Starting at the last, I have dispatched an email to the developers' list, asking. Last I knew, OpenWatcom running on Linux but targetted to PE was used. That may no longer be true.
You see what it says in the "INSTALL" file, don't ya? Use MinGW. Up to you if you want to do that or not. Apparently the MS tools are known to be a hassle.
The "Netwide" part of the fantasy is supposed to indicate that one can compile and run Nasm "anywhere". This is no longer true, if it ever was. The last MS compiler I ran was 6.00a, as I recall. It would not compile Nasm. I installed DJGPP. At one point, I was producing PEs with a Borland "bcc" compiler. I had David Lindauer's CC386 compiling Nasm, too. CC386 would optionally use Tasm or Nasm as a back end. I was amused to watch Nasm assembling "nasm.asm", etc. OK I'm easily amused. This was all a long time ago.
When Daniel Bernstein's student Jonathan Rockway discovered an exploitable overflow in Nasm, we resorted to C99 support to fix it. These days, you need a fairly modern compiler to compile Nasm without hassles. A system that supports "autoconf" and "configure" helps. I can't get an .rpm to work, so I always build Nasm with "./configure" and "make install_everything" and it's pretty painless. I don't know if MinGW would do as well for you, if you choose to go that route.
To get back to what you've got... errors on the two "%define" lines. How strange! I would think that's coming from the preprocessor(?) - nowhere near the output drivers you're trying to omit.
I suspect the complaints you're getting from inttypes.h are a "C99 issue". I can't think why that should cause trouble with "%define", either!
See what we hear from the developers' list...
Best,
Frank
