chore: release

[github-actions]

🤖 New release

• compiler_builtins: 0.1.155 -> 0.1.156
• libm: 0.2.11 -> 0.2.12

Changelog

compiler_builtins

0.1.156 - 2025-04-20

Other

• Enable icount benchmarks in CI
• Move builtins-test-intrinsics out of the workspace
• Run cargo fmt on all projects
• Flatten the libm/libm directory
• Update path to libm after the merge
• Remove the libm submodule
• Rename testcrate to builtins-test

libm

0.2.12 - 2025-04-20

Other

• Run cargo fmt on all projects
• Add a .rustfmt.toml with style edition 2024
• Flatten the libm/libm directory
• Reorganize into compiler-builtins
• Remove compiler-builtins-smoke-test
• Update .git-blame-ignore-revs after the libm merge
• Update submodules after the libm merge
• Migrate all crates except libm to edition 2024
• Introduce a virtual manifest
• Move the libm crate to a subdirectory
• Mark generic functions #[inline]
• Combine the source files for fmod
• Ensure all public functions are marked no_panic
• Account for PR_NUMBER being set to an empty string
• Ensure configure.rs changes trigger rebuilds
• Increase the timeout for extensive tests
• Require ci: allow-many-extensive if a threshold is exceeded
• Allow skipping extensive tests with ci: skip-extensive
• Cancel jobs when a new push happens
• Combine the source files for more generic implementations
• Make assert_biteq! not rely on having Int in scope
• Add NEG_NAN to Float
• Correct the normalization of subnormals
• Add regression tests for subnormal issue
• Implement rounding for the hex float parsing and prepare to improve error handling
• Add assembly version of simple operations on aarch64
• Resolve small errors identified by recent clippy
• Replace calls to core::arch intrinsics with assembly
• Upgrade all dependencies to the latest
• Gate another assertion behind compiler-builtins
• Configure out remaining formatting when compiler-builtins is set
• Ignore unused variables when compiler-builtins is set
• Resolve monomorphization errors in compiler-builtins
• Make the compiler-builtins test more accurately mirror compiler-builtins
• Pin the nightly toolchain for aarch64 jobs
• Use git ls-files rather than manually globbing for tidy
• Make fma a trait method on Float
• fma refactor 3/3: combine fma public API with its implementation
• fma refactor 2/3: move math/generic/fma.rs to math/fma.rs
• fma refactor 1/3: remove math/fma.rs
• Scale test iteration count at a later point
• Add a way to print inputs on failure
• Rename Float::exp to Float::ex
• Check exact values for specified cases
• Add roundeven{,f,f16,f128}
• Fix parsing of negative hex float literals in util
• Increase allowed offset from infinity for ynf
• Add fminimum, fmaximum, fminimum_num, and fmaximum_num
• Combine fmin{,f,f16,f128} and fmax{,f,f16,128} into a single file
• Small refactor of bigint tests
• Eliminate the use of force_eval! in ceil, floor, and trunc
• Migrate away from nonfunctional fenv stubs
• Introduce a trait constant for the minimum positive normal value
• Implement u256 with two u128s rather than u64
• Pin the nightly toolchain for i686-pc-windows-gnu
• Increase the tolerance for jn and yn
• Replace an assert! with debug_assert! in u256::shr
• Change how operators are black_boxed
• Add simple icount benchmarks for u256 operations
• Decrease the allowed error for cbrt
• Port the CORE-MATH version of cbrt
• Add an enum representation of rounding mode
• Work arouind iai-callgrind apt failures
• Uncomment some hex float tests that should work now
• Convert fmaf to a generic implementation
• Remove or reduce the scope of allow(unused) where possible
• fix exponent calculation for subnormals
• Check more subnormal values during edge cases tests
• Run standard tests before running integration tests
• Add better edge case testing for scalbn
• Add fmaf128
• Make it possible to use hf32! and similar macros outside of libm
• Improve tidy output
• Add an integration test that verifies a list of cases
• Switch musl to track master
• Ensure zero has the correct sign
• Print the hex float format upon failure
• Commonize the signature for all instances of get_test_cases
• Start converting fma to a generic function
• Add checks via annotation that lists are sorted or exhaustive
• Do not add libm_helper.rs to the sources list
• Add a check in the shared.rs that the function list is sorted
• Add scalbnf16, scalbnf128, ldexpf16, and ldexpf128
• Fix hex float trait recursion problem
• Rename EXP_MAX to EXP_SAT
• Specify license as just MIT
• also print the hex float format for outputs
