c 63:
Expression syntax in function w_more

because 

 *p->ptr++ = Py_SAFE_DOWNCAST(c, int, char);
is expanded into:
*p->ptr++ = ({ if (!((int)(char)(c) == (c))) { printf("Assertion
failed\n"); bort(); } }, (char)(c));

I know that it could be solved without _ASSERTE, but does it make
sence, because there must be a reason behind this stuff in code.

JJ