Hacker News Reader

Comment by TheCondor | Hacker News Reader

I suspect it is getting better but I've experienced some breaking changes with Typst. Nothing that was terribly difficult to fix, debugging it wasn't obvious though.

There seems to be a huge amount of folks that want typst to work. I respect TeX and LaTeX, immensely, but it's so vast and byzantine. Maybe I don't know where or what, but some kind of clean LaTeX "distribution" is needed. It seems like you could build it in to containers or something. Just have some way that sort of makes it more of an atomic unit or something. I don't know how many times I've pulled down a template started to build it and something was missing.

It's good to see innovation in this space and people using it.

View on HN

> Maybe I don't know where or what, but some kind of clean LaTeX "distribution" is needed.

Unfortunately it’s a bit like asking for a “clean” Python distribution or a “safe” subset of C++. All the baroque mess is necessary to get working the many legacy packages that many people depend upon. And not just depend upon, but depend upon with all their concomitant foibles, bugs, and learned work-arounds.

https://xkcd.com/1172/

> some kind of clean LaTeX "distribution" is needed

I don't think a distro would do much to alleviate my gripes with (La)TeX which is its language and VM. TeX-the-language violates literally every single architectural principle we have come to rally around in sane programming languages. Whether you start by reading the TeX book or by copy-pasting from similar LaTeX documents, in my experience you'll soon hit a wall. No simple if/then/else. No just-works maths. No data types. No namespaces. No modularization. Every single solution in the language that Knuth comes up for any kind of problem he solved in TeX is a new edition of facepalm.jpg, everything you don't know and look up or (more likely) post to TeX StackExchange is answered with something that will make you stare in disbelief. Nothing, absolutely nothing that you do in the language to change anything in your document is guaranteed to not accidentally affect something else in your document, and your only way to check for it is eyeballing every single page of output.

Couldn't have said it better myself. Plus, you're not supposed to notice any of this. If you do, the TeX masters will put their fingers in their ears and tell you it's PEBKAC. "Get good, n00b. RTFM. It's not happening, but if it is, it's a good thing. It has to be designed this way, actually, you're just too thick to understand why. No, I won't explain."

Also, every part of the system is like this, not just the language-qua-language. The architecture of the compiler is like this. The compiler UI is like this (yes for every TeX engine). The package management/distribution situation is like this.

It basically has two things going for it: it's highly portable, backwards-compatible, and stable, and it produces really pretty documents.

I know I'm late to this, but there is a minimal distribution of LaTeX that I used for e.g. writing my master's thesis. It's called TinyTeX.

https://yihui.org/tinytex/

Sounds like they could benefit from the R (language) build model that checks library changes against registered consumers of that library before merging.

https://jtibs.substack.com/p/if-all-the-world-were-a-monorep...

Reply on news.ycombinator.com