i've been tinkering with PCBoard source to try and see why it takes so
much CPU on WinXP ... looks like it tests for Win95 and only gives up
the timeslice every 8 iterations, to prevent some idle problem with
Win95 ... so i patched that and it stopped hogging the CPU ... but only
when using a COM port, not fossil ... fossil still hogs the CPU
so i thought of using Turbo Profiler to find the reason ... PCBoard is
too big to run on the same box with the profiler ... you have to set up
remote debugging via serial cable ... got that working with a little
test program
then i tried PCBoard ... it's too big to compile with debugging symbols
on all modules ... so i left out dbase and fido, then it compiled, but
the .exe was still 1.5 meg ... got that partly working, using passive
mode which gives a big picture of CPU use ... but to get more detail you
must use active mode ... and that crashes turbo profiler every time ...
don't know if it's because PCBoard is too big, or weird things CDC did
(they patched the Borland startup code) ... so much code, so little time
it's obvious CDC made poor design choices ... they packed too many big
features into one huge executable and relied on Borland's VROOM overlay
manager squeeze it all into 640k ... makes debugging and/or profiling
difficult if not impossible
their bad design painted them into a corner ... with no way out, i can
see why management took the money and ran
maybe a port to another bc version would open more ram.
or 32bit. but that would be a huge job.
i've been tinkering with PCBoard source to try and see why it takes so
much CPU on WinXP ... looks like it tests for Win95 and only gives up
the timeslice every 8 iterations, to prevent some idle problem with
Win95 ... so i patched that and it stopped hogging the CPU ... but only
when using a COM port, not fossil ... fossil still hogs the CPU
so i thought of using Turbo Profiler to find the reason ... PCBoard is
too big to run on the same box with the profiler ... you have to set up
remote debugging via serial cable ... got that working with a little
test program
then i tried PCBoard ... it's too big to compile with debugging symbols
on all modules ... so i left out dbase and fido, then it compiled, but
the .exe was still 1.5 meg ... got that partly working, using passive
mode which gives a big picture of CPU use ... but to get more detail you
must use active mode ... and that crashes turbo profiler every time ...
don't know if it's because PCBoard is too big, or weird things CDC did
(they patched the Borland startup code) ... so much code, so little time
it's obvious CDC made poor design choices ... they packed too many big
features into one huge executable and relied on Borland's VROOM overlay
manager squeeze it all into 640k ... makes debugging and/or profiling
difficult if not impossible
their bad design painted them into a corner ... with no way out, i can
see why management took the money and ran
true. seems most of their mess is to fix borlands errors.
Well at least I have the problem narrowed down to one source file. That >should make it easier to find a work around.
I surmise there is something funny going on inside PCBoard that
Turbo Profiler does not like.
I tried two VMs in Virtualbox 4.3.16, and it worked, I got results
in active mode.
I see rick parrish is making a 32bit port of shotgun
I work on it in my spare time when I don't feel like working
on anything else, which pretty much means not that often
It would be nice to finish it as some point
but how much source code is in there?
most old source code I found don't even work anymore.
and the little that does is not y2k compatible.
one I played with said I was 65556 years old.
i've been tinkering with PCBoard source to try and see why it takes so
much CPU on WinXP ... looks like it tests for Win95 and only gives up
the timeslice every 8 iterations, to prevent some idle problem with
Win95 ... so i patched that and it stopped hogging the CPU ... but only
when using a COM port, not fossil ... fossil still hogs the CPU
so i thought of using Turbo Profiler to find the reason ... PCBoard is
too big to run on the same box with the profiler ... you have to set up
remote debugging via serial cable ... got that working with a little
test program
then i tried PCBoard ... it's too big to compile with debugging symbols
on all modules ... so i left out dbase and fido, then it compiled, but
the .exe was still 1.5 meg ... got that partly working, using passive
mode which gives a big picture of CPU use ... but to get more detail you
must use active mode ... and that crashes turbo profiler every time ...
don't know if it's because PCBoard is too big, or weird things CDC did
(they patched the Borland startup code) ... so much code, so little time
it's obvious CDC made poor design choices ... they packed too many big
features into one huge executable and relied on Borland's VROOM overlay
manager squeeze it all into 640k ... makes debugging and/or profiling
difficult if not impossible
their bad design painted them into a corner ... with no way out, i can
see why management took the money and ran
no one would want to setup a 16 bit system to call a svga bbs
I tried running two VMs in Virtual PC 2004 (and 2007). Crash, crash,
crash, just like on real hardware. Then I tried two VMs in Virtualbox 4.3.16, and it worked, I got results in active mode.
maybe a port to another bc version would open more ram.
