🚀 go-pugleaf

In <1601@cogsci.ucsd.EDU> hartung@crl.ucsd.edu (Jeff Hartung) writes:

> In article <1992Sep10.052148.10777@monu6.cc.monash.edu.au> parry@yoyo.cc.monash.edu.au (Tom J Parry) writes:
> >It strikes me as starnge that given the current size of the Borland C++ 3.1
> >compiler package, that there is no source control system.
> >
> >Currently, the full package has every conceivable feature you could want
> >except source control and OS/2 support [oops, wrong winge]. The only

> And 32-bit code generation, and a DOS extender (32-bit would be nice, 16-bit
> would be OK, the Phar Lap "Lite" they included with the upgrade was a table
> scrap, and a shameless example of commercial crippleware), and probably a few
> other things I can't think of right now.

I'm really surprised that Borland can't generate 32-bit code now.  Or can it?
I know that the 386/486 have one-byte opcode prefixes that force the next
opcode, if it would normally be an 8-bit instruction, then to do the right
thing but with 32 bits.  (In "native" 32-bit mode, the same opcode toggles
this to mean the opposite.)  So, you can assemble code *right now* that will
run on 386/486 in normal, icky real mode that can use the full 32-bit
registers.  (I'm not sure how this works with the processor's pointers,
though.)

So how much effort would it take (and how much CPU time would it save!!) to
convert all thoses uses of long (and int!!!) to 32 bit instructions?

Or is this being done already and am I years behind??
-- 
lrwx------   1 aldavi01 emacsstu       9 Jun  6 12:43 .signature -> /dev/null