Denial

I was reading the ROM disassembly and trying to understand how BASIC uses the stack for implementing GOSUB and FOR / NEXT.

Here is an interesting quirk caused by the rather naive way that RETURN scans the stack:
Gives: ?RETURN WITHOUT GOSUB ERROR IN 1100

I am guessing this is well-known, but I was happy to see that the actual behaviour matched my reading of the code

Something else I didn't realise is SPC() and TAB() and handled only for PRINT, so A$ = SPC(10) is an error.

Naive?

That's one of the best things *correctly* handled by the stack: a pending FOR loop in a subroutine is discarded upon return from the subroutine!

What happens here? You enter the sub, open the FOR loop, return from the sub - purging the FOR loop! - and then *errorneously* re-enter the sub routine without GOSUB. No wonders you get a ?RETURN WITHOUT GOSUB error.

Actually quite sensible behaviour. When within a subroutine, you don't need to terminate all pending FOR loops to be "allowed" a RETURN. Comes handy for early exits from search processes.

You find a particularly nice example in my TRON game, where a loop in line 41 searches for an escape path for the enemy light cycle. If it succeeds, it does an early exit - otherwise it returns with the current direction unaltered, letting the enemy crash:
Here's the game. I put it into a collection of 21 games for the unexpanded VIC-20.

Oh my bad, forgot: My excuse is not enough coffee yet.

Bobbi wrote:Something else I didn't realise is SPC() and TAB() and handled only for PRINT, so A$ = SPC(10) is an error.

SPC() and TAB() only position the cursor; they don't produce characters (hence no $ on the function).

I guess for TAB that makes sense (since it uses the screen tab positions). SPC() could have been SPC$() quite easily though I guess.

Some other BASIC dialects support a SPACE$(n) or a STRING$(<string>,n) function, the latter which produces a string of n concatenated instances of <string>.

For a simple replacement of SPACE$(), a combination of a pre-initialised string SP$="<as many spaces as you think you'll need>" and LEFT$(SP$,n) is probably sufficient.

Bobbi wrote:SPC() could have been SPC$() quite easily though I guess.

SPC( and TAB( are format specifiers intended only for the use within PRINT, and for screen output. TAB( isn't even guaranteed to work on anything other than the screen, and SPC( is just implemented by outputting n spaces. Setting up a string for use within the program is slightly more complicated.

Bobbi wrote:SPC() could have been SPC$() quite easily though I guess.

SPC$() would produce actual space characters if it existed. SPC() does not.

PRINT "123←←←" SPC(3) "456"

is different than

PRINT "123←←←" CHR$(32) CHR$(32) CHR$(32) "456"

You're right.

Try out this one for a good laugh what *actually* is PRINTed:
The closing angle brackets are the control characters for CRSR RIGHT.

I discovered this easter egg earlier today (courtesy of TPUG newletter.) I assume it is well known.

LOL, must be revenge for the WAIT 6502 Easter egg

http://www.pagetable.com/?p=43

Apparently so!

Nothing spectacular here. The seeds for RND() contain enough information, that one can print ANY 5 letter string in that way.

One only needs to put out a small search program - and preferably run it on a PC in VICE with "no speed limit" - to reproduce other strings.

While we're at it - enter:
... and see the BASIC crash!

That is a good one!