[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: MKS sh problem (follow-up)

From: Werner LEMBERG
Subject: Re: MKS sh problem (follow-up)
Date: Mon, 24 Nov 2003 14:21:22 +0100 (CET)

> > Configure correctly discovers '.exe' as the executable suffix, but
> > doesn't use it when searching for C and C++ compilers, so it fails
> > to find cl in either case.  Changing every instance of 'cl' to
> > 'cl.exe' effects a quick fix, but making use of the executable
> > suffix would be more elegant as well as more portable.  Setting
> > the variables 'CC' and 'CXX' is another interim fix.
> Another fix is to set ac_executable_extensions to ".exe" (or ".exe
> .cmd .bat" if you like) in your config.site.  That variable was
> added for this exact purpose - not all Unixy shells on DOS/Windows
> return true for 'test -x foo' if only foo.exe exists.

Uh, oh, this is *completely* undocumented in autoconf -- apparently an
omission.  So we have a new bug report :-)

> I suppose autoconf could add $EXEEXT to this variable automatically,
> but that really gives a chicken-and-egg problem, as in order to
> determine the extension the compiler adds, it needs to compile &
> link a test program, and for that it needs to find the compiler
> first, for which it needs the extension.

Not necessarily.  If, say, `gcc' doesn't work as a compiler, the
configure script might try an extension of a small set of default
extensions, most notably `.exe'.  Theoretically it could be possible
that the extension of the compiler is not the extension which
compiling & linking produces.  Cross-compiling comes to my mind...


reply via email to

[Prev in Thread] Current Thread [Next in Thread]