• Introduce a wrapper type for IEEE hex float formatting
• Support parsing NaN and infinities from the hf* functions
• Revert "Temporarily pin indicatif to 0.17.9"
• Temporarily pin indicatif to 0.17.9
• Switch musl from a script download to a submodule
• Ignore specific atan2 and sin tests on i586
• Rework the available Cargo profiles
• Remove remnants of the checked feature
• Use remquo from Rug
• Use frexp from Rug
• Use az exported from Rug
• Upgrade all dependencies to the latest version
• Enable missing icount benchmarks
• Add fmodf128
• Add way to override the number of iterations for specific tests
• Increase or set CI timeouts
• Add fmodf16 using the generic implementation
• Add a generic version of fmod
• Add fminf16, fmaxf16, fminf128, and fmaxf128
• Add a generic version of fmin and fmax
• Remove an outdated note about precision
• Add roundf16 and roundf128
• Add a generic version of round
• Add a generic version of scalbn
• Change from_parts to take a u32 exponent rather than i32
• Introduce XFAILs that assert failure
• Add hf16! and hf128!
• Fix the parsing of three-item tuples in util
• Add the ability to parse hex, binary, and float hex with util
• Add rintf16 and rintf128
• Add a generic version of rint
• Adjust ceil style to be more similar to floor
• Add floorf16 and floorf128
• Add a generic version of floor
• Add ceilf16 and ceilf128
• Add a generic version of ceil
• Make Float::exp return an unsigned integer
• Shift then mask, rather than mask then shift
• Add sqrtf16 and sqrtf128
• Copy the u256 implementation from compiler_builtins
• Port the most recent version of Musl's sqrt as a generic algorithm
• Enable force-soft-floats for extensive tests
• Don't set opt_level in the musl build script
• Add a retry to the musl download
• Remove trailing whitespace in scripts, run JuliaFormatter
• Ignore files relevant to benchmarking
• Add a way to ignore benchmark regression checks
• Run wall time benchmarks with --features force-soft-floats
• Run icount benchmarks once with softfloat and once with hardfloat
• Switch to the arm-linux runner and enable MPFR
• Remove the limit for querying a baseline
• Add an xfail for recent ynf failures
• Reduce the warm up and measurement time for short-benchmarks
• Run iai-callgrind benchmarks in CI
• Add benchmarks using iai-callgrind
• Provide a way to override iteration count
• Increase the CI timeout
• Adjust precision and add xfails based on new tests
• Replace HasDomain to enable multi-argument edge case and domain tests
• Add an override for a recent failure
• Pass --max-fail to nextest so it doesn't fail fast
• Slightly restructure ci/calculate-exhaustive-matrix.py
• Change .yml files to the canonical extension .yaml
• Use cargo-nextest for running tests in CI
• Simplify and optimize fdim (#442)
• Reduce indentation in run.sh using early return
• Don't set codegen-units=1 by default in CI
• Add fdimf16 and fdimf128
• Add a generic version of fdim
• Format the MPFR manual implementation list
• Disable util and libm-macros for optimized tests
• Add truncf16 and truncf128
• Add a generic version of trunc
• Add a utility crate for quick evaluation
• Enable build-mpfr and build-musl by default
• Rename the test-multiprecision feature to build-mpfr
• Introduce arch::aarch64 and use it for rint{,f}
• Use wasm32 arch intrinsics for rint{,f}
• Add a new precision adjustment for i586 exp2f
• Add a new precision adjustment for i586 rint
• Expose C versions of libm functions in the cb crate
• Always use the same seed for benchmarking
• Add biteq and exp_unbiased to Float
• Add a release-checked profile with debug and overflow assertions
• Remove ExpInt from Float, always use i32 instead
• Reorder tests in run.sh
• Split cast into cast and cast_lossy
• Use core::arch::wasm functions rather than intrinsics
• Add tests against MPFR for remquo and remquof
• Account for optimization levels other than numbers
• Make extensive tests exhaustive if there are enough iterations available
• Increase the allowed ULP for tgammaf
• Replace "intrinsic" config with "arch" config
• Don't use intrinsics abs for f16 and f128 on wasm32
• Remove an unused feature = "force-soft-floats" gate
• Switch from using unstable-intrinsics to intrinsics_enabled
• Increase the allowed precision for failing tests on i586
• Enable MPFR tests on i586
• Only update the github ref for pull requests
• Loosen precision on i586 based on new tests
• Add an override for failing ceil/floor tests on i586
• Add domain and edge case tests to musl
