Michael Chirico

Hey #rstats #rdatatable if you're a {data.table} user please take a look at the new community survey:

https://tinyurl.com/datatable-survey

Great things getting started over at {data.table}...

Long overdue update of potools on CRAN! Tools for translating R packages.

Starring some wonderful contributions from @hadleywickham@fosstodon.org who graciously suggested+implemented a more modular workflow making distinct translation roles easier to understand + work with.

Workhorses are the new functions po_extract(), po_create() po_compile(), and po_update(). See the two new vignettes (https://cran.r-project.org/web/packages/potools/vignettes/developers.html, https://cran.r-project.org/web/packages/potools/vignettes/translators.html).

Thanks also to @maelle@mastodon.social for trying it out!

https://cran.r-project.org/web/packages/potools/index.html

Dedicated website for #rstats translations is looking slick 😀

https://contributor.r-project.org/translations/

A very funny joke the #rstats architects pulled off is to have an SEXPTYPE called LANGSXP, an R function called is.language(), and for them not to be the same thing

Some notes about reshape::rename() vs. dplyr::rename() for anyone refactoring #rstats Ancient Code:

- The API is reversed. reshape does c(old = "new"), dplyr works like c(new = "old") or 'new = old', a la mutate()
- dplyr errors by default for unfound names, so a robust implementation needs dplyr::rename(x, any_of(new = old)) if 'old' might not exist on input
- dplyr errors instead of creating duplicate names. I don't see an easy way around this, but likely mutate() was intended anyway.

TFW you removed an innocuous importFrom(pkg, foo) but the implied loadNamespace("pkg") was actually letting 'pkg' define its #rstats S4 generic on top of base::serialize() so now your serialize() calls change implicitly from pkg::serialize() to base::serialize() 😵‍💫😵‍💫😵‍💫

@maelle@mastodon.social @Lluis_Revilla@fosstodon.org @r_data_table@fosstodon.org yep... and there's likely to be funding for a new push to other major languages e.g. Spanish. Thanks to the translation team!!

Ah, fun. encodeString() and sprintf() can be faster than paste0() for '-wrapping your strings... if the inputs are huge.

@MikeMahoney218@fosstodon.org we want all(c(X, Y)) == all(X) && all(Y), the behavior of all(NULL) is required here for the empty edge case X=logical(0). Maybe the surprising part is NULL getting cast to as.logical(NULL) = logical(0).

@vincentab@fosstodon.org it's active till December 1!!

@sckottie@fosstodon.org

@dgkf@fosstodon.org @eddelbuettel@mastodon.social developer.r-project.org is the closest I can think of

@nicoj@mastodon.online @pglpm@emacs.ch I just call it the coalesce operator

maintenance update 0.3.3 of #rstats {geohashTools} is on CRAN

Main purpose is to remove references to {rgdal} now that it's off CRAN (it was previously moved to Suggests). Thanks Roger Bivand for your many years of service!

https://cran.r-project.org/web/packages/geohashTools/index.html

Could I request a fosstodon server invite from someone? Seems this here econtwitter server is due to sunset 🥲