tidyeval

[lorenzwalthert]

• I was wondering how we aim to replace standard evaluation functions like select_(~xyz) with the upcoming release of dplyr 0.6.0 and associated. Is it via rlang .data$xzy, i.e. select(.data$xyz)? I think we should aim for consitency and change it in a single commit or PR.
• In addition, I think we need to consider that not all packages from the tidyverse support tidyeval yet, e.g. see this PR for tidyr. I could not find out when that will be the case. Hence, the question is when the ideal point of transition is.

@krlmlr maybe you can advise on that.

[krlmlr]

The select_() wrappers will remain compatible for some time, but we should definitely aim for replacement once dplyr 0.6.0 hits CRAN. We can use select(!!"xyz"), this is both safe and compatible to R CMD check.

[krlmlr]

As for tidyr, we can upgrade the tidyr calls one it's been updated to tidyeval, too.

[lorenzwalthert]

Ok, thanks. I guess it would be nice to have something like the roxygen2md package since a lot of code (in general) needs to be converted eventually but if you don't know of efforts towards that I will just do it manually, for styler it's not gonna be that much as you know.

[lorenzwalthert]

As tidyr seems to be ported to the tidyverse i.e this PR is closed, should we change our function calls to select(!!"xyz") or select(.data$xyz)? I think in the dplyr vignette, they advise .data$xyz, not sure why not !!"xyz"...

[lorenzwalthert]

Well anyways we probably want to wait until it is released on CRAN anyways...

[krlmlr]

tidyr with tidyeval should be on CRAN by now, let's take another look.

[lorenzwalthert]

Well we've not used tidyr directly because of the dplyr dependency. They were ported from tidyr to the file compat-tidyr. So not sure how to proceed.