how does this compare to the TeX family or SILE mentioned here recently?
As bad as TeX is, highly imperative/script-y OCaml code is not an improvement on TeX.
You’re criticism of this code is that it’s imperative? That’s literally all code in your browser, operating system (unless I guess you’re using xmonad), lobste.rs, and most things in general.
Functional code can be great! But there’s also lots of great code that’s not functional.
TeX, though, is known to be particularly bad—I think SILE did a good job of expressing the problems. While idk how patoline compares to SILE, I’d be very surprised if they didn’t manage to improve TeX, at least as far as code quality goes.
That’s not my only criticism, it’s just the most salient one. I’ve written a 1,300 page book with LaTeX, among other things. I know where the rough spots are. Patoline solves none of them.
What are the rough spots that you want to see solved with LaTeX?;
There is also Lout http://savannah.nongnu.org/projects/lout and Nonpareil http://www.it.usyd.edu.au/~jeff/nonpareil/. Lout has one of the weirdest C codebase I’ve seen and Nonpareil is a spec without code, but I used Lout for all my typesetting in university – it has a nice syntax, is very compact (a few MBS) and easy to use.