v0.9.2

What's Changed

optype

• mark optype.Just* as @final and simplify metaclasses by @jorenham in #299
• add optype.JustDate by @jorenham in #300

optype.numpy

• numpy>=2.2 fixes for array- and dtype-likes by @jorenham in #301
• new optype.numpy.random submodule by @jorenham in #302
• common numpy literal types by @jorenham in #303

Dependencies

• update the dev deps by @jorenham in #291
• bump numpy to 2.2.3 by @jorenham in #292
• remove pre-commit by @jorenham in #296
• update development dependencies by @jorenham in #297
• bump basedmypy to 2.10.0 by @jorenham in #298

Other Changes

• prepare for v0.9.2 by @jorenham in #289

Full Changelog: v0.9.1...v0.9.2

v0.9.1

What's Changed

optype.numpy

• fix longdouble dtype name literal by @jorenham in #283
• remove float80 and complex160 from [c]longdouble by @jorenham in #285
• add onp.NDim0, which, unlike NDim, excludes 0 by @jorenham in #286
• add literal onp.To[Just]{False,True} aliases by @jorenham in #287
• generic onp.ToArrayStrict{1,2,3}D by @jorenham in #288

Dependencies

• Update the development dependencies by @jorenham in #281
• bump ruff to 0.9.4 by @jorenham in #284

Other Changes

• prepare for v0.9.1 by @jorenham in #277

Full Changelog: v0.9.0...v0.9.1

v0.9.0

Highlights

• ⚡ Faster import optype as op by deferring submodule imports (#251)
• ✨ New op.io submodule (docs) (#253)
• ✨ New and improved op.Just types (docs):
  • Just[T] - Accepts _: T iff. type(_) is T.
  • JustInt - Accepts _: int and literal ints, rejects e.g. _: bool
  • JustFloat - Accepts _: float, rejects e.g. _: int and _: numpy.float64
  • JustComplex - Accepts _: complex, rejects _: float and _: numpy.complex128
  • JustBytes - Accepts _: bytes and bytes-literals, rejects _: bytearray, _: memoryview, and _: numpy.bytes_
  • JustObject - Accepts _: object, rejects everything else. Useful for annotating sentinels like DEFAULT = object().
• ✨ New op.numpy.is_array_{0,1,2,3,n}d typeguards (docs) (#270)

Deprecations

• The op.typing.Just* types are deprecated in favor of op.Just*.

What's Changed

optype

• generic optype.HasClass: suprisingly useful! by @jorenham in #257
• move Just* from optype.typing to optype and deprecate by @jorenham in #258
• custom runtime-checkable Just* types by @jorenham in #260
• Just without redundant typing hacks by @jorenham in #261
• add JustObject by @jorenham in #264
• add JustBytes by @jorenham in #265
• change HasFunc and HasWrapped generic type params by @jorenham in #267
• change HasTypeParams type params by @jorenham in #268

optype.numpy

• prefer Just{Float,Complex} over Just[{float,complex}] in optype.numpy by @jorenham in #240
• add the missing Just types to onp.Any*DType by @jorenham in #243
• optype.numpy.is_ typeguard by @jorenham in #270
• split the private op.numpy._any_dtype module into two by @jorenham in #275
• new optype.numpy.ToDType alias by @jorenham in #276

Dependencies

• ruff 0.9.1, basedpyright 1.23.2, basedmypy 2.9.1 by @jorenham in #239
• bump basedpyright to 1.24.0 by @jorenham in #255
• bump ruff to 0.9.2 by @jorenham in #259
• bump numpy to 2.2.2 (dev-only) by @jorenham in #263
• bump pre-commit to 4.1.0 by @jorenham in #271

Continuous integration

• trigger the build & publish workflow on (pre)release by @jorenham in #238
• automatically label PR's by @jorenham in #247
• add CODEOWNERS by @jorenham in #248

