1.1.0

[robertbastian]

icu_calendar

• Fix bug in simple_week_of() around unit size (Fix bug in simple_week_of() around unit size #2951)
• Fix math in calendar (Fix math in calendar #2714)
• Add div_rem_euclid and use it in icu_calendar (Add div_rem_euclid and use it in icu_calendar #2704)
• Fix Time::from_minute_with_remainder_days to handle negatives (Fix Time::from_minute_with_remainder_days to handle negatives properly #2643) (Fix Time::from_minute_with_remainder_days to handle negatives #2702)
• doc improvements

icu_casemapping

• doc improvements

icu_collator

• doc improvements

icu_collections

• Add APIs for returning exemplar characters data (Add APIs for returning exemplar characters data #2812)
• Readable JSON inversion lists (Readable JSON inversion lists #2855)
• Add UnicodeSet that supports strings (Add UnicodeSet that supports strings #2796)
• Add documentation on CodePointTrie details and perf considerations (Add documentation on CodePointTrie details and perf considerations #2717)

icu_codepointtrie_builder

• internal improvements

icu_datetime

• internal and doc improvements

icu_decimal

• internal and doc improvements

icu_displaynames

• DisplayNames fixes (DisplayNames fixes #2918)
• Rename Territory -> Region for Displayname component (Rename Territory -> Region for Displayname component #2895)
• Transformer code for Language display names (Transformer code for Language display names #2871)
• Adding a function to get display-name for a region. (Adding a function to get display-name for a region. #2816)
• Transformer code for DisplayNames component. (Transformer code for DisplayNames component. #2635)
• doc improvements

icu_list

• Untangling list provider from logic and fixing big endian safety bug (Untangling list provider from logic and fixing big endian safety bug #2994)
• Not allocating Writeables for regex evaluation (Not allocating Writeables for regex evaluation #2991)
• doc improvements

icu_locid

• Reject duplicated extensions ([locid] Reject duplicated extensions #2893)
• More borrowing in locid's write_to_string (More borrowing in locid's write_to_string #2693)
• doc improvements

icu_locid_transform

• Clean up dependency specifications so serde isn't pulled in by default (Clean up dependency specifications so serde isn't pulled in by default #2696)
• doc improvements

icu_normalizer

• internal and doc improvements

icu_plurals

• doc improvements

icu_properties

• Add APIs for returning exemplar characters data (Add APIs for returning exemplar characters data #2812)
• Add API and testdata for Basic_Emoji property (Add API and testdata for Basic_Emoji property #2802)
• Add UnicodeSet that supports strings (Add UnicodeSet that supports strings #2796)
• Update Script property value enums (Update property value enums for Unicode 15 #2787)
• doc improvements

icu_segmenter

• Make metacrate features more specific (Make metacrate features more specific #2932)
• Remove serde dependency from segmenter with lstm feature. (Remove serde dependency from segmenter with lstm feature. #2904)
• Simplify construction of grapheme cluster break iterators (Simplify construction of grapheme cluster break iterators #2870)
• Store grapheme cluster payload instead of grapheme cluster segmenter. (Store grapheme cluster payload instead of grapheme cluster segmenter. #2864)
• #[no_std] for LSTM segmenter (#[no_std] for LSTM segmenter #2845)
• icu_segmenter: enforce clippy::indexing_slicing. (icu_segmenter denies the clippy::indexing_slicing. #2325)
• Use GraphemeClusterSegmenter in DictionarySegmenter and LstmSegmenter (Use GraphemeClusterSegmenter in DictionarySegmenter and LstmSegmenter #2716)
• Rename *BreakSegmenter to *Segmenter (Rename *BreakSegmenter to *Segmenter #2707)
• Remove unnecessary langauge check for East Asian languagne (SA property) (Remove unnecessary langauge check for East Asian languagne (SA property) #2705)
• internal and doc improvements

icu_timezone

• Adds a bytes parsing API for GMTOffset for CustomTimeZone and FFI (Adds a bytes parsing API for GMTOffset for CustomTimeZone and FFI #2943, Fix tz offset bounds #2955)
• doc improvements

icu_provider_adapters

• Add more inner_mut functions in icu_provider_adapters (Add more inner_mut functions in icu_provider_adapters #2987)
• Fix error propagation in MultiForkByErrorProvider (Fix error propagation in MultiForkByErrorProvider #2986)
• Add mutation methods to MultiForkByErrorProvider (Add mutation methods to MultiForkByErrorProvider #2972)

icu_provider_blob

• internal and doc improvements

icu_provider

• Setting correct DataError for .as_deserializing(), .as_downcasting() (Setting correct DataError for .as_deserializing(), .as_downcasting() #2993)
• doc improvements

icu_datagen

• Removing experimental feature from datagen (Removing experimental feature from datagen #3005)
• Fixing Spanish list regex (Fixing Spanish regex #2989)
• Datagen CLI improvements (Datagen CLI improvements #2950)
• Some reexports for datagen (Some reexports for datagen #2958)
• Databake improvements (Databake improvements #2906)
• Exclude certain collations by default and add option to include them (Exclude certain collations by default and add option to include them #2789)
• Allowing no keys in datagen CLI (Allowing no keys in datagen CLI #2731)
• Fixing baked datagen for no keys and keys with no data (Fixing baked datagen for no keys and keys with no data #2698)
• Internal and doc improvements

