Need a little help from a Pascal guru (Chess05 Frey/Atkin)

sje · Post by **sje** » Mon Jan 09, 2012 5:03 pm

The program source has non-local goto statements, and that's a show-stopper in Free Pascal. I don't use GNU Pascal, so it may behave differently.

I strongly suspect a problem with the old style I/O calls, all of which were designed for a non-interactive environment. Perhaps the program is stalled waiting for input or it got an end-of-file and doesn't know what to do.

Sven · Post by **Sven** » Mon Jan 09, 2012 5:21 pm

Jim Ablett wrote:I've actually compiled the program as got it running on Windows using the Gnu Pascal Compiler.
I can make one move and get one move back from the engine then it just seem to get stuck in an endless loop.
I cannot input any more moves and have to do a Ctrl+Break to get out of it.
This is really frustrating - so near yet so far.

Here's the src & semi-working windows exe > http://dl.dropbox.com/u/5047625/chess05%20iso%20src.zip

Several lines are tagged with a comment "[sam]" which I assume not to be your changes, right? I see that most of the changes are related to a variable named "jumpin", which is also involved in the main loop and in the command line parser, so there might be some correlation to the problem you have described. I did not dig deeper into it, it is my first suspicion.

I tried also to run the executable on WinXP 32bit (old Intel P4, I must admit) but it crashes immediately:

Code: Select all

      7 &#91;main&#93; ? 2720 handle_exceptions&#58; Exception&#58; STATUS_ACCESS_VIOLATION
Exception&#58; STATUS_ACCESS_VIOLATION at eip=7C921529
eax=00400000 ebx=00000000 ecx=610E889C edx=0022FFE0 esi=01CCCEE9 edi=00400000
ebp=0022EECC esp=0022EEA4 program=, pid 0, thread main
cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
Stack trace&#58;
Frame     Function  Args
0022EECC  7C921529  &#40;00400000, 00000000, 0022EEE8, 7C809F9F&#41;
0022EEDC  7C92162C  &#40;00400000, 0022EF18, 6109A582, 00400000&#41;
0022EEE8  7C809F9F  &#40;00400000, 00000001, 0022EFF0, 6104F4A6&#41;
0022EF18  6109A582  &#40;00400000, 0022FFE0, 7C91E920, 7C927C58&#41;
0022EFD8  61004A78  &#40;0022EFF0, 0022F03C, 0022EFF0, 002310A4&#41;
0022FF88  6100594F  &#40;00000000, 00000000, 00000000, 00000000&#41;
End of stack trace

which is clearly a different problem. Or have you provided a 64 bit version only?

Sven

Jim Ablett · Post by **Jim Ablett** » Mon Jan 09, 2012 6:04 pm

Sven Schüle wrote:
Jim Ablett wrote:I've actually compiled the program as got it running on Windows using the Gnu Pascal Compiler.
I can make one move and get one move back from the engine then it just seem to get stuck in an endless loop.
I cannot input any more moves and have to do a Ctrl+Break to get out of it.
This is really frustrating - so near yet so far.

Here's the src & semi-working windows exe > http://dl.dropbox.com/u/5047625/chess05%20iso%20src.zip
Several lines are tagged with a comment "[sam]" which I assume not to be your changes, right? I see that most of the changes are related to a variable named "jumpin", which is also involved in the main loop and in the command line parser, so there might be some correlation to the problem you have described. I did not dig deeper into it, it is my first suspicion.

I tried also to run the executable on WinXP 32bit (old Intel P4, I must admit) but it crashes immediately:
Code: Select all
      7 &#91;main&#93; ? 2720 handle_exceptions&#58; Exception&#58; STATUS_ACCESS_VIOLATION
Exception&#58; STATUS_ACCESS_VIOLATION at eip=7C921529
eax=00400000 ebx=00000000 ecx=610E889C edx=0022FFE0 esi=01CCCEE9 edi=00400000
ebp=0022EECC esp=0022EEA4 program=, pid 0, thread main
cs=001B ds=0023 es=0023 fs=003B gs=0000 ss=0023
Stack trace&#58;
Frame     Function  Args
0022EECC  7C921529  &#40;00400000, 00000000, 0022EEE8, 7C809F9F&#41;
0022EEDC  7C92162C  &#40;00400000, 0022EF18, 6109A582, 00400000&#41;
0022EEE8  7C809F9F  &#40;00400000, 00000001, 0022EFF0, 6104F4A6&#41;
0022EF18  6109A582  &#40;00400000, 0022FFE0, 7C91E920, 7C927C58&#41;
0022EFD8  61004A78  &#40;0022EFF0, 0022F03C, 0022EFF0, 002310A4&#41;
0022FF88  6100594F  &#40;00000000, 00000000, 00000000, 00000000&#41;
End of stack trace
which is clearly a different problem. Or have you provided a 64 bit version only?

