[batpigandme]

---

type: pre_commit_static_analysis_report
description: Results of running static analysis checks when committing changes.
report:

• task: lint_filenames
  status: passed
• task: lint_editorconfig
  status: passed
• task: lint_markdown_pkg_readmes
  status: na
• task: lint_markdown_docs
  status: na
• task: lint_markdown
  status: na
• task: lint_package_json
  status: na
• task: lint_repl_help
  status: na
• task: lint_javascript_src
  status: passed
• task: lint_javascript_cli
  status: na
• task: lint_javascript_examples
  status: na
• task: lint_javascript_tests
  status: na
• task: lint_javascript_benchmarks
  status: na
• task: lint_python
  status: na
• task: lint_r
  status: na
• task: lint_c_src
  status: na
• task: lint_c_examples
  status: na
• task: lint_c_benchmarks
  status: na
• task: lint_c_tests_fixtures
  status: na
• task: lint_shell
  status: na
• task: lint_typescript_declarations
  status: passed
• task: lint_typescript_tests
  status: na
• task: lint_license_headers
  status: passed

---

Description

What is the purpose of this pull request?

This pull request:

• Adds shared inputs for math/base/special/log1p (10 regimes of 500 points each, plus 19 edge cases), drawn by scripts/inputs.js from a seeded PRNG so they reproduce byte for byte.
• Adds fixture runners that evaluate those inputs with MPFR (the reference), Julia, Python's math and SciPy, R, Boost, Cephes, and stdlib's own C. Each set's meta records the library, its version, what it calls underneath, and the platform.
• Adds the 88 sets they generate, all from one x86_64 Linux machine with glibc 2.39.
• Adds scripts/accuracy.js, which checks that every set echoes the shared inputs bit for bit, then scores each source against MPFR by regime.
• Tests the implementation against MPFR in test.js, moves the existing Julia tests unchanged to test.julia.js, and adds test.scipy.js.

Scores against MPFR on the 5,000 regime points:

Source	Exact	1 ULP	2 ULP
stdlib (JavaScript and C, identical)	4,800	200	0
Python math, R, Boost (glibc)	4,800	200	0
Julia	4,995	5	0
SciPy	4,697	212	91
Cephes	4,590	319	91

Related Issues

Does this pull request have any related issues?

This pull request has the following related issues:

• [RFC]: Migrate math/base/special packages from relative tolerance testing to ULP difference testing (tracking issue) #11352 (bounds follow its "minimum required ULP" rule)
• build: upgrade Boost & compile C++ test fixture runners w C++14 #16545 (Boost runner builds with C++14)

Reference Packages

Did the content of this pull request follow one or more existing packages? Answering is optional.

This pull request followed these packages:

• math/base/special/fresnel (Cephes runner), math/base/special/digamma (non-finite values stored as strings)

What carried over from the reference(s)?

• Runner layout under test/fixtures/<source>/, and calling Cephes directly.

What differed, or what did you have to work out yourself?

• Shared inputs read by every runner, and a scorer that checks the echo before scoring.
• c/stdlib/ builds the package's C from its manifest.json, as compile_c_benchmark does for benchmarks; no fixture runner did this before.
• The two C runners read JSON numbers with strtod, checked exact on 205,019 values on macOS and x86_64 Linux.

Questions

Any questions for reviewers of this pull request?

• Should every transcendental package get this test layout, or only those whose current reference is inaccurate?
• Non-finite values are stored as "Infinity" and "NaN"; digamma uses "Inf". Which spelling should be standard?
• Should scripts/accuracy.js stay per package until a few packages use it, or move to a shared tool now?

Other

Any other information relevant to this pull request? This may include screenshots, references, and/or implementation notes.

• Each bound is the tightest that passes on the committed fixtures, per regime: 1 in the regimes holding stdlib's 200 points at 1 ULP, and 0 elsewhere in test.js. test.scipy.js compares against SciPy, so it reaches 2 where SciPy is 2 ULP off. The bounds describe these fixtures, not the function: sampled at 10^6 points per regime, stdlib also reaches 1 ULP in small_negative and small_positive (about 0.02% of points), where all 500 fixture points are exact.
• Separately, stdlib reaches 2 ULP in a sliver of about 3.4 × 10^9 doubles just below √2/2 − 1, where its threshold test differs from FDLIBM's, so the comment "the error is always less than 1 ulp" (lib/main.js, src/main.c) doesn't hold. That's left for its own fix.
• test.native.js is unchanged.

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

AI Assistance

When authoring the changes proposed in this PR, did you use any kind of AI assistance?

• Yes
• No

If you answered "yes" above, how did you use AI assistance?

• Code generation (e.g., when writing an implementation or fixing a bug)
• Test/benchmark generation
• Documentation (including examples)
• Research and understanding

Disclosure

If you answered "yes" to using AI assistance, please provide a short disclosure indicating how you used AI assistance. This helps reviewers determine how much scrutiny to apply when reviewing your contribution. Example disclosures: "This PR was written primarily by Claude Code." or "I consulted ChatGPT to understand the codebase, but the proposed changes were fully authored manually by myself.".

The runners, scorer, and tests were written by Claude Code (Opus 5.5), and the fixtures generated in a Claude Code session on x86_64 Linux. I set the design, reviewed every file, and checked the scores against independent reruns. This follows (to the best of my ability) the layout for "accuracy-comparison" tables across languages (AKA the accuracy-visibility project) as discussed with @kgryte.

@stdlib-js/reviewers

[stdlib-bot]

Coverage Report

Package	Statements	Branches	Functions	Lines
math/base/special/log1p	$\\color{red}523/525$ $\\color{green}+99.62\\%$	$\\color{red}41/42$ $\\color{green}+97.62\\%$	$\\color{green}3/3$ $\\color{green}+100.00\\%$	$\\color{red}523/525$ $\\color{green}+99.62\\%$

The above coverage report was generated for the changes in this PR.

[batpigandme]

"Lint Copyright Years" failure is expected. It flags test/test.julia.js ("Expected 2026, found 2018"), which holds existing Julia-fixture tests from test/test.js, moved over unchanged. Since it's moved code, it keeps the original 2018 year.

Followed precedence from other moved files (e.g., #13218 and #12964), which failed this check the same way when code moved into stats/strided.

The warnings for the fixture JSON files are only "no copyright notice found," since JSON fixtures don't carry one.