icu_provider_fs

• internal and doc improvements

icu_testdata

• stabilized metadata
• doc improvements
• CLDR 42 and ICU 72 data

bies

• Bumped writeable version

databake

• Doc improvements

fixed_decimal

• Treat strings starting with a decimal point as valid in FixedDecimal::from_str() (Treat strings starting with a decimal point as valid in FixedDecimal::from_str() #2937)
• The actual formatting part of compact decimal formatting (The actual formatting part of compact decimal formatting #2898)
• Allow noncompact CompactDecimal (Allow noncompact CompactDecimal #2889)
• CompactDecimal and ScientificDecimal (CompactDecimal and ScientificDecimal #2847)
• doc improvements

icu_pattern

• internal improvements

litemap

• internal and doc improvements

tinystr

• Add std feature and Error impl for TinyStrError (Add std feature and Error impl for TinyStrError #3009)
• internal and doc improvements

tzif

• internal and doc improvements

writeable

• Using core integer log when available (Using core integer log when available #3015)
• usize and isize implementation
• internal and doc improvements

yoke

• Add prove_covariance_manually guard for CoerceUnsized (Add prove_covariance_manually guard for CoerceUnsized #2936)
• Allow clippy::forget_copy in derive(Yokeable) impl (Allow clippy::forget_copy in derive(Yokeable) impl #2775)
• Fix soundness issue in Yoke::attach_to_cart() around implied bounds Fix soundness issue in Yoke::attach_to_cart() around implied bounds #2949

zerovec

• ZeroMap2d cursors

[Manishearth]

@robertbastian @sffc can we land #2949 and bump Yoke first? it's technically a breaking change.

I'd rather not block that on IsCovariant, a thing which is currently unused and I'd like to redesign again based on actual need.

[robertbastian]

I haven't seen any activity on that PR, and it's not in the 1.1 Blocking milestone. What's the ETA on it?

[Manishearth]

@robertbastian it should be mergeable once Shane responds

it didn't occur to me that it should be blocking, but i think it should

[Manishearth]

@robertbastian feel free to roll #3019 into your PR, but I can just merge and publish independently

[Manishearth]

Also, changelog for Yoke

• Add prove_covariance_manually guard for CoerceUnsized (Add prove_covariance_manually guard for CoerceUnsized #2936)
• Allow clippy::forget_copy in derive(Yokeable) impl (Allow clippy::forget_copy in derive(Yokeable) impl #2775)
• Fix soundness issue in Yoke::attach_to_cart() around implied bounds Fix soundness issue in Yoke::attach_to_cart() around implied bounds #2949

[sffc]

Haven't read up yet on cargo workspaces but it looks like it might be quite helpful

[sffc]

I'm not really sure what are the implications of this line

[robertbastian]

cargo ws version can be used to bump the versions. By default all crates are versioned together.

[sffc]

Should we remove these [package.metadata.workspaces] lines now?

[robertbastian]

I'd still suggest cargo ws version to bump version, it's really convenient.

[Manishearth]

Hmm, won't cargo ws prevent us from doing out-of-cycle dot releases since it wants things to be in sync? I'm perhaps not fully clear on what the tool does (at a lower level).

[Manishearth]

Can we use Cargo's regular workspace-level versioning support? It's not available on our MSRV, but that doesn't actually matter since local tests won't hit it (I think?), and cargo normalizes before publish so crates.io users don't need to worry.

I'd like to stay as close to vanilla Cargo as possible. I'm worried about introducing a dependency on a new tool which is only invoked once in a blue moon (making it much harder to catch any bugs). We have problems with cargo make at times but they're caught immediately, which makes it tolerable, I'm worried about

And plus; publishes aren't a retryable operation; if a publish succeeds but was run with the wrong options, you can't really undo that. I'm still unclear on how this tool works but it ... might be doing some Cargo.toml fixups before publishing, which seems super brittle.

I'm fine with using this tool for bumping stuff, but highly suspicious about requiring it for publishing.

[robertbastian]

I was expected cargo ws to set the versions, but apparently it doesn't. Are we sure that cargo publish doesn't work with just a path? The path fully determines the dependency's version, because it points at a local crate with one version. I was hoping to get rid of all the duplicated versions.

I don't think we can use "workspace-level versioning support" because then our MSRV CI would fail (although I guess we could do something like cargo pacakage and then run MSRV on the result).

[sffc]

cargo publish has definitely complained when I've tried in the past if a dependency didn't have a version number

I'm pretty sure cargo check will complain loudly if one project's version number is out of sync with a dependent's Cargo.toml entry.

The well-trodden path that has worked for us in previous releases is to just go ahead and update all the version numbers and then cargo publish in topological order

[Manishearth]

I was expected cargo ws to set the versions, but apparently it doesn't. Are we sure that cargo publish doesn't work with just a path? The path fully determines the dependency's version, because it points at a local crate with one version. I was hoping to get rid of all the duplicated versions.

It doesn't in my experience. And part of this is just that a local path dependency points to a specific crate, whereas Cargo.toml deps are a range.

Which means this is going to be a problem for us once we want to do patch releases and stuff, since we'll sometimes want a crate to bump its dep on, say, yoke, whereas in other cases the yoke change may be purely internal and not require a dep bump.

So I'd say we should undo these changes.

I don't think we can use "workspace-level versioning support" because then our MSRV CI would fail (although I guess we could do something like cargo pacakage and then run MSRV on the result).

Yeah, I was hoping Cargo would silently ignore workspace = true locally but it still notices it on older versions.

[robertbastian]

The well-trodden path that has worked for us in previous releases is to just go ahead and update all the version numbers and then cargo publish in topological order

I was hoping to improve on that because that's a lot of stupid work, but alas not.

[sffc]

Also, if you discover any further issues while publishing (not unlikely), so long as they are only Cargo.toml changes, I've usually pushed them directly to the main branch and asked for a post-submit review

[robertbastian]

I'm not sure I can push to main.