Out of memory?

[Frozone212]

I typed in the entirety of Hidden Maze for the Commodore PET and i get an error in 810

Out of Memory

 

Any ideas?

maze chase.d64

 

Configuration:

2001 8n 4k

IO: 2048 bytes

40 column etc

Winvice 3.5 win 10

Edited July 5, 2021 by Frozone212
added configuration

[carlsson]

You have a lot of recursive GOSUB without RETURN. The stack can only hold a smaller amount of nested subroutine calls. I would suggest a RETURN on line 670. I don't know what 730 is supposed to do but as it stands it makes up for an infinite loop until you ran out of stack.

 

Also I believe 4K RAM is too little for your program, since you're allocating an array M(630) which itself would take up 3162 bytes out of the 3069 available. If you change it to M%(630), it would "only" use 1269 bytes but be a little bit slower plus that you can only store numbers -32768 to +32767. Now it seems that you're only using it for numbers 0-255 anyway, so you could set up a buffer in RAM that is as large as you need and change pointers so you don't overwrite it with variables. That means assignments to M(X) would be replaced by POKE M+X and fetching values would be PEEK(M+X).

 

With the current program, even an 8K PET would barely fit it but 16K or more would probably be fine, once you have solved the GOSUB issues that are independent on how much RAM you have.

 

   10 goto 400
  100 rem
  110 gosub 730:printct:forj=0to3:d=a:c=s2120 fori=1to7:pokec,m(d)
  130 pokec-e(j),m(d-d(j)):pokec+e(j),m(d+d(j))
  140 ifm(d)=wlthen160
  150 d=d+a(j)/2:c=c+e(j+1):nexti
  160 nextj:pokes2,240:m(a)=240:return
  200 rem
  210 fori=1toh:forj=2toh+1:pokes2,j
  220 m((i*(h+2))+j)=wl:nextj:nexti
  310 m(a)=4
  320 j=int(rnd(r)*4):z=j:pokes2,j
  330 b=a+a(j):ifm(b)=wlthenm(b)=j:m(a+a(j)/2)=hl:a=b:goto320
  340 j=(j+1)*-(j<3):ifj<>zthen330
  350 j=m(a):m(a)=hl:ifj<4thena=a-a(j):goto320
  360 t1=(3*h)+5:t2=int(rnd(r)*2):ifint(t2/2)*2=t2thent1=(2*h)+6
  362 q1=-1:q2=-(h+1):q3=h+3
  364 ift1=(2*h)+6thenq1=-q1:q2=-q2:q3=-q3
  370 m(z)=69:m(z+q2)=wl:m(z-q1)=wl:m(z+q3)=wl
  380 fori=1toh
  382 m(3*(h+2)+4+int(rnd(r)*(h-5))*(h+2)+int(rnd(r)*(h-5)))=hl
  384 nexti:return
  400 rem
  410 gosub 600:gosub 730:gosub 200:gosub 100
  415 geta$:ifa$=""thenct=ct+1:goto 415
  420 j=-( (a$="w")+2(a$="s")+3*(a$="d") )
  425 ifa$="?" then 500
  435 a2=a+a(j)/2
  440 ifm(a2)=69 then800
  445 ifm(a2)<>hlthen415
  450 m(a)=hl:a=a2:gosub100:goto415
  510 gosub 730:f=0:fori=1toh:forj=2toh+1:l=(i*(h+2))+j
  520 pokes+j+f,m(l):nextj
  600 rem
  605 vl=40:h=23:ff=vl
  610 a(0)=2:a(1)=-(h+2):a(2)=-2:a(3)=(h+2)*2
  620 d(0)=h+2:d(1)=1:d(2)=-(h+2):d(3)=-1
  630 e(0)=vl:e(1)=1:e(2)=-vl:e(3)=-1:e(4)=vl
  640 wl=160:hl=32:s2=32768+vl*12+int(vl/2):ct=0
  650 s=s2-vl*((h+1)/2)+ff-(h+3)/2
  660 a=(h+2)*(h+1)/2+(h+3)/2:dimm(630)
  730 rem
  810 gosub 730:print "your score";ct:end
  900 rem
  920 print "hidden maze":print
  930 print "get out as quickly as you can using control keys"
  940 print "w is up, s is down and d is right"
  950 print "press ? to show full maze"
  960 print "this costs points"
  980 r=0:print  "{rvon}{wht}{del}{CTRL-U}{rvon}{swlc}"
  990 return

[bluejay]

Tl;dr: if you’re out of memory, get yourself more. 
 

 

Why would you use a 4K system in emulation when you can have as much as you want?

[carlsson]

If one is developing with a certain machine in mind to run the software on later, it probably is a good idea to emulate the same spec during development. But as I told, this is a two-fold problem.

 

10 a=0

15 a=a+1:gosub15

 

It crashes out when a=24 so that is the maximum number of nested levels if you have no other values on the stack. In the real world, the number probably is a bit smaller. Normally that isn't a problem if you properly return from every subroutine but if you apply spaghetti code and GOTO yourself back to the main program, eventually this will be a problem.

[Frozone212]

i'll try to find a disk version. Typing it is a nightmare

[carlsson]

Aha. I didn't trigger on "typed in". I thought it was an own creation.

 

Well, the listing is here: https://archive.org/details/1982-12-compute-magazine/page/n153/mode/2up

 

It explains line 730 where you changed a PRINT:RETURN into a REM statement. I didn't check which other places that differ, but in general probably most of it is fine except the article doesn't say how little RAM you can expect the program to run on. I still think 8K would be a shoehorn but you can try.

[Frozone212]

it says bad subscript in 330. any ideas?

Edited July 7, 2021 by Frozone212
caught another one

[bluejay]

3 hours ago, Frozone212 said:

it says bad subscript in 330. any ideas?

It says “BAD SUBSCRIPT  The program was trying to reference an element of an array whose number is outside of the range specified in the DIM statement.”

 

Perhaps a typo on where you defined whatever array it’s referring to?

[carlsson]

Print the values of J and B when it happens so you can see whichever of those are out of range.

 

Actually up until 5 minutes ago, I didn't know that Commodore BASIC accepts that you have both floating point variable A and the matrix A(N) at the same time. I knew that A and A% or A and A$ may coexist, but not that matrix names were separate too. I suppose that means you can have A$ and A$(N) at the same time too.