@Dennis What is ypur opinion on adding image based languages like Piet? Say as XPM3 format image becauce they can be typed out by hand fairly easily then fed through ImageMagick before going to the interpreter?
Whether you take it as an xpm file or a hexdump I don't think it really matters, all it would change is whether you use ImageMagick or xxd to do conversion to the format expected by the interpreter.
@Dennis Is this an accurate description of the technology behind tio.run#apl-dyalog? seunshare'd Dyalog 16.0 Unicode/Classic on 2 or 3 load-balanced SELinux arenas
@Adám I'd say Dyalog 16.0 Unicode/Classic in an SELinux sandbox on 4 load-balanced arenas. While TIO doesn't use the sandbox command (because it's slow), it does emulate it with seunshare.
I'd have to start saving that. I gave you data for the last 4 weeks since that's what I have. Those logs alone (only the web server) weigh 240 MB, so they get pruned periodically.
Btw compilation printed a lot of warnings. The sign differences and parens suggestions may be safe to ignore, but the lack of a return statement in int implicit_input(int, int) is undefined behavior.
@Dennis BEGIN isn't needed, and the Hello, World! I found on Wikipedia didn't include it, but I can't say which way is more idiomatic for a language from almost five decades ago.
@Dennis Btw, the new format has two hashes ##. Some programs, like MS Office chop URLs before the second hash, apparently reasoning that that is a bookmark, and the rest is garbage.
Everything before # is the URL known to the browser. All TIO requests are now to https://tio.run/, so every permalink lets you fetch the HTML page from cache. With %23 (or the old /nexus/<language> format), you download the page for each permalink (or each language).
@Dennis As part of my presentation, I'll click links in my PowerPoint slide. I've made sure to use uncached TIO URLs so that TIO will not bring the result up upon page load. This way, I can click Execute and the audience can see the Output appear.