I have Borland C++ 4.52, but it needs Windows (3.1 or higher) to run. I
want PCBoard to compile on DOS, so that means getting Borland C++ 3.1 to work.
or 32bit. but that would be a huge job.
Yes, more time than I have.
woah, I see rick parrish is making a 32bit port of shotgun.
I see rick parrish is making a 32bit port of shotgun
At https://www.randm.ca/bbs/gunshot/ he says:
I work on it in my spare time when I don't feel like working on
anything else, which pretty much means not that often
IOW,
It would be nice to finish it as some point
A nice daydream, though one that will never materialize.
in a BBS torrent I have. I don't know who put the torrent together, but
it's large, about 440 meg, and has most any BBS software you could want.
I don't usually keep the torrent online, but if you want it, let me
know.
aa4re abbs acit acropolis adtbbs affinity afterhours aibbs asgard86 astrodome atlas auntie aurora autobbs axis b0badel backdoor baphomet
bbasm bbbs bbs4c bbsx blacklightning boyan brandx captains celerity
i've been tinkering with PCBoard source to try and see why it takes so
much CPU on WinXP ... looks like it tests for Win95 and only gives up
the timeslice every 8 iterations, to prevent some idle problem with
Win95 ... so i patched that and it stopped hogging the CPU ... but only
when using a COM port, not fossil ... fossil still hogs the CPU
so i thought of using Turbo Profiler to find the reason ... PCBoard is
too big to run on the same box with the profiler ... you have to set up remote debugging via serial cable ... got that working with a little
test program
then i tried PCBoard ... it's too big to compile with debugging symbols
on all modules ... so i left out dbase and fido, then it compiled, but
the .exe was still 1.5 meg ... got that partly working, using passive
mode which gives a big picture of CPU use ... but to get more detail you
must use active mode ... and that crashes turbo profiler every time ...
don't know if it's because PCBoard is too big, or weird things CDC did
(they patched the Borland startup code) ... so much code, so little time
it's obvious CDC made poor design choices ... they packed too many big features into one huge executable and relied on Borland's VROOM overlay manager squeeze it all into 640k ... makes debugging and/or profiling difficult if not impossible
their bad design painted them into a corner ... with no way out, i can
see why management took the money and ran
someone did start a pcboard clone once. some French hacker group.
but they never finished it. I think it was just a core engine to run the ppes. and everything else the bbs had had to be a ppe format.
heck, even wwiv is 32bit now.
and some of it even works too
i've been tinkering with PCBoard source to try and see why it takes so
much CPU on WinXP ... looks like it tests for Win95 and only gives up
the timeslice every 8 iterations, to prevent some idle problem with
Win95 ... so i patched that and it stopped hogging the CPU ... but only
when using a COM port, not fossil ... fossil still hogs the CPU
so i thought of using Turbo Profiler to find the reason ... PCBoard is
too big to run on the same box with the profiler ... you have to set up remote debugging via serial cable ... got that working with a little
test program
then i tried PCBoard ... it's too big to compile with debugging symbols
on all modules ... so i left out dbase and fido, then it compiled, but
the .exe was still 1.5 meg ... got that partly working, using passive
mode which gives a big picture of CPU use ... but to get more detail you
must use active mode ... and that crashes turbo profiler every time ...
don't know if it's because PCBoard is too big, or weird things CDC did
(they patched the Borland startup code) ... so much code, so little time
it's obvious CDC made poor design choices ... they packed too many big features into one huge executable and relied on Borland's VROOM overlay manager squeeze it all into 640k ... makes debugging and/or profiling difficult if not impossible
their bad design painted them into a corner ... with no way out, i can
see why management took the money and ran
| Sysop: | DaiTengu |
|---|---|
| Location: | Appleton, WI |
| Users: | 910 |
| Nodes: | 10 (1 / 9) |
| Uptime: | 117:10:14 |
| Calls: | 12,128 |
| Calls today: | 2 |
| Files: | 186,512 |
| Messages: | 2,230,122 |