Sven

Hi Sven,

Sorry forgot to mention it's compiles with cygwin version of gpc so it'll need cygwin1.dll. I've update the package now in case you want to re-download.

Jim.

Jim Ablett · Post by **Jim Ablett** » Mon Jan 09, 2012 6:08 pm

sje wrote:The program source has non-local goto statements, and that's a show-stopper in Free Pascal. I don't use GNU Pascal, so it may behave differently.

I strongly suspect a problem with the old style I/O calls, all of which were designed for a non-interactive environment. Perhaps the program is stalled waiting for input or it got an end-of-file and doesn't know what to do.

Thanks Steven,

an I/O problem is what I suspected myself. Still it would be interesting if someone could get it working properly.

Jim.

Sven · Post by **Sven** » Mon Jan 09, 2012 6:28 pm

sje wrote:The program source has non-local goto statements, and that's a show-stopper in Free Pascal. I don't use GNU Pascal, so it may behave differently.

According to "ByteChess.txt" inside the package, GNU Pascal can handle those non-local goto statements. I also found no mentioning of that as a known problem at the GNU pascal site.

Sven

Carey · Post by **Carey** » Tue Jan 10, 2012 5:28 am

Jim Ablett wrote:I've actually compiled the program as got it running on Windows using the Gnu Pascal Compiler.
I can make one move and get one move back from the engine then it just seem to get stuck in an endless loop.
I cannot input any more moves and have to do a Ctrl+Break to get out of it.
This is really frustrating - so near yet so far.

Here's the src & semi-working windows exe > http://dl.dropbox.com/u/5047625/chess05%20iso%20src.zip

thanks,
Jim.

Jim

I'm the idiot who typed that sucker in and got it to work. I did play several complete games and it seemed to work tolerably well.

I don't know where you got your copy, but you might not have gotten the final version.

Even on my now defunct website, I had a minor correction text file about a typo in the castling constants that I never got around to applying to the source itself.

I'll email you the 'final' source zip file as well as the last error I found.

When I compiled it, I probably did it for DOS, but I have no idea what version of the compilers I used.

Carey

Jim Ablett · Post by **Jim Ablett** » Tue Jan 10, 2012 2:38 pm

Carey wrote:
Jim Ablett wrote:I've actually compiled the program as got it running on Windows using the Gnu Pascal Compiler.
I can make one move and get one move back from the engine then it just seem to get stuck in an endless loop.
I cannot input any more moves and have to do a Ctrl+Break to get out of it.
This is really frustrating - so near yet so far.

Here's the src & semi-working windows exe > http://dl.dropbox.com/u/5047625/chess05%20iso%20src.zip

thanks,
Jim.
Jim

I'm the idiot who typed that sucker in and got it to work. I did play several complete games and it seemed to work tolerably well.

I don't know where you got your copy, but you might not have gotten the final version.

Even on my now defunct website, I had a minor correction text file about a typo in the castling constants that I never got around to applying to the source itself.

I'll email you the 'final' source zip file as well as the last error I found.

When I compiled it, I probably did it for DOS, but I have no idea what version of the compilers I used.

Carey

Here is new working windows executable compiled using Carey's updated src > http://dl.dropbox.com/u/5047625/chess05-ja.zip

All credit is due to Carey who has done a lot of work on bugfixing the src and now you should be able to compile a working binary with
Turbo Pascal & Free Pascal (untested) as well as Gnu Pascal which I used.

I made a few cosmetic changes to the console output and tweaked the internals a bit so it plays a stronger game with longer thinking time.

Many thanks for your help Carey,

Jim.

Jim Ablett · Post by **Jim Ablett** » Tue Jan 10, 2012 10:57 pm

Please re-download.
Found buffer overflow in eval.
(my fault - increased search depth too much),

http://dl.dropbox.com/u/5047625/chess05-ja.zip

best,
Jim.

nanochess · Post by **nanochess** » Wed Jan 11, 2012 8:51 pm

I'm fact I have the source of Chess 0.5 from "The Byte Book of Pascal", the listings are faded, I've typed it one time

and managed to get it compiled, I don't remember what I did with it, but it was so slow and not as strong as I thought

.
Any idea of ELO rating with your modifications?

Carey · Post by **Carey** » Thu Jan 12, 2012 4:21 am

Jim Ablett wrote:Please re-download.
Found buffer overflow in eval.
(my fault - increased search depth too much),

http://dl.dropbox.com/u/5047625/chess05-ja.zip

best,
Jim.

Laugh... Way to go, Jim.

I guess you forgot this thing was designed back in the days when people would be running it on interpreted UCSD Pascal and 8080's...

(grin)

Need a little help from a Pascal guru (Chess05 Frey/Atkin)

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki

Re: Need a little help from a Pascal guru (Chess05 Frey/Atki