This week in R (2024-10-11)

Happy Friday everyone! Here's some news from the world of R this week:

• A Shiny new Assistant:?Posit have released Shiny Assistant, an instance of the Claude 3.5 Sonnet LLM which is specialised in Shiny app development in both R and Python. Read the announcement here and play around with assistant here.
• Funky file systems: Danielle Navarro has written a very entertaining blog post about {fs}and why it is an underappreciated package in R data science workflows. If you are using file paths at all or developing and deploying on different OS', you should definitely read it!
• Duck Duck loose: Stephen Turner has written a blog post comparing DuckDB to base R and {readr}/{dplyr}, and found DuckDB to be 125x faster than base R! The post includes some information about {duckplyr}, the package for interacting with DuckDB using {dplyr}-like syntax.
• Tinker Tailor: The tidymodels team have announced the latest addition to their ecosystem - {tailor}. This new package "is to postprocessing as {recipes} is to preprocessing" and is something that has been highly anticipated. The package is not yet on CRAN but you can check out the Github here.
• Shiny in Production:?I attended another excellent conference in Newcastle this week, congrats to Jumping Rivers and all the sponsors and speakers for putting on a great event. Some personal favourites were: ?? Tan Ho 's reflections on a large-scale shiny app project; Gareth Burns ' journey using shiny as an interactive education tool; Alasdair Morgan 's lightning talk on correcting resource usage on Posit Workbench, and Katy Morgan's discussion on integrating Gen AI into an auditing shiny app. Talks were recorded and next year's instalment of the conference has already been confirmed so keep an eye out here for announcements.
• Fun fact:??did you know, there were 1203 downloads of R versions 3.x.x or even 2.x.x from the RStudio CRAN mirror in the last 30 days?! Thankfully that only accounts for less than 0.2% of all R downloads in that time period.

r_downloads <- cranlogs::cran_downloads("R", when = "last-month")

old_old_versions <- r_downloads |>
  dplyr::filter(stringr::str_detect(version, "^[32]"))

sum(old_old_versions$count)
#> [1] 1203

sum(old_old_versions$count) / sum(r_downloads$count)
#> [1] 0.001489366        

New versions:

• {data.table} v1.16.2 the latest release of the super fast table package includes various bug fixes.
• {mirai} v1.3.0 - the exciting async package adds some new features for working with daemons and other bug fixes.

I post updates like this every week so if you're interested feel free to follow. Comment below if there's something interesting you found out this week too!