I've finally solved one.

Another idea is to DIM it one larger than it needs to be and store the lengths in the first elements. So, if you require a 5 x 10 array, instead of DIM A[5,10], use DIM A[6,11], then store the real data in [1,1] through [5,10], and store the size of the first dimension (5) in [0,1], and the size of the second dimension (10) in [1,0]. Yes, for larger arrays it'll result lots of wasted memory.

Not that I'm aware of. :( Perhaps, during creation you could store it in a variable? Otherwise maybe you could DIM it one larger than it needs to be, store some sort of end-of-data as the last value, then at a future point, walk through the array until you find that value?

Classic 15-tile slide puzzle using only 311 bytes. It may not look like valid code, but it is. When run, the solution is shown. Press a button to scramble the tiles. Then use the directional buttons to move the tiles, or B to quit. A message will appear when you've reached the solution. Key: 4KV4E8HV

Same mistake I often make. For SmileBasic, arrays use [square brackets] instead of (round brackets).

Sorry, I was writing based on memory. Looking at your code again, a better place for the VSYNC would be after the line that contains "BTN=="

Does an error message appear, or are you assuming it's not working because it immediately goes to @LOSE? If the FOR - NEXT loop was intended to create a delay, it should contain a VSYNC. Just after the BUTTON() would probably be a good place for it.

"INC SY -1" in the line above it may be the problem. If the intent is to decrease the value of SY by 1, then replace that with "DEC SY".

Can you add a rotation feature to GPUTCHR, so text can be written sideways or upside-down?

For each chr in the string, calc its location on screen, GSAVE the part of the screen that'll be behind it, rotate those pixels the opposite direction, then GLOAD them back to the screen, GPUTCHR the chr to the screen, GSAVE it, rotate it the right direction, GLOAD it back to the screen.

There are ways to do it, but I've not found any that are as easy as it should be. (1) Use FONTDEF to create your custom sideways / upside-down characters, then use GPUTCHR. - or - (2) Use GPUTCHR to display each character in normal orientation then use GSAVE to copy their pixels into an array, rotate the pixels within the array, then use GLOAD to write them back to the screen.

Finally!

Thanks for the tip about RANDOMIZE. I got it down to 471 bytes.

Oh. I see you found it already. Well, I hope they fix it.

I found the official bug reporting topic. The fastest way to locate it is by its search for its author, MiiVerse user smile5884, then looking at his most recent discussion.

2nd post, the output. It seem that if the destination of the assignment is not subjected to some operation that changes it, it remains a pointer to the original, vs a copy of it.

I agree. 1st post, the code.

Oh. I'll see if I can reproduce the bug. If I recall correctly, SmileBoom had a Discussion topic here for collecting bug reports.

The comment in your code asked if we could do it in 600 bytes. Here's my go at it. Excluding the comment, it's 573 bytes. After launching the game, press a button to scramble the tiles. Use the direction buttons to slide tiles, or press B to exit. The game will detect when the solution is found. After that, press a button to exit.

It depends on what what's being programmed, and how tight the code has to be vs how much time you have to finish the project. Higher level languages are generally faster for development, but their output binary can be much larger than a low-level language's.