Subj : Re-release
To   : Allen Prunty
From : mark lewis
Date : Tue May 10 2016 12:40 am


09 May 16 00:06, you wrote to Sean Dennis:

 AP> You can run 32 bit on a 64 bit system... but older 16 bit doors will
 AP> not run on a 64 bit system but sometimes will run on a 32 but.  If
 AP> they werew written in Turbo Pascal there's a patch that may or may not
 AP> slow them down enough to work properly.

are you thinking of the Runtime Error 200 patches? if so, they don't "slow them 
down enough"... they adjust the way the timing is done for the DELAY 
function... i have my own patch for the CRT library but it is more of a 
replacement routine since i also have the sources to the CRT lib... i simply 
replaced the DELAY routine with one of my own and recompiled it... existing 
binary files need binary patching in the same way though... that is what all 
the various patching programs take care of...

the original problem is that they decided that since they knew how long a NOP 
instruction took, they would execute X number of them and it should equal the 
time passed on the clock... as machines got faster, they executed all the NOPs 
in the same time increment and the result between the starting time reading and 
the ending one was zero... when that was fed, without checking if the result 
was zero, to the division routine to figure out the proper DELAY synchronizer 
value, the result was the Divide by Zero Runtime Error 200... even if they had 
checked for the result being zero before the division there wasn't much they 
could do other than to run the NOP loops again... different patches handle this 
different ways... ones that i'm familiar with simply replace the DELAY routine 
and skip the calibration routine...

)\/(ark

Always Mount a Scratch Monkey

.... We don't do no dups!
---
 * Origin:  (1:3634/12.73)

.