The comment is about the display layer. Yes [almost] all computer languages are Turing complete; the distinction is about the level at which they produce drawing instructions. Some simply set particular bits in memory that cause pixels to light up. That's essentially how the original Mac worked. More abstract is producing instructions to "draw a rectangle" or "rotate all future lines by 20 degrees." That's the "display list" approach and it's more flexible than manipulating raw pixels.

More flexible still is producing commands that can contain loops like Postscript. Now we're in tricky territory, because it's easy to make a mistake that e.g. produces an infinite loop and the whole display locks up. And it's much harder to reason about such things as the update rate of the drawing process because the drawing commands are themselves Turing complete.

This doesn't matter when you're merely printing because there is no refresh. Not so with an interactive display. That's why the "Display" in "Display Postscript" was a big deal.

I think PostScript was considered “too powerful” for its own good by the industry, which is why it was mostly replaced by PDF (initially a subset of it).

With PostScript, you can have a document that renders differently every time you open it (on purpose). Or a file that will never finish rendering or printing, like the infinite loop you mentioned.

PostScript infinite possibilities are cool and cute, but not all that practical for a graphics description language.

> With PostScript, you can have a document that renders differently every time you open it (on purpose). Or a file that will never finish rendering or printing, like the infinite loop you mentioned.

There was a font, wasn't there -- was it Just van Rossum's Beowulf -- that drew itself slightly differently every time? Can't find a reference now.

edit: it was!

https://www.fontshop.com/families/ff-beowolf

Oh my god. Thanks for this.