Other Changes

• more consistent style for imports and exports by @jorenham in #246
• lazy submodule imports by @jorenham in #251
• optype.io by @jorenham in #253
• stop using __future__.annotations and typing.TYPE_CHECKING by @jorenham in #256
• cleaner ruff.lint.isort config by @jorenham in #262
• cleaner private module structure by @jorenham in #266
• format pyproject.toml by @jorenham in #269
• update the auto-release changelog config by @jorenham in #272
• globally ignore common inline typechecker ignores by @jorenham in #273

Full Changelog: v0.8.0...v0.9.0

v0.8.0

What's Changed

optype.numpy

• accept more valid numpy array-likes in optype.numpy.To*{1,2,3,N} by @jorenham in #214
• onp.Any[U]Int{DType,Array} for np.int_ by @jorenham in #224
• include opt.Just[{int,float,complex}] in onp.Any{Int,Float,Complex}{Array,DType} by @jorenham in #225
• new onp.compat module, and Any as shape-type default on numpy<2.1 by @jorenham in #226
• new onp.Matrix and onp.MArray[{0,1,2,3}D] aliases by @jorenham in #228
• onp.AnyBytes8DType for np.dtype("c") by @jorenham in #230
• new onp.ToJust{Bool,Float,Complex} aliases by @jorenham in #231
• new onp.To[Just]{Float64,Complex128}* aliases by @jorenham in #232

optype.pickle

• Added missed optype documentation by @kam193 in #212

optype.typing

• add JustFloat and JustComplex to optype.typing by @jorenham in #215

Dependencies

• migrate from codespell to typos by @jorenham in #213
• bump basedmypy to 2.8.0 and basedpyright to 1.22.1 by @jorenham in #218
• update development dependencies by @jorenham in #223

Continuous integration

• test & typecheck all of the supported numpy versions by @jorenham in #227
• PyPI trusted publishing by @jorenham in #234

Other Changes

• update the security policy by @jorenham in #216
• automatically generated release notes configuration by @jorenham in #233

New Contributors

• @kam193 made their first contribution in #212

Full Changelog: v0.7.3...v0.8.0

v0.7.3

Highlights

• optype is now on conda-forge, thanks to @lucascolley (docs)
• [optype.numpy] The CanArrayND type now optionally accepts a second shape-type argument (docs)
• [optype.numpy] New Array0D and CanArray0D aliases for 0-dimensional numpy arrays (not scalars) (docs)
• [optype.numpy] New To*Strict{1,2,3}D array-like types with "strict" shape, useful for non-overlapping shape-type overloads (docs)

Fixes

• Fix ImportError on python >= 3.13 without typing_extensions installed

What's Changed

• fix license metadata by @jorenham in #202
• bump ruff to 0.8.1 by @jorenham in #203
• Conda installation instructions by @jorenham in #204
• fix typing_extensions import on python>=3.13 by @jorenham in #205
• optional onp.CanArrayND shape-type param, improved shape-type compatibility by @jorenham in #206
• onp.To*Strict{1,2,3}D: array-like aliases with strict shape-type by @jorenham in #207
• fix (more) typing_extensions imports on python>=3.13 by @jorenham in #208
• 0-d numpy array aliases by @jorenham in #209

Full Changelog: v0.7.2...v0.7.3

v0.7.2

Highlights

• optype.numpy.To*3D - Array-like aliases for 3-D array-likes (docs)
• optype.numpy.ToJustInt* - Scalar- and array-like aliases for just integer array-likes, i.e. bool and np.bool_ will be rejected (docs)

Fixes

• #194 - workaround for a mypy bug with recursive sequences types

What's Changed

• ruff 0.8.0 and basedpyright 1.22.0 by @jorenham in #195
• workaround for a mypy bug with recursive sequences types by @jorenham in #196
• export SequenceND from optype.numpy by @jorenham in #197
• 3-d aliases for arrays and array-likes in optype.numpy by @jorenham in #198
• ToJustInt scalar- and array-likes in optype.numpy by @jorenham in #199

