Merge branch 'main' into canload

components/calendar/src/islamic.rs

@@ -2149,7 +2149,7 @@ mod test {
         }
     }
 
-    #[ignore]
+    #[ignore] // slow
     #[test]
     fn test_days_in_provided_year_observational() {
         let calendar = IslamicObservational::new();
@@ -2179,7 +2179,7 @@ mod test {
         );
     }
 
-    #[ignore]
+    #[ignore] // slow
     #[test]
     fn test_days_in_provided_year_ummalqura() {
         let calendar = IslamicUmmAlQura::new();

components/collator/tests/tests.rs

@@ -622,16 +622,12 @@ fn test_ja_chooon_kigoo() {
     }
 }
 
-// TODO: This test should eventually test fallback
 // TODO: Test Swedish and Chinese also, since they have unusual
 // variant defaults. (But are currently not part of the test data.)
-#[ignore]
 #[test]
 fn test_region_fallback() {
     // There's no explicit fi-FI data.
-    let locale: Locale = "fi-u-co-standard".parse().unwrap();
-
-    // let locale = locale!("fi-FI").into();
+    let locale = locale!("fi-FI");
 
     let collator = Collator::try_new(&locale.into(), CollatorOptions::new()).unwrap();
     assert_eq!(collator.compare("ä", "z"), Ordering::Greater);
@@ -694,8 +690,8 @@ fn test_vi() {
     }
 }
 
-#[ignore]
 #[test]
+// See DatagenProvider test_zh_non_baked for gb2312 and big5han tests
 fn test_zh() {
     // Note: ㄅ is Bopomofo.
 
@@ -735,20 +731,6 @@ fn test_zh() {
         assert_eq!(collator.compare("佰", "ㄅ"), Ordering::Less);
         assert_eq!(collator.compare("不", "把"), Ordering::Greater);
     }
-    {
-        let locale: Locale = "zh-u-co-gb2312".parse().unwrap();
-        let collator = Collator::try_new(&locale.into(), CollatorOptions::new()).unwrap();
-        assert_eq!(collator.compare("艾", "a"), Ordering::Greater);
-        assert_eq!(collator.compare("佰", "a"), Ordering::Greater);
-        assert_eq!(collator.compare("ㄅ", "a"), Ordering::Greater);
-        assert_eq!(collator.compare("ㄅ", "ж"), Ordering::Greater);
-        assert_eq!(collator.compare("艾", "佰"), Ordering::Less);
-        // In GB2312 proper, Bopomofo comes before Han, but the
-        // collation leaves Bopomofo unreordered, so it comes after.
-        assert_eq!(collator.compare("艾", "ㄅ"), Ordering::Less);
-        assert_eq!(collator.compare("佰", "ㄅ"), Ordering::Less);
-        assert_eq!(collator.compare("不", "把"), Ordering::Greater);
-    }
     {
         let locale: Locale = "zh-u-co-stroke".parse().unwrap();
         let collator = Collator::try_new(&locale.into(), CollatorOptions::new()).unwrap();
@@ -785,18 +767,6 @@ fn test_zh() {
         assert_eq!(collator.compare("佰", "ㄅ"), Ordering::Less);
         assert_eq!(collator.compare("不", "把"), Ordering::Less);
     }
-    {
-        let locale: Locale = "zh-u-co-big5han".parse().unwrap();
-        let collator = Collator::try_new(&locale.into(), CollatorOptions::new()).unwrap();
-        assert_eq!(collator.compare("艾", "a"), Ordering::Greater);
-        assert_eq!(collator.compare("佰", "a"), Ordering::Greater);
-        assert_eq!(collator.compare("ㄅ", "a"), Ordering::Greater);
-        assert_eq!(collator.compare("ㄅ", "ж"), Ordering::Less);
-        assert_eq!(collator.compare("艾", "佰"), Ordering::Less);
-        assert_eq!(collator.compare("艾", "ㄅ"), Ordering::Less);
-        assert_eq!(collator.compare("佰", "ㄅ"), Ordering::Less);
-        assert_eq!(collator.compare("不", "把"), Ordering::Less);
-    }
     // TODO: Test script and region aliases
 }
 
@@ -1044,7 +1014,6 @@ fn test_tr_primary() {
     }
 }
 
-#[ignore]
 #[test]
 fn test_lt_tertiary() {
     let left = [
@@ -1079,7 +1048,6 @@ fn test_lt_tertiary() {
     }
 }
 
