🚀 go-pugleaf

Hi all,

From a private email:

...

>I'm also still working on a 386+ only TURBO.TPL - I've increased the
>granularity of the DOS unit by splitting most files into two
"get"/"set"
>parts, and in one case, TIME.ASM, into four parts, TIMEGDAT.ASM,
>TIMESDAT.ASM, TIMEGTIM.ASM & TIMESTIM.ASM! I've also changed the code
>for Intr/MsDos to the BP7 version, issuing a real interrupt, so that
it
>will work under windoze. I've also done some work on CRT, most
notably
>I've removed the Delay initialisation to Delay (it's done on the
first
>call (or a Delay(0)) so that Delay can be smartlinked out completely.
>I've extracted Pedt S' code using TWU1, and I like his use of a
detected
>enhanced keyboard, but rather than his way of detecting it, I think I
>will use one of the newer interrupts for it, 

Any comments on this remark?

>given that the code will only run on at least a 386!
>
>I've also added a small amount of 386 code to OVERLAY and made some
>minimal optimizations to it, but as yet I haven't split it up - which
>should give some smartlink-savings, as I guess that some procedures
in
>it are used very infrequently!
>
>As for the SYSTEM unit itself, sofar I've modified and/or added 386
>instructions to 41 of the files and I've also decided to add the last
of
>BP7's files, appropriately named "LAST.ASM", which clears out the
unused
>portion of the datasegment. It's also very likely that I will (try
to)
>remove the re-entrancy problems in the original RTL FPU code and
>heapmanager!

...

Q1) How much interest would there be in such a unit, or a similar 
one for BP7? (TPL60N20 will be based on the TPL60N19 code by Norbert 
Juffa and a tiny bit of SYS60A code by Eagle Performance Software 
(aka Jim LeMay))

Q2) My disassemblies of TPC.EXE (TP 6.00 & TP 6.01) are complete and 
assemble, after some massaging to byte-4-byte equivalent versions 
(except for the relocation table). I'm ready to create a VER65
version,
where BASM will be able to assemble a lot of 386(+) instructions. I 
could do the same for BP7's compilers, but except for the command line
compiler this will be a lot harder. Would there be any interest in
such
a hacked compiler?

BTW, I will create two versions of some (or all?) of the units, a
normal one and a test/debug one that has NOPs before most and after
some RTL routines, allowing you to patch in FAR JUMPS for user exits,
like Andy Cooper's HeapChk routine (MEMORY.SWG, file 85) and my own
MyWrite unit that intercepts the Longint write routine to emulate
procedures/functions with a variable number of parameters.

Regards,

Robert
-- 
Robert AH Prins
prino@bigfoot.com

PS: Does anyone know how I can get TLINK to remove the uninitialized
datasegment from a linked file, it's included as lots of zeroes in my
reassembled versions of TPC.EXE and I don't like them one bit!