style with styler

[lorenzwalthert]

Styling is done with a version of styler containing both r-lib/styler#150 and r-lib/styler#149.

According to the vignette programming with dplyr, there is no space around bang-bang operators.

[lorenzwalthert]

Reference: r-lib/styler#93.

[krlmlr]

@hadley @lionel-: Should we add the space after the bang-bangs to the tidyverse style guide?

[lorenzwalthert]

I opened tidyverse/style#38 for that.

[lionel-]

Personally I add a space but I think other people don't.

[krlmlr]

Can we keep the which on the same line with the switch( here? This is one of the few special cases.

[lorenzwalthert]

We could but it would involve creating a text_before column in our parse data and it is not covered by the tidyverse style guide. In addition, I was not sure which of the below versions would be correct:

switch(cond,
  x, 
  y
)

or

switch(cond,
       x, 
       y
)

[krlmlr]

I thought the switch and the arguments are part of the same nest (=sub-tibble of the nested parse data)?

I'm fine with

switch(cond,
  x, 
  y
)
ifelse(cond,
  true_part,
  false_part
)

A simpler solution that seems enough for now: The line break after the opening paren is added only if strict = TRUE. I'm not sure what the current setup is.

[lionel-]

It's the former. The idea is that it is a special case of this style:

list(
  a,
  b
)

where the main arguments go onto the first line. There could be one, two or three of them. A rough guide is that anything before ... is on the first line, but that does not need to be always the case.

[krlmlr]

Thanks @lionel- for chiming in. I like the idea that anything before the triple dots may stay after the opening paren, we should be able to cover most cases (apart from ifelse()). Unfortunately, doing this accurately also seems to require knowledge about the callee's argument list, which we don't have yet.

Let's simply avoid adding a line break for strict = FALSE .

[lorenzwalthert]

No, the function call itself not.

test <- c("switch(abc,", "xy,z)")
styler:::create_tree(test)
#>                                              levelName
#> 1  ROOT (token: short_text [lag_newlines/spaces] {id})
#> 2   °--expr:  [0/0] {20}                              
#> 3       ¦--expr:  [0/0] {3}                           
#> 4       ¦   °--SYMBOL_FUNCTION_CALL: switc [0/0] {1}  
#> 5       ¦--'(': ( [0/0] {2}                           
#> 6       ¦--expr:  [0/0] {6}                           
#> 7       ¦   °--SYMBOL: abc [0/0] {4}                  
#> 8       ¦--',': , [0/0] {5}                           
#> 9       ¦--expr:  [1/0] {12}                          
#> 10      ¦   °--SYMBOL: xy [0/0] {10}                  
#> 11      ¦--',': , [0/0] {11}                          
#> 12      ¦--expr:  [0/0] {17}                          
#> 13      ¦   °--SYMBOL: z [0/0] {15}                   
#> 14      °--')': ) [0/0] {16}

But anyway, we can go into the first child and check text if pd$token_before[2] is a "SYMBOL_FUNCTION_CALL", so we don't need text_before. I think we do something similar in other cases.

[lorenzwalthert]

This was for testing purposes only.