blog about C API

[tdhock]

@aitap since you are familiar with the C API issue Rdatatable/data.table#6180
I was wondering if you may be interested to write a blog for the Raft about the current state of the R C API and how data.table uses it? If so, could you please write a general / gentle introduction, so that even people who are not C API experts (like me) could understand.
Target readers would be any current/future data.table developers, as well as other R package developers using the C API.

[aitap]

Thank you for the opportunity! I will do my best. We might need a few editing rounds. Is there a target word count? How much of the focus should be on the "non-API" entry points LEVELS, SETLENGTH, SET_GROWABLE_BIT, SET_TRUELENGTH, STRING_PTR, TRUELENGTH?

[tdhock]

I'm not sure you need to worry about a target word count. Please take as much space as you need, in order to explain how these different non-API entry points are used, and how/why. If possible please explain which ones can't be easily changed without sacrificing some kind of efficiency.
My recent duckdb vs polars comparison is probably too long https://rdatatable-community.github.io/The-Raft/posts/2024-10-17-duckdb_polars_reshape-toby_hocking/

[aitap]

Does this skeleton roughly match what you had in mind for the post? Would you like me to also cover the functions data.table has already got rid of (isFrame, [UN]SET_S4_OBJECT, SET_TYPEOF, NAMED)?

[tdhock]

looks great.
I like the historical introduction. I think it would be worth adding links to the R-devel thread which started the C API declarations earlier this year, which I believe is https://stat.ethz.ch/pipermail/r-devel/2024-April/083349.html
yes I think it would be worth discussing the functions data.table has already got rid of, with a brief explanation about how/why they were easy/quick to do.

[tdhock]

done thanks!

[tdhock]

this is highlighted on Rweekly