Eli Zaretskii wrote:
>>> What exactly do the makefiles/scripts expect that make
>>> c:\mingw\bin\../foo undesirable?  Commands and scripts that work on
>>> Windows should generally grok both forward- and back-slashes alike, so
>>> having mixed forward- and back-slashes in a file name is quite normal.
Jared Maddox wrote:
>From reading the post, it looks like the scripts expect *nix style
>paths instead of Windows style paths. Is that what you meant, LM?

Correct the scripts appear to be expecting Linux/Unix style forward slash
format only.  They can't handle the backslash or drive letters.  So, the mixed
slashes format is normal with MinGW?  There's nothing wrong with this?

Earnie wrote:
>Yes, but what scripts?  In general dependency tracking happens in many
>projects without issue so a better understanding of what is happening
>needs to happen before we try to guess what is wrong.

One project is a Linux program I'm trying to port, so I guess I shouldn't
be surprised that it chokes on the Windows syntax.  The first time I ran
into this issue was with SDL 2.0 (which is a cross-platform library).  It
was having trouble with the OpenGL header files being in a directory under
c:\mingw...  I asked on the SDL mailing list and they said they were having
no such trouble on their machines and that my installation of MinGW was
probably broken.  I'd just recently installed the latest version of MinGW
when I hit the issue too and several other projects I'd built on the machine
gave me no problems with compiling and linking.  So, I thought that was
unlikely.  It did lead me to wonder whether there was something
that could be set up with MinGW that could cause gcc to list dependencies using
a more Linux like syntax or whether the problem was with their scripts after
all.  Has anyone been able to build SDL 2.0 on Windows with MinGW?  I was
able to get it to build if I wiped out the .d files, but otherwise it failed.