Answering msg from Paul Edwards to rowan crowe,
on Saturday June 24 1995 at 23:09

 PE> Hi Rowan.  As I believe I mentioned to you, I believe I solved
 PE> the problem with the assembler calls from Watcom by declaring
 PE> them all as "cdecl", and also saving all the registers, instead
 PE> of just the ones that Borland required to be saved.

    I still think that the default of passing by registers is a no-no, but
I guess once you are familiar with Watcom you will know its quirks.

 PE> The bit of assembler that I was unable to convert from Borland
 PE> to Watcom ASM (despite the fact that Watcom claims to be MASM
 PE> compatible, and the code supposedly works with MASM) is
 PE> spawn.asm.  If MSQ310.ZIP makes it to you, do you think you
 PE> could take a look and see if there is some unusual code in
 PE> there that would stop it working on a "generic" MSDOS
 PE> assembler?  It is only that that is preventing me from using
 PE> Watcom to compile the DOS version.

    If it's the same spawn.asm that came with the other releases, then I
already have it:

;       --- Version 3.0 91-05-27 17:56 ---
;
;       SPAWN.ASM - Main function for memory swapping spawn call.
;
;       Public Domain Software written by
;               Thomas Wagner
;               Ferrari electronic GmbH
;               Beusselstrasse 27
;               D-1000 Berlin 21
;               Germany


    It's bloody 45k in size! But I'll have a look. Are you assembling it correctly?

; Assemble with
;
; tasm  /DPASCAL spawn,spawnp           - Turbo Pascal (Tasm only), near
; tasm  /DPASCAL /DFARCALL spawn,spawnp - Turbo Pascal (Tasm only), far
; ?asm  spawn;                          - C, default model (small)
; ?asm  /DMODL=large spawn              - C, large model

    I presume you'd use /DMODL=large spawn

    What sort of errors is the assembler giving you, are they
"global" or does it spit out line numbers. I can't help much
unless I know where to start looking....

    It's possible that Watcom doesn't like jamming directives, in that case
it would be fairly easy to modify the code to remove all of the conditional
assembly and just create the C large model version.

    I don't have MASM so I can't test its "MASMablity" ... which
is a pity as I'll need it soon to play with the compiler I'm working on.
Anyone got a copy of MASM they'd like to give me? ;-)

---