Add guidance about GitHub language detection

[Bisaloo]

Feel free to move this issue if you think it belongs to another repo but I think it would make a worthwhile addition to the grooming section.

Some packages (including rOpenSci packages such as skimr, rsnps or rdflib) are mistakenly classified as HTML repos by GitHub. This hinders discoverability as they cannot be found in a search with the language: r filter.

It would be very useful to add pointers about how people can mitigate this issue (e.g. https://github.com/github/linguist#overrides).

[Bisaloo]

Closing as this is apparently already the case. (I should definitely search more before opening new issues)

[maelle]

so https://ropensci.github.io/dev_guide/grooming.html#github-linguist is sufficient then? if not feel free to reopen.

[Bisaloo]

Yes, I'd say so. It's pretty close to what I had in mind anyways.

If I were to modify it, I would maybe more explicitly say that repos that don't follow this tip might not appear in search results with the language: r filter, as I mentioned in my first comment.

The current phrasing ("more discoverable") might not be clear enough to someone who didn't encounter this issue first hand.

[Bisaloo]

On a related note, are you sure that files in docs are ignored by default?

If you store html files in non standard places (not in docs/, e.g. in vignettes/)

See e.g. https://github.com/ropensci/rdflib. I seem to remember that we also had to manually override linguist settings for this in pavo.

Not 100% sure though. I would need to check.

[Bisaloo]

Hum, according to linguist docs, yes, it should be ignored.

https://github.com/github/linguist#documentation

I'm not sure why rdflib is marked as HTML then. Maybe https://github.com/ropensci/rdflib/blob/master/slides/slides.html?

[maelle]

can you open a small PR for the updated phrasing please? :-)