Datagen fallback modes and singletons (#3487)

components/calendar/src/provider.rs

@@ -71,8 +71,8 @@ pub struct EraStartDate {
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
 #[icu_provider::data_struct(
-    marker(JapaneseErasV1Marker, "calendar/japanese@1"),
-    marker(JapaneseExtendedErasV1Marker, "calendar/japanext@1")
+    marker(JapaneseErasV1Marker, "calendar/japanese@1", singleton),
+    marker(JapaneseExtendedErasV1Marker, "calendar/japanext@1", singleton)
 )]
 #[derive(Debug, PartialEq, Clone, Default)]
 #[cfg_attr(

components/collator/src/provider.rs

@@ -195,7 +195,8 @@ impl<'data> CollationDataV1<'data> {
     "collator/dia@1",
     extension_key = "co",
     fallback_by = "collation",
-    fallback_supplement = "collation"
+    fallback_supplement = "collation",
+    singleton,
 ))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_collator::provider))]
@@ -216,7 +217,7 @@ pub struct CollationDiacriticsV1<'data> {
 /// including in SemVer minor releases. While the serde representation of data structs is guaranteed
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
-#[icu_provider::data_struct(CollationJamoV1Marker = "collator/jamo@1")]
+#[icu_provider::data_struct(marker(CollationJamoV1Marker, "collator/jamo@1", singleton))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_collator::provider))]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]
@@ -420,7 +421,11 @@ impl CollationMetadataV1 {
 /// including in SemVer minor releases. While the serde representation of data structs is guaranteed
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
-#[icu_provider::data_struct(CollationSpecialPrimariesV1Marker = "collator/prim@1")]
+#[icu_provider::data_struct(marker(
+    CollationSpecialPrimariesV1Marker,
+    "collator/prim@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_collator::provider))]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]

components/datetime/src/lib.rs

@@ -215,7 +215,7 @@ mod tests {
         check_size_of!(5752 | 4544, TypedDateFormatter::<Gregorian>);
         check_size_of!(6744 | 5408, TypedDateTimeFormatter::<Gregorian>);
 
-        check_size_of!(80, DateTimeError);
+        check_size_of!(88, DateTimeError);
         check_size_of!(176, FormattedDateTime);
         check_size_of!(16, FormattedTimeZone::<CustomTimeZone>);
         check_size_of!(160, FormattedZonedDateTime);

components/locid_transform/src/fallback/mod.rs

@@ -290,17 +290,12 @@ impl LocaleFallbacker {
     ///
     /// [📚 Help choosing a constructor](icu_provider::constructors)
     #[cfg(feature = "data")]
-    pub const fn new() -> Self {
-        Self {
-            likely_subtags: DataPayload::from_static_ref(
-                crate::provider::Baked::SINGLETON_FALLBACK_LIKELYSUBTAGS_V1,
-            ),
-            parents: DataPayload::from_static_ref(
-                crate::provider::Baked::SINGLETON_FALLBACK_PARENTS_V1,
-            ),
-            collation_supplement: Some(DataPayload::from_static_ref(
-                crate::provider::Baked::SINGLETON_FALLBACK_SUPPLEMENT_CO_V1,
-            )),
+    #[allow(clippy::new_ret_no_self)] // keeping constructors together
+    pub const fn new() -> LocaleFallbackerBorrowed<'static> {
+        LocaleFallbackerBorrowed {
+            likely_subtags: crate::provider::Baked::SINGLETON_FALLBACK_LIKELYSUBTAGS_V1,
+            parents: crate::provider::Baked::SINGLETON_FALLBACK_PARENTS_V1,
+            collation_supplement: Some(crate::provider::Baked::SINGLETON_FALLBACK_SUPPLEMENT_CO_V1),
         }
     }
 

components/locid_transform/src/provider/canonicalizer.rs

@@ -8,7 +8,7 @@ use icu_provider::prelude::*;
 use tinystr::UnvalidatedTinyAsciiStr;
 use zerovec::{VarZeroVec, ZeroMap, ZeroSlice};
 
-#[icu_provider::data_struct(AliasesV1Marker = "locid_transform/aliases@1")]
+#[icu_provider::data_struct(marker(AliasesV1Marker, "locid_transform/aliases@1", singleton))]
 #[derive(PartialEq, Clone, Default)]
 #[cfg_attr(
     feature = "datagen",

components/locid_transform/src/provider/directionality.rs

@@ -6,7 +6,11 @@ use super::*;
 use icu_provider::prelude::*;
 use zerovec::ZeroVec;
 
-#[icu_provider::data_struct(ScriptDirectionV1Marker = "locid_transform/script_dir@1")]
+#[icu_provider::data_struct(marker(
+    ScriptDirectionV1Marker,
+    "locid_transform/script_dir@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(
     feature = "datagen",

components/locid_transform/src/provider/expander.rs

@@ -7,7 +7,11 @@ use icu_locid::subtags::{Language, Region, Script};
 use icu_provider::prelude::*;
 use zerovec::ZeroMap;
 
-#[icu_provider::data_struct(LikelySubtagsV1Marker = "locid_transform/likelysubtags@1")]
+#[icu_provider::data_struct(marker(
+    LikelySubtagsV1Marker,
+    "locid_transform/likelysubtags@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(
     feature = "datagen",
@@ -57,7 +61,11 @@ pub struct LikelySubtagsV1<'data> {
     pub und: (Language, Script, Region),
 }
 
-#[icu_provider::data_struct(LikelySubtagsForLanguageV1Marker = "locid_transform/likelysubtags_l@1")]
+#[icu_provider::data_struct(marker(
+    LikelySubtagsForLanguageV1Marker,
+    "locid_transform/likelysubtags_l@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(
     feature = "datagen",
@@ -123,9 +131,11 @@ impl<'data> LikelySubtagsForLanguageV1<'data> {
     }
 }
 
