Integrated reprex reminders/suggestions

Early days, but is there perhaps a way to do reprex suggestion/reminders for users' first few screenshots of code. Obviously, these could be ignored, and perhaps the conditions for triggering said suggestion could be refined, but it might be helpful to see it (at the very least) mentioned in context.

5 Likes

Another option (in combination, or on its own if your suggestion is hard to implement in Discourse) would be pinned reminders in the most likely categories for code-specific questions (#tidyverse, #shiny, the unfortunately named #package-develoment, and possibly #rstudio-ide). I'd prefer your more just-in-time option, though.

1 Like

This is a great idea, and perhaps the reminder could include a link to a walkthrough of how to make a reprex.

1 Like

Yeah, I'd argue that the 20 minutes of @jennybryan on this rOpenSci community call/video (below) could have the highest ROI of a new R-coder's life (or not so new…:flushed:)

But, attention spans being what they are, @njtierney's post is gif-filled and fantastically useful
http://www.njtierney.com/post/2017/01/11/magic-reprex/

I point to these rather than the tidyverse reprex guide, and/or the SO thread just because they're less text-heavy! I think those are great as well.

3 Likes

This is great, thank you so much for sharing @mara.

I endorse all these ideas, but this is a hard problem that SO has different tacks on for years, but still faces nonstop. Given this is RStudio, an approach that may be possible here would be to make code chunks run (with suitable restrictions for security and resources) when rendering Markdown. That way users could not just copy and paste errors, but actually show them, and it would be much easier to automatically flag non-reproducible examples and add an extra dialogue before they could be posted.

Making this happen is clearly difficult—but possible—and if anybody can make it work well, it's RStudio. And obviously it'd be incredibly awesome.

For some reason the "but possible" :link: isn't working… Definitely would be curious to see what you've got in mind! :blush:

Thanks for letting me know, fixed! It's just the DataCamp R-Fiddle, which proves R is runnable on a website.