Unvendor the vignettes/ directory for R packages

[jennybc]

Description

In an R package, the vignettes/ directory holds executable documents that are rendered into documentation at package build time. Currently linguist incorrectly labels this folder as vendored code, which has multiple downstream downsides, including causing exclusion of the files below vignettes/ from code search. (It seems like linguist might use the "vendored" label pragmatically, but then other projects use this data and take the label at face value?)

This was introduced in 9ae0bdb, which was wrong IMO. (Or, at least, has become wrong as other folks started to use linguist's vendored list as if it really means vendored).

The executable documents in vignettes/ tend to be RMarkdown documents, a file type that that is already (correctly) handled elsewhere in linguist:

In languages.yml:

https://github.com/github/linguist/blob/16a7a528ab840dd24188edda671d31881a67e527/lib/linguist/languages.yml#L5428-L5439

The executable documents we keep in vignettes/ are compiled to PDF or html files that live in inst/doc/, which is already handled (by which I mean "excluded"), thanks to #4068:

https://github.com/github/linguist/blob/16a7a528ab840dd24188edda671d31881a67e527/lib/linguist/documentation.yml#L19

linguist is correctly ignoring the rendered documentation but is incorrectly labelling the source as vendored code, which it's not.

Checklist:

• I am adding new or changing current functionality
  • I have added or updated the tests for the new or changed functionality.

It's not clear to me that this functionality is tested. Sorry if that's just user error.

[jennybc]

If you want to see a bunch of wild-caught R packages, we have a read-only mirror of CRAN here: https://github.com/cran.

Some packages, but not all, have vignettes/ and that mostly contains .Rmd files. When vignettes/ is present, the corresponding rendered documentation is found int inst/doc/.

[lildude]

I know nothing about R but your explanation makes sense and checks out.

[hadley]

@lildude any sense of how long it will take this change to percolate in code indexing on github.com?

[lildude]

Changes don't appear on GitHub until a new Linguist release is made and then deployed. I'm in the process of putting together the next release now. Repo's will also need a push for them to be reanalysed with the newer Linguist knowledge to take effect.

Search won't update until go-enry pulls in the lastest Linguist release.

[hadley]

@lildude thanks!