-#[icu_provider::data_struct(
-    LikelySubtagsForScriptRegionV1Marker = "locid_transform/likelysubtags_sr@1"
-)]
+#[icu_provider::data_struct(marker(
+    LikelySubtagsForScriptRegionV1Marker,
+    "locid_transform/likelysubtags_sr@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(
     feature = "datagen",
@@ -177,7 +187,11 @@ impl<'data> From<LikelySubtagsV1<'data>> for LikelySubtagsForScriptRegionV1<'dat
     }
 }
 
-#[icu_provider::data_struct(LikelySubtagsExtendedV1Marker = "locid_transform/likelysubtags_ext@1")]
+#[icu_provider::data_struct(marker(
+    LikelySubtagsExtendedV1Marker,
+    "locid_transform/likelysubtags_ext@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(
     feature = "datagen",

components/locid_transform/src/provider/fallback.rs

@@ -12,7 +12,11 @@ use zerovec::ZeroMap;
 use zerovec::ZeroMap2d;
 
 /// Locale fallback rules derived from likely subtags data.
-#[icu_provider::data_struct(LocaleFallbackLikelySubtagsV1Marker = "fallback/likelysubtags@1")]
+#[icu_provider::data_struct(marker(
+    LocaleFallbackLikelySubtagsV1Marker,
+    "fallback/likelysubtags@1",
+    singleton
+))]
 #[derive(Default, Clone, PartialEq, Debug)]
 #[cfg_attr(
     feature = "datagen",
@@ -55,7 +59,11 @@ pub const DEFAULT_SCRIPT: Script = script!("Latn");
 pub const DEFAULT_REGION: Region = region!("ZZ");
 
 /// Locale fallback rules derived from CLDR parent locales data.
-#[icu_provider::data_struct(LocaleFallbackParentsV1Marker = "fallback/parents@1")]
+#[icu_provider::data_struct(marker(
+    LocaleFallbackParentsV1Marker,
+    "fallback/parents@1",
+    singleton
+))]
 #[derive(Default, Clone, PartialEq, Debug)]
 #[cfg_attr(
     feature = "datagen",
@@ -74,7 +82,8 @@ pub struct LocaleFallbackParentsV1<'data> {
 /// Key-specific supplemental fallback data.
 #[icu_provider::data_struct(marker(
     CollationFallbackSupplementV1Marker,
-    "fallback/supplement/co@1"
+    "fallback/supplement/co@1",
+    singleton,
 ))]
 #[derive(Default, Clone, PartialEq, Debug)]
 #[cfg_attr(

components/normalizer/src/provider.rs

@@ -27,7 +27,11 @@ use zerovec::ZeroVec;
 /// including in SemVer minor releases. While the serde representation of data structs is guaranteed
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
-#[icu_provider::data_struct(CanonicalDecompositionDataV1Marker = "normalizer/nfd@1")]
+#[icu_provider::data_struct(marker(
+    CanonicalDecompositionDataV1Marker,
+    "normalizer/nfd@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_normalizer::provider))]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]
@@ -46,8 +50,12 @@ pub struct DecompositionDataV1<'data> {
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
 #[icu_provider::data_struct(
-    CompatibilityDecompositionSupplementV1Marker = "normalizer/nfkd@1",
-    Uts46DecompositionSupplementV1Marker = "normalizer/uts46d@1"
+    marker(
+        CompatibilityDecompositionSupplementV1Marker,
+        "normalizer/nfkd@1",
+        singleton
+    ),
+    marker(Uts46DecompositionSupplementV1Marker, "normalizer/uts46d@1", singleton)
 )]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_normalizer::provider))]
@@ -95,8 +103,12 @@ impl DecompositionSupplementV1<'_> {
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
 #[icu_provider::data_struct(
-    CanonicalDecompositionTablesV1Marker = "normalizer/nfdex@1",
-    CompatibilityDecompositionTablesV1Marker = "normalizer/nfkdex@1"
+    marker(CanonicalDecompositionTablesV1Marker, "normalizer/nfdex@1", singleton),
+    marker(
+        CompatibilityDecompositionTablesV1Marker,
+        "normalizer/nfkdex@1",
+        singleton
+    )
 )]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_normalizer::provider))]
@@ -118,7 +130,7 @@ pub struct DecompositionTablesV1<'data> {
 /// including in SemVer minor releases. While the serde representation of data structs is guaranteed
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
-#[icu_provider::data_struct(CanonicalCompositionsV1Marker = "normalizer/comp@1")]
+#[icu_provider::data_struct(marker(CanonicalCompositionsV1Marker, "normalizer/comp@1", singleton))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_normalizer::provider))]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]
@@ -138,7 +150,11 @@ pub struct CanonicalCompositionsV1<'data> {
 /// including in SemVer minor releases. While the serde representation of data structs is guaranteed
 /// to be stable, their Rust representation might not be. Use with caution.
 /// </div>
-#[icu_provider::data_struct(NonRecursiveDecompositionSupplementV1Marker = "normalizer/decomp@1")]
+#[icu_provider::data_struct(marker(
+    NonRecursiveDecompositionSupplementV1Marker,
+    "normalizer/decomp@1",
+    singleton
+))]
 #[derive(Debug, PartialEq, Clone)]
 #[cfg_attr(feature = "datagen", derive(serde::Serialize, databake::Bake), databake(path = icu_normalizer::provider))]
 #[cfg_attr(feature = "serde", derive(serde::Deserialize))]