List/place for all tidyeval resources?

tidyeval

#1

Hi Everyone!

Super stoked to see this community group, it looks like a super helpful support, thanks for making it. :slight_smile:

Over the past few months since rlang has come out, and since tidyeval has been rolled out amongst the tidyverse R packages, in addition to the (really nice!) documentation for rlang and programming with dplyr, there have been some really nice blog posts that provide some nice examples of using tidyeval. For example, Edwin Thoen’s write up, and then Mara Averick’s really nice round up of tidyeval resources

I was considering adding a special tidyeval resource section to the resources part of my website to collect these, partly for my own resources, but also for others. But I wonder if perhaps it would be best if there was a vetted place for all things tidyeval to live?

Cheers,

Nick


#2

Hi! I agree with that. I had the project of trying to put all resources I encounter about tidyverse packages in one place with links. Like Mara’s round up but not just for tidyeval. Still a project though…

A resource section or place somewhere could allow to capitalize on all the posts about tidyverse that are great examples to discover tidyverse :package: and improve their usage.I find myself to search frequently in my own resource stash (list of blog post save in different place)

Seems a place and a discussion to put this idea. Title is about tidyeval but close enough.


#3

Maybe we could add a section under https://www.tidyverse.org/learn/ ? My main concern is that user contributed content of this nature requires constant maintenance as links break or things go out of date. But maybe this is something the tidyverse developer advocate could work on with the community (once we’ve hire them).

Part of the problem with tidy eval is that we don’t currently have any good central resources because we’re still figuring out the best way to teach it. I’m about to start writing a chapter on it for the 2nd edition of Advanced R which will hopefully fill in some of the gap. You can follow along at https://adv-r.hadley.nz/tidy-eval, but it’ll probably be a week or two before there’s anything terribly useful there.


#4

Yes, so I really like this idea, and, @njtierney, I think a resources section is a great idea. I’m new to/bad at blogging, and that post in particular is really not a “blog post” (I’ve added three or four items to it since first publishing, and :bulb:, this isn’t really what a blog is for-- I had my medium and content all wrong)!

The format of that post (literally just a collection of tweets) was born of laziness/selfishness (I screenshot for tweets bc then I can recall and retrieve the content more easily). So, I’m happy to help in whatever way.

Summary: that post is a :-1: format, the posts in it are :+1:- let’s do it differently!


#5

This specific problem is something I really want to work on. I love when @jennybryan annotates sunset sections of the STAT545 site, and have a few ideas (but nothing great :confused:) re a way of indicating “freshness” of packages and info.

I did a not-so-great job of asking about this on Twitter last week, but what I was trying to illustrate was the general idea of somehow noting “not new, but still good.”

Better res. images here: https://imgur.com/a/kRGhG


Highlighting bookdown.org book updates? 📚
#6

You are good at blogging and also good at the internet more generally. Don’t sell yourself short!


#7

I’m not terribly familiar with Discourse, having mostly just used it here. That said, it appears that people with a high enough “trust level” (3) can create wiki-style posts that are editable by all standard users. That would be one option to workshop this kind of list in the short term.

That said, assuming that RStudio is using the standard requirements for becoming that trust level, only RStudio employees will have that ability for the next 49 days or so (minimum). So, one of them would have to get the ball rolling.


#8

Thank you :blush:, what I should have said was something closer to “I haven’t quite gotten used to the medium of blogging, yet…” which, in this case, resulted in my updating something that people assume is static. :wink:


#9

I think with the help of community and a specific tooling, a section with links toward blog post can be maintained up to date, all the more is RStudio team member are involved.

With all the work you promote toward learning experience, blogging, sharing, and making website, it has developed awesome content on the web about use-case with tidyverse :package:. Sometime I feel that some blogposts are as useful as a good vignette and it is very pleasant to see some usage of a :package: when you want to learn one. Moreover, there are some content which deals with kind of hidden functionalities or just unknown functions that are not all explained in the :package: documentation. For example, purrr is a real toolbox where map is well known, sometime map_dfr or map_dfc not so much and possibly or safely less.

My idea is to try to make easy to find example content with tidyverse, with the ability for search by tidyverse packages. There already blog aggregators but they are sometimes to wide in topics and do not point enough to their source. Above all, I find it hard to search for a good example in a blogpost on a package. Currently, I often search into my saved list (with pocket, or likes on twitter) and for the tidyverse, I even google something adding @mara twitter handle (@dataandme) because I know the content will be blog post about #rstats. Recently, I tried to follow her step and retweet some content but focusing on saying which package it is a good example for. like this:


I use name of package and :package: emoji to be able to find them more easily later with twitter API. I find it more useful if we can add somewhere the links with title, short description and tags. Not already found time to begin but still in my head.


#10

This could be an interesting thing to try to automate and/or make a “plugin” for that works well with blogdown. Assuming the posts include code, this could just be a matter of looking at what library(foo) :package:s are in there (since sessionInfo will obviously include utils, etc. that aren’t as relevant). @milesmcbain’s deplearning package might also be a good starting point for this.

:confused: downside to any of this is, of course, that it would depend on people being consistent (which is never a reliable mechanism), but food for thought nonetheless…


#11

I had in mind something less automated, let’s say semi-automated.

Of course using blogdown and with a way to automat publishing from few elements as a link short description and tags. However, the choosing of the link and the classification by packages would be check manually.

I think it kind of mimic what could be done on Twitter : choose a link, wrote a small description with the link, tweet. Unless your Workflow on Twitter if fully automated already.

I think it is a manageable way to deal with consistency variation from different post. What do you think? Too manual Workflow?
Should we give a try @mara?


#12

Oh, my workflow isn’t even mildly-efficient, it’s about as far from automated as one can get :joy:.

I’m game to give it a whirl (I’ve been trying out pulling together some of my tweets by :package: in blog format). For now, I’ve been trying to tag packages in tweets, but even if I don’t include them in those, it’s easy enough to note for things I’m reading anyhow! :+1: