Restyle expression-derived source with styler

[jennybc]

Expression input generally gets processed with formatR::tidy_source() due to the addition of this line in the .R source file:

knitr::opts_chunk$set(tidy = TRUE, tidy.opts = list(indent = 2))

But this does not play well with the bang-bang syntactic sugar from tidyeval:

formatR::tidy_source(
  text = c(
    "nameshift <- c(SL = 'Sepal.Length')",
    "head(dplyr::rename(iris[,1:2], !!!nameshift), 3)"
  )
)
#> nameshift <- c(SL = "Sepal.Length")
#> head(dplyr::rename(iris[, 1:2], !(!(!nameshift))), 3)

styler is under development and its non-invasive quality works to our advantage:

styler::style_text(
  c(
    "nameshift <- c(SL = 'Sepal.Length')",
    "head(dplyr::rename(iris[,1:2], !!!nameshift), 3)"
  )
)
#> [1] "nameshift <- c(SL = \"Sepal.Length\")"            
#> [2] "head(dplyr::rename(iris[, 1:2], !!!nameshift), 3)"

Switch to that once it's on CRAN.

cc @krlmlr @lorenzwalthert

[lorenzwalthert]

Glad to see it works for you. We haven't payed particular attention to tidyeval / rlang features yet, so let us know if you find something missing / wrongly formatted. A CRAN submission is planned for late August.

[jennybc]

How firm is your timeline @lorenzwalthert? If it's pretty good, I could switch over now and have dev reprex depend on dev styler, which might allow a few of us to accumulate a bit of reprex + styler + tidyeval experience w/o even trying.

[krlmlr]

We should release to CRAN as part of the GSoC project, we have no reason to expect major delays. It would be awesome to give styler a bit of exposure before that, but I remember a few minor problems we wanted to fix before announcing a stable-ish version. @lorenzwalthert: Can you please comment?

[lorenzwalthert]

Yes, @krlmlr is right. We fixed all remaining (minor) problems now except for token-dependent indention (i.e. indention that is not just the regular 2, 4, etc. spaces but depends on the length of other tokens). That means styler will format

a <- function(x,
              y) {
}

and the like as

a <- function(x,
  y) {
}

However we should also be able to fix that (see PR #104) soon, so we will submit to CRAN by the end of August.

[jennybc]

So we can just start using it here with no fanfare, knowing we won't have a non-CRAN dependency for long. I will leave any active publicity to you two.

[lorenzwalthert]

Ok, great.

[krlmlr]

I think we could give styler a try now, it seems to work well and we're about to submit to CRAN. Should we use styler exclusively, or keep formatR as an option?

[jennybc]

I lean towards just switching over to styler.

I am trying to get googledrive submitted to CRAN and can do some maintenance here right after.

[jennybc]

Re-styling is no longer mandatory for expression input, so closing this.

Re-styling with styler will, instead, be made available for all input modes: #108.