For the sake of future googlers here are explicit answers to the questions I raised and then some more.
On Sun, Aug 28, 2011 at 2:43 AM, Jean-Claude Beaudoin <email@example.com>
Should I understand that gcc in MinGW (as of now) is strictly an 8-bit ANSI character application
as Microsoft would call it (as opposed to a UNICODE character application)?
And that I should give up on the idea of gcc handling Unicode filenames on the
command line, or in #include directives for that matter?
After a bit of experimentation and some reading of the gcc source code the answers are:
Yes and yes.
Also, gcc in MinGW is unable to handle being installed in a directory whose any part of
the path leading to it contains a non-ASCII Unicode character. In such a case, gcc is simply
unable to find its subcomponents (such as cc1, as, ld...) and terminates on startup.
The usual warning about blank space characters in the installation path of MinGW is
clearly too weak and should be upgraded to warn about any non-ASCII Unicode characters.
(More precisely the problem is with any Unicode character that is not present in the
ANSI code page). Most probably here the only safe bet is to restrict oneself to strict 7-bit ASCII.