search

r-lib / lintr

Static Code Analysis for R
https://lintr.r-lib.org
Other
1.19k stars 184 forks source link

Should we lint code in `glue::glue()` expressions? #968

Open AshesITR opened 2 years ago

AshesITR commented 2 years ago

Inspired by #942. Should be possible to do IINM, but not sure how to best report lints found within glue expressions. stringr::str_interp() is another interpolation function with similar semantics.

To obtain a list of lints for a particular glue expression, we could lint(..., text = paste(glue_exprs, collapse = "\n")) and then translate the lints obtained to location info in the original file. object_usage_linter() would need special care not to report undefined variables which are defined in an outer scope, i.e. outside of the glue call.

MichaelChirico commented 2 years ago

FWIW I think we should ignore str_interp(), since it's marked as superseded:

https://stringr.tidyverse.org/reference/str_interp.html

Its successor, str_glue() should be treated same as glue() since it's just a wrapper:

stringr::str_glue
function (..., .sep = "", .envir = parent.frame()) 
{
    glue::glue(..., .sep = .sep, .envir = .envir)
}
eitsupi commented 2 years ago

Since it is mentioned in the README of glue about stringr::str_glue(), it would be great if it is treated in the same way as glue::glue(). https://github.com/tidyverse/glue/blob/d47d6c7701f738f8647b951df418cfd5e6a7bf76/README.md?plain=1#L50-L52