[INFRA] updating remark, CIs, contributor docs

[sappelhoff]

was also trying to get a global install to work, but it seems tricky.

Perhaps it'd be easier to do a local install both here in the CI, and as an advertisement in the CONTRIBUTING guide.

We'd then add node_modules, package.json and package-lock.json to our .gitignore file so that nobody accidentally commits these folders/files, because they are created during local npm installs.

See also: https://github.com/remarkjs/remark-lint

[Remi-Gau]

does runnning remark still mess us the macro code mentioned in that other PR (see comment ).

Or should I:

• A) open an issue about this (and take care of it in another PR)
• B) see if we can get the linter to ignore those sections?

[sappelhoff]

does runnning remark still mess us the macro code mentioned in that other PR (see comment ).

I didn't check -- but I think that's still an issue.

I would go for A) --> opening an issue and doing a PR next time we have time and energy for it

but you can also push to this PR if you'd rather do B)

[Remi-Gau]

does runnning remark still mess us the macro code mentioned in that other PR (see comment ).

I didn't check -- but I think that's still an issue.

I would go for A) --> opening an issue and doing a PR next time we have time and energy for it

but you can also push to this PR if you'd rather do B)

I quickly check locally and adding :

<!-- lint disable -->

Does not seem to prevent remark to try to fix those "macros": so I won't look further into it for now, but I will add a message in the CONTRIBUTING.md to tell contributors to be careful which hunk they commit after fixing.

[Remi-Gau]

Strangely enough even though remark does not flag files as problematic it still fixes stuff when asked to.

The most annoying (potentially problematic)

• changing unordered list from using - to using *
• changing ordered list from using 1. to actually using the number of the item
• changes literal hyperlinks URLs from [URL](URL) to <URL>

Clearly not the end of the world but someone using this fix could end up committing quite a few extra things

Could look into a finer control of remark But I would flag that as low priority for the reason that we have not seen that happening all over the place yet.

---

More problematic the tendency to escape all _ and & with a \ in all the URLs.

---

In general, from a new user perspective, all these extra changes could really look like they broke everything and could be very confusing.

[Remi-Gau]

LGTM

[sappelhoff]

Thanks for review and suggestions Remi!