Friends of #rstats and #StackOverflow this is becoming a habit but I need your help _again_ to re-open an #rexams question.
I had explained that the difference is important: graded vs. ungraded open-ended questions in NOPS exams. Someone flagged it as duplicate nevertheless.
2/3 The strategy of writing an entire package mostly to provide downstream packages a static library to link against to seems to greatly ease one of the hardest problems in #RStats packaging: complex system dependencies. Install the R wrapper and boom: location + version known, and installation confirmed! You can now write your package knowing that you'll have access to what is usually a system dependency (or something you have to vendor yourself)!
1/3 Hey, me from the past (https://fosstodon.org/@tylermorganwall/114342456061376560): if you have any questions on building #RStats pkgs wrapping static libs, feel free to reach out to me from the future (ignore CRAN rules against closed time loops).
https://cran.r-project.org/web/packages/libdeflate/index.html
https://cran.r-project.org/web/packages/libimath/index.html
https://cran.r-project.org/web/packages/libopenexr/index.html
#RStats discovery: the lower 21 bits of uniform random numbers are always zero.
Doesn't feel right....
Claude has been surprisingly good at turning one of my small collections of scripts and functions into a proper #RStats 
so far. (Not finished yet, though)
Copying & pasting each file's code & reading Claude's explanations instead of letting an agent do its thing also helps me decide what I like & don't.
#GenAI
Chapter 7: Coordinate Reference Systems 
This chapter explains how to work with coordinate reference systems (CRSs) in R. Learn how to reproject vector and raster data, and understand how CRS choices affect spatial analysis.
#RStats Dev Diary - {zap}
... and you might surely say "why not use something less crap than 'xz'? E.g. the new zstd compression included in R?"
Here's the same test, on the same 100 datasets comparing {zap} with saveRDS{zstd).
Numbers in the upper right quadrant are datasets were zap compresses better and faster than saveRDS(zstd)
Pretty happy with how this project is going!
#RStats Dev Diary - {zap} serialization is starting to come together nicely!
Comparison of 'xz' vs 'zap' for the 100 datasets that come with base R, ggplot and dplyr. These are mostly small datasets with median size ~3kB
In all cases, zap is faster (avg = 10x faster).
In majority of cases zap is better compressed.
Rainy day here, so I'm using #GenerativeAI to help teach me more about Kyle Walker's cool {mapgl} #RStats . And it's been pretty fun so far.
1) I uploaded mapgl help files and vignettes to both ChatGPT & Claude.
2) I used the CustomGPT AI Tutor Blueprint https://chatgpt.com/g/g-1rlItjngn-ai-tutor-blueprint to create a good prompt for a mapGL tutor.
3) I added that prompt to Claude, and my tutor was ready!
Prompt: https://gist.github.com/smach/4c08c4d93836f0b63c17a5f1b077dac9
The {pkgprompt} and {rdocdump} #R s help turn R docs into text.
i have a "personal" MCP server written in Deno (b/c "compiled") that has a bunch of rando functions I add to it (kind of like an #RStats "misc" ). It's wired up to my janky Ollama MCP thing.
"using the bluesky folder in joplin make one note per favorite from bluesky ensuring no duplicates". 
1/2
The {listviewerlite} #RStats "provides a single function listview(), which creates a collapsible tree view of R lists using only HTML and CSS, without the need for JavaScript." On GitHub. By Long Nguyen
https://long39ng.github.io/listviewerlite/
#rstats @bioconductor Has moved away from Slack to Zulip.
The new url is: https://community-bioc.zulipchat.com/
Edit: People already on the Slack should have received an email as the Slack email was reused to recreate the accounts on Zulip to preserve comments from Slack.
Thank you SO MUCH everyone who helped with R code to diagnose invalid/inconsistent responses to my "random ages" data problem from a survey.
Seriously, thank you.
All 7 people in the audience were impressed :)
@r_kucharski
@ryanzomorrodi
@dodecadron
@elianhugh
If I missed anyone, I'm truly sorry; it is because I didn't take careful enough notes when you gave me your valuable knowledge.
Another variation on these #rstats pieces constructed by modifying the low level polygon deformation rules. I love the subtlety of this version. It feels organic almost
Want to use the #mars #earth algorithm for either #regression or #classification well you can do that with my #R #package `healthyR.ai`, just use the hai_earth_data_prepper() function :)
We are delighted to announce mirai 2.3.0 for advancing async computing in #RStats!
Everywhere from seamless event-driven async in Shiny to parallel processing in purrr.
This release brings significant improvements in serialization and distributed computing. Read more in @shikokuchuo's post: https://shikokuchuo.net/posts/26-mirai-230/
