Hi Pawel,
This is subtle...
"use32" is an alias for the *directive* "bits 32" - the actual raw form is "[bits 32]" - "bits 32" without the brackets is also a macro. This is what you'd use - at the top of the file, typically - to tell Nasm to emit 32-bit code.
What you've got there is not a "directive", but an "attribute" or "property" of the "section"/"segment" (also aliases) directive. In this case, "bits 32" isn't going to do anything. The "alias" can be seen in "standard.mac" - it defines "use32" as "[bits 32]", not the other way around...
"use32" was made an "alias" for convenience, but in this case, it only causes confusion...
If you look in the manual under "output formats", "extensions to the segment directive", instead of in the "directives" section, "bits", etc., you'll see the difference.
By rights, Nasm should complain about "section foo bits 32"... apparently it doesn't... ("progbits"/"nobits" *are* valid...)
FWIW, you may also want to add "class=code" to your ".code" section. I don't know what it does, but Ollydbg, at least, is happier when you do it.
"-f win32" is the more "modern" way to make a Windows file (yeah, I know it doesn't know "import"...) "-f obj" will work, but since it defaults to 16-bit, there are some extra hoops to jump through.
In this case, "well, don't change it" is the correct answer.
Best,
Frank
