Wrapping at infix operator in long if statement chain

[gadenbuie]

Just bumped into this in the wild (i.e. code I didn't write in chromote). I expected no change to the following code,

if (
  domain %in% private$session$get_auto_events() &&
    private$event_callback_counts[[domain]] == 1 &&
    isTRUE(private$event_enable_domains[[domain]])
) {
  private$session$debug_log("Enabling events for ", domain)
  private$session[[domain]]$enable()
}

but air format introduces a line break after the %in%:

if (
  domain %in%
    private$session$get_auto_events() &&
    private$event_callback_counts[[domain]] == 1 &&
    isTRUE(private$event_enable_domains[[domain]])
) {
  private$session$debug_log("Enabling events for ", domain)
  private$session[[domain]]$enable()
}

It's not a new change, this repros back to air 0.1.1.

[gadenbuie]

Smaller reprex:

# before
if (x %in% letters && this_needs_to_be_long_enough && this_also_needs_to_be_long_enough) {
  'result'
}

# after
if (
  x %in%
    letters &&
    this_needs_to_be_long_enough &&
    this_also_needs_to_be_long_enough
) {
  'result'
}

[lionel-]

hmm I think this is to be expected?

[gadenbuie]

This seems inconsistent

# before
if (x %in% letters && y %in% some_other_collection_of_letters && x == who_knows_what) {
  'result'
}

# after
if (
  x %in%
    letters &&
    y %in% some_other_collection_of_letters &&
    x == who_knows_what
) {
  'result'
}

[kpagacz]

Out of the box, tergo formats this the way you want it:

if (
  domain %in% private$session$get_auto_events() &&
    private$event_callback_counts[[domain]] == 1 &&
    isTRUE(private$event_enable_domains[[domain]])
) {
  private$session$debug_log("Enabling events for ", domain)
  private$session[[domain]]$enable()
}

and the smaller reprex:

# before
if (
  x %in% letters &&
    this_needs_to_be_long_enough &&
    this_also_needs_to_be_long_enough
) {
  "result"
}

# after
if (
  x %in% letters &&
    this_needs_to_be_long_enough &&
    this_also_needs_to_be_long_enough
) {
  "result"
}

Check out tergo (it also has an R package) if you haven't already! (I am the author!)

[jayhesselberth]

+1 for no break after infix operators.

The break makes the lhs and rhs look like operands, esp. mixed with && or ||.

[DavisVaughan]

@kpagacz tergo formats this similar to air with line_length = 80.

# before
if (x %in% letters && this_needs_to_be_long_enough && this_also_needs_to_be_long_enough) {
  'result'
}

# after (tergo)
if (
  x %in%
    letters &&
    this_needs_to_be_long_enough &&
    this_also_needs_to_be_long_enough
) {
  'result'
}

# after (air)
if (
  x %in%
    letters &&
    this_needs_to_be_long_enough &&
    this_also_needs_to_be_long_enough
) {
  'result'
}

# before
if (x %in% letters && y %in% some_other_collection_of_letters && x == who_knows_what) {
  'result'
}

# after (tergo)
if (
  x %in%
    letters &&
    y %in%
      some_other_collection_of_letters &&
    x ==
      who_knows_what
) {
  'result'
}

# after (air)
if (
  x %in%
    letters &&
    y %in% some_other_collection_of_letters &&
    x == who_knows_what
) {
  'result'
}

# With `line_length = 80`
tergo ~/Desktop/test2.R ~/Desktop/tergo.toml

[DavisVaughan]

@jayhesselberth %>% is an infix operator similar to |>. %||% is an infix operator similar to ||. %*% is an infix operator similar to *. I'd argue those should chain in the same way as other non-infix operators (certainly for the pipe). %in% is the one that feels weird here, I think.

[jayhesselberth]

Yeah, maybe %in% just needs special treatment.

[kylebutts]

I have other examples of this.

Formatted:

tibble(
  tmdb_production_company_name = res$production_companies[[1]]$name,
  tmdb_production_company_country = res$production_companies[[
    1
  ]]$origin_country
)

new_var_with_very_long_name_to_cause_overflow <- (1.2382 +
  20.3402 +
  12120.3402 * 2) /
  sqrt(20.3920^2 + 5.1292^2)

My preference would be for it to look something like

tibble(
  tmdb_production_company_name = res$production_companies[[1]]$name,
  tmdb_production_company_country = 
    res$production_companies[[1]]$origin_country
)

new_var_with_very_long_name_to_cause_overflow <- 
  (1.2382 + 20.3402 + 12120.3402 * 2) /
  sqrt(20.3920^2 + 5.1292^2)

I think a general rule would be to see if you could fit in a new line, to not fully expand?