Dear Unicon users,

Few months ago I posted the link to one my AI program.

http://chem.pmf.hr/~kazimir/Finder.html

That program was my main reason to migrate from C++ on Icon and later Unicon. The program was tested extensively and worked for total of nearly 100 days without substantial problems. Some of the experiences will be, hopefully, published in the next issues of The Generator.

In further versions of the program, dynamic data structures, i.e. graphs have to be used more extensively. However, some concerns were raised a year ago, when the program that behaves less than satisfactory was posted here. C. Jeffery announced the fix, but one which was reported work properly only on Unix.

These days issue becomes critical for me. I'm sending similar program for a test again. The program constructs dynamic structures resembling "real ones" closely. I need help of someone who is willing to execute it on some Linux Unicon (until it crashes or start "thrashing") and send the program output publicly or privately. Also, if someone uses other Windows or Linux programming tools (Jcon, Python, Lisp, Ruby, Wrapl, anything) regularly, and is willing to translate and test the program and send information, it will help me to make informed decision.


Sincerely, Kazimir Majorinc



Appendix 1: the program
##################################################################
$define RANDOM ?
$define CARDINAL *

procedure main()

  writes("Do you want to test (1) lists or (2) sets? ")

  if read()=1
    then {CONTAINER:=list; INSERT:=put}
    else {CONTAINER:=set; INSERT:=insert}

  t:=&time
  nodes := CONTAINER()
  edges:=0
  write("Nodes, edges, seconds.")

  repeat
  {  new_node:=CONTAINER()
     every 1 to 2*sqrt(CARDINAL(nodes)) do
     { old_node:=RANDOM( nodes )
       INSERT(new_node, old_node)
       INSERT(old_node, new_node)
       edges+:=1
     }
     INSERT(nodes, new_node)
     if CARDINAL( nodes ) % 500 =0 then
       write( CARDINAL( nodes), ", ", edges, ", ", (&time-t)/1000. )
  }

end
#####################################################################
Appendix 2: program output on Windows XP, 256 MB, 2.4 GHz P4.

C:\Unicon>crash
Do you want to test (1) lists or (2) sets? 1
Nodes, edges, seconds.
500, 14641, 0.015
1000, 41640, 0.046
1500, 76682, 0.109
2000, 118225, 0.203
2500, 165375, 0.281
3000, 217550, 0.359
3500, 274291, 0.453
4000, 335265, 0.5620000000000001

[K.M.'s note: the program now crashes silently.]

C:\Unicon>crash
Do you want to test (1) lists or (2) sets? 2
Nodes, edges, seconds.
500, 14641, 0.047
1000, 41640, 0.297
1500, 76682, 0.797
2000, 118225, 1.515
2500, 165375, 2.718
3000, 217550, 4.578

Run-time error 307
File crash.icn; Line 23
inadequate space in block region
Traceback:
   main()
   insert(set_1993(147),set_3440(16)) from line 23 in crash.icn

#################################################
Appendix 3; Program output on Windows XP, 512 MB, 2.8 GHz P4

C:\downloads>crash
Do you want to test (1) lists or (2) sets? 1
Nodes, edges, seconds.
500, 14641, 0.015
1000, 41640, 0.046
1500, 76682, 0.078
2000, 118225, 0.14
2500, 165375, 0.187
3000, 217550, 0.25
3500, 274291, 0.406
4000, 335265, 0.468
4500, 400191, 0.5620000000000001
5000, 468854, 0.64
5500, 541051, 0.734
6000, 616616, 0.843
6500, 695419, 0.953

[K.M.'s note: the program now crashes silently.]

C:\downloads>crash
Do you want to test (1) lists or (2) sets? 2
Nodes, edges, seconds.
500, 14641, 0.047
1000, 41640, 0.219
1500, 76682, 0.61
2000, 118225, 1.188
2500, 165375, 2.11
3000, 217550, 3.125
3500, 274291, 4.516
4000, 335265, 6.344
4500, 400191, 9.141
5000, 468854, 12.141
5500, 541051, 15.891
6000, 616616, 20.156
6500, 695419, 25.39
7000, 777322, 32.484
7500, 862210, 41.421

[K.M.'s note: the program now crashes silently.]



-------------------------------------------------------
This SF.Net email is sponsored by: InterSystems CACHE
FREE OODBMS DOWNLOAD - A multidimensional database that combines
robust object and relational technologies, making it a perfect match
for Java, C++,COM, XML, ODBC and JDBC. www.intersystems.com/match8
_______________________________________________
Unicon-group mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/unicon-group

Reply via email to