use ICU to define sorting

NEWS.md

@@ -1,3 +1,7 @@
+# git2rdata 0.3.1
+
+* Use `icuSetCollate()` to define a standardised sorting.
+
 # git2rdata 0.3.0
 
 ## New features

R/datahash.R

@@ -50,22 +50,18 @@ datahash <- function(file) {
 #' @noRd
 #' @return a named vector with the old locale
 set_c_locale <- function() {
-  old_ctype <- Sys.getlocale(category = "LC_CTYPE")
-  old_collate <- Sys.getlocale(category = "LC_COLLATE")
-  old_time <- Sys.getlocale(category = "LC_TIME")
-  Sys.setlocale(category = "LC_CTYPE", locale = "C")
-  Sys.setlocale(category = "LC_COLLATE", locale = "C")
-  Sys.setlocale(category = "LC_TIME", locale = "C")
-  return(c(ctype = old_ctype, collate = old_collate, time = old_time))
+  icuSetCollate(
+    locale = "en_GB", case_first = "lower", normalization = "on",
+    case_level = "on"
+  )
+  return(c())
 }
 
 #' Reset the old locale
 #' @param locale the output of `set_c_locale()`
 #' @return invisible `NULL`
 #' @noRd
 set_local_locale <- function(locale) {
-  Sys.setlocale(category = "LC_CTYPE", locale = locale["ctype"])
-  Sys.setlocale(category = "LC_COLLATE", locale = locale["collate"])
-  Sys.setlocale(category = "LC_TIME", locale = locale["time"])
+  icuSetCollate(locale = "default")
   return(invisible(NULL))
 }

tests/testthat/test_b_special.R

@@ -19,7 +19,7 @@ expect_is(
 )
 expect_equal(
   names(output)[1],
-  "9e5edf55ceadd2c148d6d715ea5d12cc8e1538d8"
+  "1d135a85dc9beff3223d6c79f0d8975b559afca7"
 )
 old_locale <- git2rdata:::set_c_locale()
 dso <- ds[order(ds$a), , drop = FALSE] # nolint
@@ -64,7 +64,7 @@ expect_equal(
 )
 expect_equal(
   names(output)[1],
-  "9e5edf55ceadd2c148d6d715ea5d12cc8e1538d8"
+  "1d135a85dc9beff3223d6c79f0d8975b559afca7"
 )
 expect_identical(
   names(output),