Procedural Generated Mazes

I have created a small program that generates a random maze. This was the result of some research into an unknown future project. Does or can anyone else make use of this for whatever purpose.

It can save off the generated maze data to import or merge with your own projects. It is very basic so feel free to tamper with it. I would be interested in what it can be used for or how it can evolve.

It uses an algorithm called a recursive backtracker and the algorithm is in a text file in the app and is also embedded in the code as comments.

It is coded in BASIC and is reasonably fast. Eg.

These are whilst it is drawing it. Anyhow there we have it. Let me know

Greg

Edited because Textile keeps lying to me

VDU23,128,1,2,4,8,16,32,64,128,23,129,128,64,32,16,8,4,2,1
REP.VDU127+RND(2):U.0

Amazing…

https://10print.org/ is relevant.

10print

<Guy writes book on one line of code>
<Creates website for book of one line of code>
<Website sells T-shirts containing one line of code>
<One line of code is now clothing people and generating an income>

<looks at wasted life>

VDU23,128,1,2,4,8,16,32,64,128,23,129,128,64,32,16,8,4,2,1 REP.VDU127+RND(2):U.0

Missing Variable

or, more helpfully, missing semicolon.

They’re two separate lines mate, but feel free to concatenate them with a colon.

Nice , but is there a solution? :-)
For those interested in mazes I had compiled Gareth McCaughan’s version.
don’t know the algorithm used, and interestingly the program can give the solution…
Drawing files are in Postscript format.
With C source
https://jeanmichelb.riscos.fr/Download/MazeGMcC.zip

http://www.riscos-digitalcd.net/games/games.zip contains !Laby_Man a little BASIC game which generates mazes. I can’t remember where maze generator code came from though.

Only one problem with the one liner, it doesn’t generate a maze. Looks nice though.

Tough crowd!

Perhaps you’d prefer this:

And that was how many lines? ;)

@nemo
Nice. BASIC ?

@nemo

On closer inspection the maze fails. Sorry for the bearer of such news :-)

Rick requested

how many lines?

It was supposed to be one long line, but Greg noticed

the maze fails

Heisenbug! How did that happen? Not possible.

I’m going to blame a cosmic ray, because every time I retry it and test with a flood fill it’s perfect. Seriously, how did that happen?

The advantage of recording something as a GIF is you can play it back in slow motion.

The bug was in the placement of The Very First Pixel. sigh.

They’re two separate lines mate

Very Interesting.

On a Pi 4 with Netsurf (which I was using), I see your post as one line.

VDU23,128,1,2,4,8,16,32,64,128,23,129,128,64,32,16,8,4,2,1 REP.VDU127+RND(2):U.0

Iris shows only one line as well.

What do you use to see it as two lines?

For me, I only see two separate lines, if I view it on Windows.

For me, I only see two separate lines, if I view it on Windows.

They’re run into a single line in Firefox and Chromium on Linux. That’s why we have bc.. to do code blocks:

VDU23,128,1,2,4,8,16,32,64,128,23,129,128,64,32,16,8,4,2,1
REP.VDU127+RND(2):U.0

Being able to edit a post a day later has both pros and cons.

The Iris screen is still open and it shows one line.
Netsurf now shows two lines.

Being able to edit a post a day later has both pros and cons.

Maybe, that is the reason it now looks right on Windows.

I feel that’s fairly unlikely, given that it’s still displaying as one line on Linux long after you posted that.

The original is wrapped in <p><code> tags, with a line break in the intended place. Most browsers seem to be flattening that out as whitespace, which I think they’re entitled to do.

Mine, by virtue of the bc.. that I used, is wrapped in <pre> tags, and so the line break is retained.

The moral of which is: if you want to post multiple lines of formatted code, use bc.. in front of the first line and follow up with p. if there’s extra commentary to follow.

I apologize.

it just looks different depending on which browser you use.

The moral of which is: if you want to post multiple lines of formatted code, use bc.. in front of the first line and follow up with p. if there’s extra commentary to follow.

Useful advice. It’s a shame the Formatting Help (below) doesn’t list that and the Textile reference linked to is past its use-by date and no longer available.

Textile is rubbish, and was lying to me. Fixed the formatting of the two-liner for posterity.

it just looks different depending on which browser you use

It seems to be the width of the column (which can be adjusted by the draggable input box at the bottom) that is the greatest contributor – I typed two lines, it showed two lines, I had no reason to think it was not two lines. But it turns out that though Textile respects newlines in plain text, it removes them from <code>sections</code>… which is insane.

These
are
new
lines.

These are also new lines.

I now understand why that’s happening. But it’s particularly rubbish in the <code> case.

Heisenbug

The bug in the placement of the first pixel of the emergent-message maze was caused by unwarranted precision – a / where a DIV was intended.

Textile respects newlines in plain text, it removes them from [code]sections[/code]… which is insane.

The solution is to wrap the <code> tags in <pre> tags.

Textile is rubbish.

Fixed that for you.

I think a good 5% ¹ of posts on this forum are about Textile not doing what it’s supposed to, or doing something completely gonzo when it does do what it is (mostly) supposed to. Like the double spaced lines issue.

which is insane.

It will convert anything that looks like a URL into an actual link, which in the case of pre/code sections, just comes out completely wrong. That’s why I sometimes have code snippets that say hxxp://, it’s not a fancy new protocol, it’s simply to stop Textile doing stupid things.

As I’ve been adding bits to this post, I’ve been watching Alan’s post above get ever closer to what he wanted. But, really, one shouldn’t have to fight the parser…

¹ Figure plucked out of my arse.