Full Changelog: v0.7.1...v0.7.2

v0.7.1

What's Changed

• fix onp.ToArray{1,2}D names by @jorenham in #193

Full Changelog: v0.7.0...v0.7.1

v0.7.0

Highlights

• [BREAKING] The optype.numpy.Any* aliases no longer accept bare scalars types
• Coercible scalar- and array-like type aliases optype.numpy.Array{1,2,3,N}D (docs)
• The new optype.dlpack module with DLPack types and enums (docs)
• Improved the readability of the optype.numpy types
• Experimental Just and JustInt in optype.typing (docs)

What's Changed

• development dependency updates by @jorenham in #160
• native toml tox config by @jorenham in #161
• switch from poetry to uv by @jorenham in #162
• switch from single to double quotes for black compatibility by @jorenham in #163
• enforce ruff format by @jorenham in #164
• clean up the internal module structure by @jorenham in #165
• update pre-commit hooks by @pre-commit-ci in #166
• update development dependencies by @jorenham in #168
• optype.dlpack: DLPack types and enums by @jorenham in #169
• optype.array_api - dtypes by @jorenham in #170
• Revert "optype.array_api - dtypes" by @jorenham in #176
• PEP 735 dependency groups, and fixed mypy config by @jorenham in #177
• fix optype.numpy.CanArray by @jorenham in #178
• optype.types.Deprecated by @jorenham in #179
• simplified optype.numpy dtypes by @jorenham in #182
• fix and improve the optype.numpy.Any* docs by @jorenham in #183
• improve readability of optype.numpy type aliases and protocols by @jorenham in #184
• optype.numpy.Array{1,2,3,N}D aliases by @jorenham in #185
• workaround a PEP 696 bug in typing_extensions.Protocol affecting CanSequence by @jorenham in #186
• add SPEC 0 badge by @jorenham in #187
• bump basedpyright to 1.21.1 (pyright 1.1.389) by @jorenham in #188
• bump ruff to 0.7.4 by @jorenham in #189
• experimental Just and JustInt types in optype.typing by @jorenham in #191
• optype.numpy.To* coercible scalar- and array-likes by @jorenham in #192

New Contributors

• @pre-commit-ci made their first contribution in #166

Full Changelog: v0.6.1...v0.7.0

v0.6.1

This is a bugfix release with no breaking changes.
However, because optype now uses less typing.Any, which could cause type-checkers to report errors that were previously hidden.

Highlights

• Stricter type-checking with basedmypy.
• Removed almost all typing.Any uses. See numpy/numpy#27211 for why Any is evil.
• Fixed several typing errors with numpy 2.1.

What's Changed

• Farewell git-flow by @jorenham in #151
• Add security policy by @jorenham in #152
• Add a sponsor button by @jorenham in #153
• basedmypy by @jorenham in #154
• Bump NumPy to 2.1.0 by @jorenham in #155
• basedmypy badge by @jorenham in #156
• optype 0.6.1 by @jorenham in #157

Full Changelog: v0.6.0...v0.6.1

v0.6.0

Highlights

• Mypy support (strict mode)
• Beartype support
• New standard library modules
  • optype.copy
  • optype.dataclasses
  • optype.inspect
  • optype.json
  • optype.pickle
  • optype.string
  • optype.typing
• Improvements to optype.numpy (it's still optional)
  • NumPy 2.1 support
  • Improved docs
  • Shape type aliases`
  • Support for numpy.dtypes.StringDType (numpy >= 2)
  • Reworked Any*Array and Any*DType

Breaking changes

• Increased the minimum Python version from 3.10.0 to 3.10.1
• Increased the minimum typing_extensions version from 4.7 to 4.8
• Moved CanCopy, CanDeepcopy and CanReplace from optype._ to optype.copy._
• Moved HasDataclassFields from optype._ to optype.dataclasses._
• Moved CanGetnewargs[Ex], CanReduce[Ex], and Can{Get,Set}state from optype._ to optype.pickle._
• Any*Array and Any*DType for int_, float64 and complex128 don't include int, float, and complex anymore. See numpy/numpy#27032 (comment) for why this was done.

Full Changelog: v0.5.0...v0.6.0