[R] open_dataset - format is unclear

[csgillespie]

Describe the bug, including details regarding any error messages, version, and platform.

library(arrow)
?open_dataset

gives the function sig as

...
format = c("parquet", "arrow", "ipc", "feather", "csv", "tsv", "text", "json"),

However, the docs in ?open_dataset don't mention json. Which would mean it's not supported?

---

If I was being pedantic, write_dataset() is slightly odd.

• CSV == test in open_, but not in write_
• open_ reads in JSON / tsv, but write_ doesn't

Component(s)

R

[thisisnic]

Thanks @csgillespie, this kind of feedback from our user community is super helpful!

the docs in ?open_dataset don't mention json.

Looks like we missed mentioning JSON in the docs when we added that functionality - thanks for spotting that!

CSV == test in open_, but not in write_

Sorry, I'm a but unsure what you mean by this, mind rephrasing?

open_ reads in JSON / tsv, but write_ doesn't

Writing JSON files isn't currently supported in the arrow R package as it's not implemented in Arrow C++; if it's something that would be useful, you could open a C++ feature request ticket?

Writing tsv files is supported, though the released package's version of the docs don't reflect that. However, this is fixed in the dev version (see dev docs here).

[thisisnic]

I've opened #38216 for documenting the option of JSON in the format parameter

[thisisnic]

Aha, I see, CSV == text. Yep, we should update that too, here's the ticket #38217

[thisisnic]

Interested in submitting a PR at all @csgillespie?