-#[ignore]
 #[test]
 fn test_lt_primary() {
     let left = ["ž"];

components/datetime/src/helpers.rs

@@ -38,7 +38,7 @@ macro_rules! size_test {
             };
         }
         #[test]
-        #[cfg_attr(not(icu4x_run_size_tests), ignore)]
+        #[cfg_attr(not(icu4x_run_size_tests), ignore)] // Doesn't work on arbitrary Rust versions
         fn $id() {
             let size = core::mem::size_of::<$ty>();
             let success = match option_env!("CI_TOOLCHAIN") {
@@ -67,7 +67,7 @@ macro_rules! size_test {
             };
         }
         #[test]
-        #[cfg_attr(not(icu4x_run_size_tests), ignore)]
+        #[cfg_attr(not(icu4x_run_size_tests), ignore)] // Doesn't work on arbitrary Rust versions
         fn $id() {
             let size = core::mem::size_of::<$ty>();
             let expected = $size;

components/datetime/src/neo.rs

@@ -57,7 +57,7 @@ macro_rules! gen_any_buffer_constructors_with_external_loader {
                 &provider.as_downcasting(),
                 &ExternalLoaderAny(provider),
                 locale,
-                $($arg),+
+                $($arg.into()),+
             )
         }
         #[doc = icu_provider::gen_any_buffer_unstable_docs!(BUFFER, Self::$compiled_fn)]
@@ -74,7 +74,7 @@ macro_rules! gen_any_buffer_constructors_with_external_loader {
                 &provider.as_deserializing(),
                 &ExternalLoaderBuffer(provider),
                 locale,
-                $($arg),+
+                $($arg.into()),+
             )
         }
     };
@@ -394,7 +394,7 @@ impl<C: CldrCalendar, R: TypedDateTimeMarkers<C> + IsRuntimeComponents> TypedNeo
             &crate::provider::Baked,
             &ExternalLoaderCompiledData,
             locale,
-            components,
+            components.into(),
             length,
         )
     }
@@ -440,7 +440,7 @@ impl<C: CldrCalendar, R: TypedDateTimeMarkers<C> + IsRuntimeComponents> TypedNeo
             provider,
             &ExternalLoaderUnstable(provider),
             locale,
-            components,
+            components.into(),
             length,
         )
     }
@@ -451,7 +451,7 @@ impl<C: CldrCalendar, R: TypedDateTimeMarkers<C>> TypedNeoFormatter<C, R> {
         provider: &P,
         loader: &L,
         locale: &DataLocale,
-        components: impl Into<NeoComponents>,
+        components: NeoComponents,
         length: NeoSkeletonLength,
     ) -> Result<Self, LoadError>
     where
@@ -478,7 +478,7 @@ impl<C: CldrCalendar, R: TypedDateTimeMarkers<C>> TypedNeoFormatter<C, R> {
             &R::DateTimePatternV1Marker::bind(provider),
             locale,
             length,
-            components.into(),
+            components,
         )
         .map_err(LoadError::Data)?;
         let mut names = RawDateTimeNames::new_without_fixed_decimal_formatter();
@@ -980,7 +980,7 @@ impl<R: DateTimeMarkers + IsRuntimeComponents> NeoFormatter<R> {
             &crate::provider::Baked,
             &ExternalLoaderCompiledData,
             locale,
-            components,
+            components.into(),
             length,
         )
     }
@@ -1081,7 +1081,7 @@ impl<R: DateTimeMarkers + IsRuntimeComponents> NeoFormatter<R> {
             provider,
             &ExternalLoaderUnstable(provider),
             locale,
-            components,
+            components.into(),
             length,
         )
     }
@@ -1092,7 +1092,7 @@ impl<R: DateTimeMarkers> NeoFormatter<R> {
         provider: &P,
         loader: &L,
         locale: &DataLocale,
-        components: impl Into<NeoComponents>,
+        components: NeoComponents,
         length: NeoSkeletonLength,
     ) -> Result<Self, LoadError>
     where
@@ -1169,7 +1169,7 @@ impl<R: DateTimeMarkers> NeoFormatter<R> {
             &R::DateTimePatternV1Marker::bind(provider),
             locale,
             length,
-            components.into(),
+            components,
         )
         .map_err(LoadError::Data)?;
         let mut names = RawDateTimeNames::new_without_fixed_decimal_formatter();

components/experimental/src/dimension/provider/pattern_key.rs

@@ -123,7 +123,7 @@ impl AsULE for PatternKey {
             PatternKey::Decimal(value) => {
                 let sign = if value < 0 { 0b0010_0000 } else { 0 };
                 debug_assert!(value > -32 && value < 32);
-                (0b01 << 6) | sign | (value as u8 & 0b0001_1111)
+                (0b01 << 6) | sign | (value.unsigned_abs() & 0b0001_1111)
             }
             PatternKey::Power { power, count } => {
                 let power_bits = {
@@ -230,6 +230,14 @@ fn test_pattern_key_ule() {
             count: CompoundCount::Two,
         }
     );
+
+    let decimal_neg_1 = PatternKey::Decimal(-1);
+    let decimal_neg_1_ule = decimal_neg_1.to_unaligned();
+    assert_eq!(decimal_neg_1_ule.0, 0b0110_0001);
+
+    let decimal_neg_1 = PatternKey::from_unaligned(decimal_neg_1_ule);
+    assert_eq!(decimal_neg_1, PatternKey::Decimal(-1));
+
     // Test invalid bytes
     let unvalidated_bytes = [0b1100_0000];
     assert_eq!(

components/experimental/src/personnames/specifications/derive_missing_initials.rs

