Removing Windows 95-related code in posixmodule.c

Now that Windows 9x is no longer supported (per PEP 11), would it be appropriate to replace the ANSI ("A") versions of Windows OS calls in posixmodule.c with their Unicode ("W") equivalents and remove the unicode_file_names() function that determines which version to call? There are a number of places in that module where we try to decide whether to call ANSI or UNICODE versions (because only the ANSI versions were fully supported on Windows 9x). This is still fine on Win32, where the ANSI versions are still supported for backwards compatibility, but it causes issues with Windows CE, which doesn't have the ANSI versions at all. I've been maintaining some fairly ugly patches to cope with this in my company's Python 2.5.x for CE build, and I'd love to drop them as we move to a post-2.6 version. Thanks! -- Tim Lesher <tlesher@gmail.com>

Tim Lesher wrote:
Now that Windows 9x is no longer supported (per PEP 11), would it be appropriate to replace the ANSI ("A") versions of Windows OS calls in posixmodule.c with their Unicode ("W") equivalents and remove the unicode_file_names() function that determines which version to call?
Not directly. The A versions are still used if you pass byte strings as file names. It would be possible to convert byte strings to Unicode strings (using CP_ACP), but I would only want to consider that if the code complexity does not increase under that approach (which I doubt). FWIW, the unicode_file_names function *is* removed in Python 2.7. Regards, Martin
participants (2)
-
"Martin v. Löwis"
-
Tim Lesher