Vidar @vidaringa.bsky.social · 29/09/2025So I guess we're all waiting right now for @simonpcouch.com to run his test using sonnet 4.5 #rstats And maybe gpt-5-codex? :) 030
Vidar @vidaringa.bsky.social · 17/07/2025Trying out Positron from @posit.co. I think I can get used to it. But I will miss Rstudio. Autocomplete might be more like Rstudio, i.e. typing ggp (then hit tab) will result in ggplot2:: instead of ggplot() 020
Vidar @vidaringa.bsky.social · 27/05/2025Interesting. I’ve used the shiny assistant. Really like it. I’m still only using the chat interface in claude. Need to try github copilot. 110
Vidar @vidaringa.bsky.social · 27/05/2025I agree. Claude for R and chatgpt for almost everything else :) 000
Vidar @vidaringa.bsky.social · 27/05/2025I do see value in Claude, I've use sonnet 3.7 a lot and now 4.0. I was quite surprised that I got ~800 lines R shiny app that worked from first try after just one (quite detailed) prompt. @simonpcouch.com has done some evaluation: www.simonpcouch.com/blog/ I hope we get some replies from R people 120
Vidar @vidaringa.bsky.social · 27/05/2025I agree with Hadley — I think Claude is better with R than ChatGPT in general. I haven’t spent enough time with Gemini 2.5 to judge. What do you think, #rstats? What’s your preferred LLM for R code? 472
Vidar @vidaringa.bsky.social · 25/05/2025So what is your current workflow. Do you use llm for any of the programming for shiny? 000
Vidar @vidaringa.bsky.social · 25/05/2025Those creating R shiny apps, how important do you think it is to know shiny well? I’m reading Mastering Shiny right now as I want to understand how this works. But at the same time, a good prompt gives me a complete code for a relatively complex shiny app and it worked right away. #rstats 560