A basic BASIC question.

JohnnyRockets · Post by **JohnnyRockets** » Mon Nov 05, 2012 3:45 pm

Hi!

Considering the following program listing from the VIC-20 Commodore Programmer's Reference Manual:

Here is an inefficient program that doesn't use GOSUB:
100 PRINT "THIS PROGRAM PRINTS

Boray · Post by **Boray** » Mon Nov 05, 2012 3:56 pm

Every basic command word, such as GOSUB or PRINT takes ONE byte in memory, so maybe the second program really takes less memory. You could make it use even less memory by putting more than one command on the same line using a : ... other than that, using Gosub in that example would clearly be better if you decide to change the delay loop. Then you only need to change it in one place.

Witzo · Post by **Witzo** » Mon Nov 05, 2012 4:03 pm

The GOSUBS that jump to the loop actually each take up less memory than the wait loop.

Also, basic is placed in memory as 'tokens', one byte (thanks Boray) codes that represent each basic keyword.

Enter both and type
?FRE(0)
after entering each to see how many bytes in memory are left.

Pedro Lambrini · Post by **Pedro Lambrini** » Mon Nov 05, 2012 4:03 pm

Why is it a gosub and not a goto? That's something I've never really understood...

Witzo · Post by **Witzo** » Mon Nov 05, 2012 4:08 pm

GOSUB remembers where to go back after a RETURN.
GOTO can't be followed by a RETURN. Going back would require a new GOTO pointing to the line after the first GOTO.
In the second example, after jumping to the wait loop, you don't know where to return to to print the next line. Was it the first, second or third line that was just printed? Conveniently, GOSUB has noted in memory where the next line was.

Pedro Lambrini · Post by **Pedro Lambrini** » Mon Nov 05, 2012 4:32 pm

Aah, thank you for your clear and concise answer. I can't tell you how silly I feel now! Still, you learn something new every day and today I have you to thank for that.

Witzo · Post by **Witzo** » Mon Nov 05, 2012 4:37 pm

Glad to be of service!
Usually it's me asking for advice from the local thaumaturgs.

Boray · Post by **Boray** » Mon Nov 05, 2012 4:48 pm

Here is my text on basic optimization:
http://user.tninet.se/~pug510w/datormuseum/optim.html

Maybe of more help for non-beginners though...

Witzo · Post by **Witzo** » Mon Nov 05, 2012 4:59 pm

Boray wrote:Here is my text on basic optimization:
http://user.tninet.se/~pug510w/datormuseum/optim.html

Maybe of more help for non-beginners though...

Awesome. I never knew any of those things except for squeezing as much as possible on a line.
What is the maximum line length? Two screen lines?

Boray · Post by **Boray** » Mon Nov 05, 2012 5:03 pm

It's four lines on a vic-20 and two lines on a c64, but you can make them even longer if you use command word abbreviations when writing the line. But then it will be hard to edit them afterwards. I think the very maximum is 255 or 256 characters.

Witzo · Post by **Witzo** » Tue Nov 06, 2012 3:53 am

Boray wrote:It's four lines on a vic-20 and two lines on a c64, but you can make them even longer if you use command word abbreviations when writing the line. But then it will be hard to edit them afterwards. I think the very maximum is 255 or 256 characters.

Yes, it comes back to me now.
The two-character one-normal-one-shifted abbreviations that expanded beyond the fourth line on screen.

wimoos · Post by **wimoos** » Tue Nov 06, 2012 5:01 am

The lines in the program in memory (with the tokenized keywords) have a maximum length of 255 bytes. Typing a line will never reach that length. POKEing around will, but that's extremely cumbersome.

Using shifted characters will indeed help. Funny thing is that when you shift the last character of a keyword, it will not be recognized ! That is because in the reference table of keywords in ROM, the last characters are already "shifted" as to signify the end of the keyword. Comparing the typed-in keyword with the keyword in memory stops when a difference of $80 is found. When both last characters are shifted, the difference is 0, which explains this odd behaviour.

Regards,

Wim

JohnnyRockets · Post by **JohnnyRockets** » Tue Nov 06, 2012 3:37 pm

Witzo wrote:The GOSUBS that jump to the loop actually each take up less memory than the wait loop.

Also, basic is placed in memory as 'tokens', one byte (thanks Boray) codes that represent each basic keyword.

Enter both and type
?FRE(0)
after entering each to see how many bytes in memory are left.

Thanks, and very interesting!

The difference was:

3376 free without the gosubs

3390 free with the gosubs

= 14 byte difference

What is taking up those 14 bytes? Or is that mystery going to be found by examining the machine code?

Witzo · Post by **Witzo** » Tue Nov 06, 2012 4:58 pm

My guess it's the copies of the wait loop, FOR L = 1 TO 500 : NEXT; 3 tokens FOR-TO-NEXT, several numbers, several times over.
The gosubs are each only one token for GOSUB and the target line number.

JohnnyRockets · Post by **JohnnyRockets** » Tue Nov 06, 2012 5:15 pm

Witzo wrote:My guess it's the copies of the wait loop, FOR L = 1 TO 500 : NEXT; 3 tokens FOR-TO-NEXT, several numbers, several times over.
The gosubs are each only one token for GOSUB and the target line number.

Makes good sense! I appreciate it!