murlsn> What's the version of Windows you're using?
That is win95.
murlsn> Try using MinGW tools only, with Cygwin bash renamed to sh.exe and set
murlsn> in a directory that IS in the path, and use just plain SHELL=/bin/sh
murlsn> That works just fine for me on Windows 2000. I can run linker lines
murlsn> that don't even fit in a 80x50 console screen without problems. If I don't
murlsn> do the sh stuff as above, make creates batch files in order to run the
murlsn> commands, and Windows 2000 allows up to 4K of arguments only, which is not
murlsn> enough for me. If you're using something like Windows 95, I'm sure there
murlsn> are much worse restrictions.
Well, it seems I had a djgpp set env variable, plus I had an old mingw make,
which I replaced for the one that comes in Mr. Khan's 2.95.2 package. Oddly,
this make waits for quite some seconds before it starts working. However, I
discovered some flags were silently ignored, as evidenced by some linking and
Suppose I had:
gcc -many flags and defines here- -c foo.c -o foo.o
Compilation did not fail, but some flags were not passed to gcc/ld. I still
don't know whether this was due to the presence of a -I flag without a value,
or to the presence of -s and -pipe, or to all them at a time.
As an aside, I defaulted to the Bloodshed package, since I could not
convince Mr. Khan's one about the existence of crt2.o.