chore: release v0.12.0 (#77)

* chore: release v0.12.0

* Improve changelogs

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: Frank McSherry <fmcsherry@me.com>

Author: github-actions[bot]

CHANGELOG.md

@@ -7,6 +7,35 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.12.0](https://github.com/frankmcsherry/columnar/compare/columnar-v0.11.1...columnar-v0.12.0) - 2026-03-25
+
+### Added
+
+- Structured decoding via `DecodedStore`, a zero-allocation random-access view into indexed-encoded data, with constant-instruction-count field access regardless of tuple width ([#78](https://github.com/frankmcsherry/columnar/pull/78), [#79](https://github.com/frankmcsherry/columnar/pull/79))
+- `FromBytes::validate` implementations for `Fixeds`, `Strides`, `Empties`, `Bools`, and `Discriminant`; `Stash::try_from_bytes` as the validated constructor ([#85](https://github.com/frankmcsherry/columnar/pull/85))
+- `Discriminant::is_heterogenous()` and homogeneous enum optimization: enum containers now skip per-element discriminant and offset metadata when all elements share the same variant
+- `no_std` support (with `alloc`) ([#94](https://github.com/frankmcsherry/columnar/pull/94))
+- `BorrowedOf` type alias, mirroring the existing `ContainerOf` ([#81](https://github.com/frankmcsherry/columnar/pull/81))
+- `Strides::pop` for unsealing lists at merge boundaries ([#98](https://github.com/frankmcsherry/columnar/pull/98))
+- `Strings::get_str()` convenience method for when you want `&str` with explicit validation ([#88](https://github.com/frankmcsherry/columnar/pull/88))
+- Trait implementations for `Repeats` and `Lookback` types ([#91](https://github.com/frankmcsherry/columnar/pull/91))
+- Additional `Bytes`/`Stash` properties and methods ([#97](https://github.com/frankmcsherry/columnar/pull/97))
+
+### Changed
+
+- `Strings::Ref` changed from `&str` to `&[u8]` — UTF-8 validation at index time caused up to 17x slowdown and blocked compiler optimizations ([#86](https://github.com/frankmcsherry/columnar/pull/86))
+- `element_sizes` now returns `Result`, defaulting to `Err` for unimplemented types — previously missing implementations silently accepted any byte length ([#80](https://github.com/frankmcsherry/columnar/pull/80))
+- Decoding pipeline preserves `u64` alignment throughout, making field casts infallible and enabling dead-code elimination of unused fields ([#78](https://github.com/frankmcsherry/columnar/pull/78))
+- Refreshed `Trees` container and its JSON example to current conventions ([#92](https://github.com/frankmcsherry/columnar/pull/92), [#93](https://github.com/frankmcsherry/columnar/pull/93))
+- Removed vestigial `const N: usize` generic from `Repeats` ([#91](https://github.com/frankmcsherry/columnar/pull/91))
+
+### Removed
+
+- `HeapSize` trait, replaced by `AsBytes` which exposes actual byte slices rather than capacity metrics ([#87](https://github.com/frankmcsherry/columnar/pull/87))
+- `EncodeDecode` trait and `Sequence` encoding format, superseded by the `indexed` module ([#78](https://github.com/frankmcsherry/columnar/pull/78))
+- `from_u64s`, `decode_u64s`, and `from_byte_slices` methods, replaced by `DecodedStore`/`from_store` ([#79](https://github.com/frankmcsherry/columnar/pull/79))
+- `inspect` module
+
 ## [0.11.1](https://github.com/frankmcsherry/columnar/compare/columnar-v0.11.0...columnar-v0.11.1) - 2026-01-17
 
 ### Other

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "columnar"
-version = "0.11.1"
+version = "0.12.0"
 authors = ["Frank McSherry <fmcsherry@me.com>"]
 description = "Conversion from arrays of complex structs to simple structs of arrays"
 edition = "2021"
@@ -19,7 +19,7 @@ members = ["columnar_derive", "test-no-std"]
 serde = { version = "1.0", optional = true, features = ["derive"] }
 smallvec = { version = "1.13.2", features = ["const_generics"] }
 bytemuck = { version = "1.20", features = ["min_const_generics"] }
-columnar_derive = { path = "columnar_derive", version = "0.11" }
+columnar_derive = { path = "columnar_derive", version = "0.12" }
 
 [dev-dependencies]
 bencher = "0.1.5"

columnar_derive/CHANGELOG.md

@@ -7,6 +7,19 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.12.0](https://github.com/frankmcsherry/columnar/compare/columnar_derive-v0.11.1...columnar_derive-v0.12.0) - 2026-03-25
+
+### Added
+
+- Derive support for `from_store` (structured decoding) ([#79](https://github.com/frankmcsherry/columnar/pull/79))
+- Derive support for `extend_for_self` on enum types ([#82](https://github.com/frankmcsherry/columnar/pull/82))
+
+### Changed
+
+- Generated code uses fully scoped paths (`::core::result::Result`, etc.) to avoid naming conflicts and support `no_std` ([#96](https://github.com/frankmcsherry/columnar/pull/96))
+- Generated `FromBytes` implementations use compile-time-constant byte slice counts ([#78](https://github.com/frankmcsherry/columnar/pull/78))
+- Consolidated scalar fields and streamlined naming in generated code ([#88](https://github.com/frankmcsherry/columnar/pull/88))
+
 ## [0.11.0](https://github.com/frankmcsherry/columnar/compare/columnar_derive-v0.10.2...columnar_derive-v0.11.0) - 2025-09-29
 
 ### Other

columnar_derive/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "columnar_derive"
-version = "0.11.1"
+version = "0.12.0"
 authors = ["Frank McSherry <fmcsherry@me.com>"]
 description = "Derive macros for columnar crate"
 edition = "2021"