@@ -110,7 +110,6 @@ mod tests {
     }
 
     #[test]
-    #[ignore] // not clear from the spec whether this test is correct
     fn test_multi_3_initial_should_still_only_be_2() -> Result<(), PersonNamesFormatterError> {
         let mut person_data = LiteMap::new();
         person_data.insert(
@@ -127,7 +126,9 @@ mod tests {
         };
         let result =
             super::derive_missing_initials(&person_name, &requested_field, "{0}.", "{0} {1}");
-        assert_eq!(result, "M. J.");
+
+        // TODO(#3077): broken, this should be equal
+        assert_ne!(result, "M. J.");
         Ok(())
     }
 }

components/icu/examples/tui.rs

@@ -16,9 +16,9 @@ use icu::timezone::CustomTimeZone;
 use icu_collections::codepointinvlist::CodePointInversionListBuilder;
 use std::env;
 
-fn print<T: AsRef<str>>(_input: T) {
-    #[cfg(debug_assertions)]
-    println!("{}", _input.as_ref());
+#[cfg(not(debug_assertions))]
+macro_rules! println {
+    ($($arg:tt)*) => {};
 }
 
 #[no_mangle]
@@ -40,9 +40,9 @@ fn main(_argc: isize, _argv: *const *const u8) -> isize {
         })
         .unwrap_or(5);
 
-    print(format!("\nTextual User Interface Example ({locale})"));
-    print("===================================");
-    print(format!("User: {user_name}"));
+    println!("\nTextual User Interface Example ({locale})");
+    println!("===================================");
+    println!("User: {user_name}");
 
     {
         let dtf = TypedZonedDateTimeFormatter::<Gregorian>::try_new(
@@ -56,7 +56,7 @@ fn main(_argc: isize, _argv: *const *const u8) -> isize {
 
         let formatted_dt = dtf.format(&today_date, &today_tz);
 
-        print(format!("Today is: {formatted_dt}"));
+        println!("Today is: {formatted_dt}");
     }
 
     {
@@ -68,9 +68,9 @@ fn main(_argc: isize, _argv: *const *const u8) -> isize {
         let only_latin1 = user_name.chars().all(|ch| latin1_set.contains(ch));
 
         if only_latin1 {
-            print("User name latin1 only: true");
+            println!("User name latin1 only: true");
         } else {
-            print("User name latin1 only: false");
+            println!("User name latin1 only: false");
         }
     }
 
@@ -79,8 +79,8 @@ fn main(_argc: isize, _argv: *const *const u8) -> isize {
             .expect("Failed to create PluralRules.");
 
         match pr.category_for(email_count) {
-            PluralCategory::One => print("Note: You have one unread email."),
-            _ => print(format!("Note: You have {email_count} unread emails.")),
+            PluralCategory::One => println!("Note: You have one unread email."),
+            _ => println!("Note: You have {email_count} unread emails."),
         }
     }
 

components/icu/src/datagen.rs

@@ -71,9 +71,10 @@ icu_registry::registry!(cb);
 ///
 /// ```no_run
 /// # use icu_provider::DataMarker;
+/// # use std::path::Path;
 /// # fn main() -> Result<(), Box<dyn std::error::Error>> {
 /// assert_eq!(
-///     icu::markers_for_bin("target/release/my-app")?,
+///     icu::markers_for_bin(Path::new("target/release/my-app"))?,
 ///     std::collections::HashSet::from_iter([
 ///         icu::list::provider::AndListV2Marker::INFO,
 ///         icu::list::provider::OrListV2Marker::INFO,
@@ -82,11 +83,8 @@ icu_registry::registry!(cb);
 /// # Ok(())
 /// # }
 /// ```
-pub fn markers_for_bin<P: AsRef<Path>>(path: P) -> Result<HashSet<DataMarkerInfo>, DataError> {
-    let file = std::fs::read(path.as_ref())?;
-    let file = file.as_slice();
-
-    markers_for_bin_inner(file)
+pub fn markers_for_bin(path: &Path) -> Result<HashSet<DataMarkerInfo>, DataError> {
+    markers_for_bin_inner(&std::fs::read(path)?)
 }
 
 #[test]

components/segmenter/tests/css_word_break.rs

@@ -154,7 +154,7 @@ fn wordbreak_keepall() {
 }
 
 #[test]
-#[cfg_attr(not(feature = "lstm"), ignore)]
+#[cfg(feature = "lstm")]
 fn wordbreak_keepall_lstm() {
     // from css/css-text/word-break/word-break-keep-all-003.html
     let s = "และและ";

ffi/capi/src/provider.rs

@@ -69,7 +69,9 @@ pub mod ffi {
             Ok(Box::new(convert_buffer_provider(
                 icu_provider_fs::FsDataProvider::try_new(
                     // In the future we can start using OsString APIs to support non-utf8 paths
-                    core::str::from_utf8(path).map_err(|_| ICU4XDataError::Io)?,
+                    core::str::from_utf8(path)
+                        .map_err(|_| ICU4XDataError::Io)?
+                        .into(),
                 )?,
             )))
         }