Sort default_undesirable docs robustly

[MichaelChirico]

Have been getting some spurious diffs vs. main on this. Is anyone else seeing it? I'm a bit surprised because the default* objects are coming from modify_defaults(), which uses platform_independent_order() already:

lintr/R/with.R

Line 59 in fdbb9bf

res <- res[platform_independent_order(names(res))]

Maybe this is a call to improve platform_independent_order()? That's pretty rudimentary indeed:

lintr/R/utils.R

Lines 206 to 207 in fdbb9bf

	platform_independent_order <- function(x) order(tolower(gsub("_", "0", x, fixed = TRUE)))
	platform_independent_sort <- function(x) x[platform_independent_order(x)]

[IndrajeetPatil]

Is anyone else seeing it?

Yes, indeed, I was seeing these diffs in this .Rd file while working on some other PRs.

[AshesITR]

Me too. I manually roll those changes back before committing. Similarly for some snapshot tests (Windows recreates them with CRLF endings on at least one of my machines)

[MichaelChirico]

OK, switched to using LC_COLLATE for full robustness.

Per r-lib/withr#179, it's also possible to see some interference with LC_COLLATE envvar, but not adding that complication yet since this is only needed for dev.

[codecov-commenter]

Codecov Report

Attention: 1 lines in your changes are missing coverage. Please review.

Comparison is base (4b59aac) 98.53% compared to head (2352a9b) 98.53%.

❗ Current head 2352a9b differs from pull request most recent head f2062ea. Consider uploading reports for the commit f2062ea to get more accurate results

Files	Patch %	Lines
R/zzz.R	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2455   +/-   ##
=======================================
  Coverage   98.53%   98.53%           
=======================================
  Files         126      126           
  Lines        5676     5676           
=======================================
  Hits         5593     5593           
  Misses         83       83           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[Bisaloo]

In case that's useful, method = "radix" will always use C locale for character strings. It can be a convenient way to get locale-independent sorting without directly tweaking the LOCALE settings

[IndrajeetPatil]

@Bisaloo This is great!

With the default sorting method

callr::r(
  function() withr::with_collate("C", sort(c("a", "_a", "A"))),
  env = c(LC_COLLATE = "C")
)
#> [1] "A"  "_a" "a"

callr::r(
  function() withr::with_collate("en_US", sort(c("a", "_a", "A"))),
  env = c(LC_COLLATE = "en_US")
)
#> [1] "_a" "a"  "A"

With the radix sorting method

callr::r(
  function() withr::with_collate("C", sort(c("a", "_a", "A"), method = "radix")),
  env = c(LC_COLLATE = "C")
)
#> [1] "A"  "_a" "a"

callr::r(
  function() withr::with_collate("en_US", sort(c("a", "_a", "A"), method = "radix")),
  env = c(LC_COLLATE = "en_US")
)
#> [1] "A"  "_a" "a"

^{Created on 2023-12-16 with reprex v2.0.2}

[MichaelChirico]

In case that's useful, method = "radix" will always use C locale for character strings. It can be a convenient way to get locale-independent sorting without directly tweaking the LOCALE settings

awesome idea! thank you {data.table} 😅