• Add test infrastructure for f16 and f128
• Add fabsf16, fabsf128, copysignf16, and copysignf128
• Enable f16 and f128 when creating the API change list
• Run extensive tests in CI when relevant files change
• Update precision based on failures from extensive tests
• Add extensive and exhaustive tests
• Add more detailed definition output for update-api-list.py
• Add tests against MPFR for ilogb and ilogbf
• Increase the precision for jn and jnf
• Rename unstable-test-support to unstable-public-internals
• Update precision based on new test results
• Rewrite the random test generator
• Add an iterator that ensures known size
• Streamline the way that test iteration count is determined
• Add a way for tests to log to a file
• Add tests against MPFR for scalbn{f} and ldexp{f}
• Add tests against MPFR for frexp and frexpf
• Add tests against MPFR for modf and modff
• Clean up integers stored in MpTy
• Sort ilogb with other precise operations
• Change to exhaustive matching for default_ulp
• Use intrinsics for abs and copysign when available
• Rename generic abs to fabs
• Always emit f16_enabled and f128_enabled attributes
• Add missing functions to the macro list
• Use rustdoc output to create a list of public API
• Forward the CI environment variable when running in Docker
• Remove lossy casting in logspace
• Set the allowed FMA ULP to 0
• Don't run push CI on anything other than master
• Use CheckCtx in more places
• Move CheckBasis and CheckCtx to a new run_cfg module
• Add ALL, from_str and math_op to Identifier
• Add new trait implementations for Identifier and BaseName
• Include shared.rs in libm_test::op
• Move the macro's input function list to a new module shared
• Add a way to plot the output from generators
• Update allowed precision to account for new tests
• Add tests for edge cases
• Add interfaces and tests based on function domains
• Introduce a float extension trait and some numerical routines
• Add an 8-bit float type for testing purposes
• Remove an is_nan workaround that is no longer needed
• Update and slightly refactor some of the Float trait
• Always enable unstable-float in CI
• Add f16 and f128 configuration from compiler-builtins
• Introduce generic abs and copysign
• Change from -latest to named CI images
• Allow Clippy lints in compiler-builtins-smoke-test
• Fix new clippy::precedence lints
• Replace string function name matching with enums where possible
• Rename associated type helpers, add OpITy
• Introduce helper types for accessing trait items
• Fix a bug in abs_diff
• Remove tests against system musl
• Use https: links in README.md
• Move some numeric trait logic to default implementations
• Change the multiprec_ prefix to mp_
• Change default ULP to use enum matching
• Rename Name to Identifier to avoid some ambiguity of "name"
• Change the CheckCtx constructor to take a Name enum
• Correct the proc macro to emit pub functions
• Rework tests to make use of the new MathOp trait
• Introduce a op module with struct representations of each routine
• Adjust how the proc macro emits types and add an enum
• Fix clippy lints in crates/ and enable this on CI
• Resolve clippy errors in libm tests and check this in CI
• Add some more basic docstrings (#352)
• Introduce hf32! and hf64! macros for hex float support
• Enable clippy for libm in CI
• Fix errors reported by Clippy in libm
• Replace libm_test::{Float, Int} with libm::{Float, Int}
• Expose the support module publicly with a test feature
• Update libm Float and Int with functions from the test traits
• Change prefixes used by the Float trait
• Check benchmarks in CI
• Remove libm-bench
• Add benchmarks against musl libm
• add support for loongarch64-unknown-linux-gnu
• Rename canonical_name to base_name
• Add float and integer traits from compiler-builtins
• Move architecture-specific code to src/math/arch
• Update select_implementation to accept arch configuration
• Add an "arch" Cargo feature that is on by default
• Vendor cfg_if::cfg_if!
• Update libm-test/build.rs to skip directories
• Rename the special_case module to precision and move default ULP
• Run tests against MPFR on CI where possible
• Add a test against MPFR using random inputs
• Create interfaces for testing against MPFR
• Combine the WASM CI job with the others
• Make use of select_implementation
• Introduce a select_implementation macro
• Introduce math::arch::intrinsics
• Replace feature = "unstable-intrinsics" with intrinsics_enabled
• Move the existing "unstable" feature to "unstable-intrinsics"

---

This PR was generated with release-plz.