Just had an awesome idea; in #gemini you could inspect the characters in code blocks and make a very good educated guess if it's pointless ASCII art or if it actually has code in. Not only can I have the current setting in Ariane to collapse all code blocks, I can have an additional one that will just skip rendering any code blocks that look like ASCII art altogether.
I can't think of legitimate code syntax where there'd be say 5 consecutive identical non-whitespace characters. It might be that simple.
@oppen i thought i had a counter-example from one of my students, but they 'only' used 4 consecutive characters for some variable names 😅
@oppen a possible legitimate code syntax with that amount of repetition would be brainf*ck code (i don't know how common would that kind of content be in gemini, though 😅)
anyways, now you left me wondering: wouldn't that criterion result in false negatives? ascii art that doesn't resort to repetition.
@chirrolafupa I'd use it as one of a few rules, and include some common character blocks in another check too. As soon as I can get to a computer I'm going to copy all the ASCII art I can find in Geminispace and start. Writing some tests
@oppen Lisp and Scheme easily accumulate more than 5 consecutive ")"''s.
@vincent I'll exclude ( and )
@oppen I can't think of any other examples. Brilliant heuristic!
@oppen I don't think you'd ever expect to see that many "(", in contrast. You could probably get a way with just a ")" special case.
@oppen Printing out a row of dashes.
Thinking about this some more, it'd be easier to have an array of 'normal' characters present in a wide range of programming languages and calculate a percentage from that. Some ASCII art is made up of just normal characters, I could accept that, or could try and look for a valid combination of letter and symbols. This is a perfect TDD scenario.