On 7/7/2022 9:59 AM, C. Masloch wrote:
On at 2022-07-07 09:39 -0700, Ralf Quint wrote:
Similar with NASM, where for some weird reasons, they made the
assembler case-sensitive, which I would consider utter nonsense (also
among my griefs with C(++)). And it really bites you if you are
trying to link assembler modules with other programming languages. It
also (deliberately) doesn't support some assembler instructions (as
per Intel specs) just because it doesn't fit into their parser (LODS,
MOVS, ...),
Fair enough on all the other points, but the support for string
instructions is actually there. It just isn't done with explicit
operands to the instruction to indicate the operand size and segment
override (whereas I believe the symbol offset address you can specify
to MASM is ignored). Instead, NASM supports eg lodsb / lodsw / lodsd,
as well as segment override and a16 or a32 as prefixes. So for example
you could have "a32 es lodsd". This syntax used to be documented in
their instruction set reference, as forked by me. [1]
As I mentioned, this is one of those changes that pretty much prevent
you from assembling A LOT of older x86 source code from the early days
of DOS (and x86).
Microsoft's MASM maintained full compatibility with Intel's own
assembler, ASM86. And after all, Intel is the one entity that
designed/defined x86. At least NASM did not follow the stupid idea of
AT&T (and gas, to me a typical Stallman idiosyncrasy, but apparently
AT&T got on that trip before he got involved) to use a complete
different syntax...
I would much rather use one of those Open Source assemblers (btw, which
A86 is not!) like WASM/JWASM that stick to the Intel/Microsoft syntax
and compatibility than to start converting each and every piece of
assembly language source for DOS...
Ralf
_______________________________________________
Freedos-user mailing list
Freedos-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-user
