search

openjournals / joss

The Journal of Open Source Software
https://joss.theoj.org
MIT License
1.51k stars 184 forks source link

[proposal]: Use quarto/myst for paper submissions? #1323

Open sneakers-the-rat opened 6 months ago

sneakers-the-rat commented 6 months ago

Myst is lovely, and many ppl have also been switching to using quarto to write documents, which is also lovely! It seems like as a software journal we might want to support these formats, and we could certainly preserve backwards compatibility/multiple formats if we want to so it's not an either/or but both/and!

I can make a more detailed proposal if folks are interested and make a PR, but I think it would be great if we could have PDF and web-first versions of all our papers and start trying to win vs. commercial journals on ethics, price, and tech <3

Switching Costs

General

HTML

PDF

rowanc1 commented 6 months ago

The scipy proceedings will be moving to using MyST (https://mystmd.org) for 2024 submissions, and back-porting previous articles (which were written mostly in RST and LaTeX). The new proceedings website for SciPy is here: https://proceedings.scipy.org, the american geophysical union is also starting to embrace MyST (presentation).

Happy to help field MyST questions – I am one of the developers on MyST and on the steering council for the executable books/myst project. 🚀

jedbrown commented 6 months ago

I'm a fan of MyST and would like to see high-quality HTML output. One issue will be that it's not uncommon for current authors to inline some LaTeX bits and that would need to be forbidden if we're publishing the same source also as HTML. There are some examples in the sample paper, but that list is far from exhaustive. In general, we would need to somehow verify both versions, and that sounds error-prone.

There has come up previously on Slack; see this thread that discussed citations (MySTmd looks like closer feature parity than MyST-parser) and potentially losing PDF/A-3b compliance. I think if we can solve these technical issues, then it would be exciting to proceed.

sneakers-the-rat commented 6 months ago

Could we do a 2 stage render and preprocess LaTeX, or else could we use an HTML template that includes MathJaX? Taking a look at the citations, i think we could also preproc those.

Or, we could have two document modes? people who want to write in MyST can do so, and we can keep rendering for current format?

edit: starting a checklist of problems in OP

edit2: also @rowanc1 hi! and good eye <3

jromanowska commented 6 months ago

I have been using Quarto for a while, but MyST seems also very well suited for the purpose of JOSS papers!

mstimberg commented 6 months ago

One issue will be that it's not uncommon for current authors to inline some LaTeX bits and that would need to be forbidden if we're publishing the same source also as HTML.

Just mentioning here that this already causes issues for the JATS (=XML) output with the current system every once in a while. This is always a bit annoying since it only gets built after a recommend-accept. E.g. this issue with multiple labels in multi-line equations (Slack) – which was relatively easy to workaround, though.

mstimberg commented 6 months ago

Actually, if we are only interested in providing an HTML version (as opposed to having richer authoring tools with Myst/Quarto), then another option might be to have a nice reader that directly uses the JAST output. eLife invested in their eLife Lens reader a while ago (example rendering: https://lens.elifesciences.org/00778), but it doesn't seem to be developed anymore: https://github.com/elifesciences/lens