Use Blob2 in more places (#4281)

Fixes #4216
unicode-org · Nov 14, 2023 · f2d98b1 · f2d98b1
1 parent 5774ab6
commit f2d98b1
Show file tree

Hide file tree

Showing 7 changed files with 18 additions and 10 deletions.
diff --git a/components/datetime/tests/data/blob.postcard b/components/datetime/tests/data/blob.postcard
diff --git a/components/datetime/tests/data/gen.sh b/components/datetime/tests/data/gen.sh
@@ -2,6 +2,6 @@
 cargo run -p icu_datagen -- \
 --keys "datetime/gregory/datelengths@1" "datetime/gregory/datesymbols@1" "datetime/timelengths@1" "datetime/timesymbols@1" "decimal/symbols@1" "time_zone/formats@1" "time_zone/specific_short@1" \
 --locales en \
---format blob \
+--format blob2 \
 --out $(dirname $0)/blob.postcard \
 --overwrite
diff --git a/provider/blob/README.md b/provider/blob/README.md
diff --git a/provider/blob/src/export/mod.rs b/provider/blob/src/export/mod.rs
@@ -17,7 +17,7 @@
 //! let mut blob: Vec<u8> = Vec::new();
 //!
 //! // Set up the exporter
-//! let mut exporter = BlobExporter::new_with_sink(Box::new(&mut blob));
+//! let mut exporter = BlobExporter::new_v2_with_sink(Box::new(&mut blob));
 //!
 //! // Export something
 //! DatagenDriver::new()
@@ -27,6 +27,8 @@
 //!     .unwrap();
 //!
 //! // communicate the blob to the client application (network, disk, etc.)
+//! # let golden_blob = std::fs::read(concat!(env!("CARGO_MANIFEST_DIR"), "/tests/data/v2.postcard")).unwrap();
+//! # assert_eq!(blob, golden_blob);
 //! ```
 //!
 //! The resulting blob can now be used like this:

diff --git a/provider/blob/src/lib.rs b/provider/blob/src/lib.rs
@@ -5,13 +5,13 @@
 //! `icu_provider_blob` contains [`BlobDataProvider`], a [`BufferProvider`] implementation that
 //! supports loading data from a single serialized blob.
 //!
-//! To build blob data, use the `--format blob` option of [`icu_datagen`]:
+//! To build blob data, use the `--format blob2` option of [`icu_datagen`]:
 //!
 //! ```bash
-//! $ icu4x-datagen --keys all --locales full --format blob --out data.postcard
+//! $ icu4x-datagen --keys all --locales full --format blob2 --out data.postcard
 //! ```
 //!
-//! For examples, see the specific data providers.
+//! You can also use `--format blob` if you need to support ICU4X versions prior to 1.4.
 //!
 //! [`ICU4X`]: ../icu/index.html
 //! [`BufferProvider`]: icu_provider::BufferProvider

diff --git a/provider/datagen/README.md b/provider/datagen/README.md
diff --git a/provider/datagen/src/lib.rs b/provider/datagen/src/lib.rs
@@ -25,7 +25,7 @@
 //!     .with_all_locales()
 //!     .export(
 //!         &DatagenProvider::new_latest_tested(),
-//!         BlobExporter::new_with_sink(Box::new(
+//!         BlobExporter::new_v2_with_sink(Box::new(
 //!             File::create("data.postcard").unwrap(),
 //!         )),
 //!     )
@@ -468,6 +468,11 @@ pub enum Out {
         fingerprint: bool,
     },
     /// Output as a postcard blob to the given sink.
+    ///
+    /// This supports only version 1 of the blob format. Please use [`DatagenDriver`] with
+    /// [`BlobExporter`] to export to blob format version 2.
+    ///
+    /// [`BlobExporter`]: crate::blob_exporter::BlobExporter
     Blob(Box<dyn std::io::Write + Sync>),
     /// Output a module with baked data at the given location.
     Baked {
@@ -610,6 +615,7 @@ pub fn datagen(
                                 )?)
                             }
                             Out::Blob(write) => {
+                                // Note: no blob v2 support in legacy API
                                 Box::new(blob_exporter::BlobExporter::new_with_sink(write))
                             }
                             Out::Baked {