ffmeta 0.1.1

0.1.1 - 2026-10-10

Build

• hatchling 1.27.0 at least, the first writing PEP 639's metadata
• hatchling's floor said exactly -- 2.4 is 1.27.0's default

Tests

• pytest: strictness as strict = true, which pytest 9.0 honours

ffman 0.1.1

0.1.1 - 2026-10-10

Features

• image: ffman's release publishes its image to GHCR -- nixpkgs' ffmpeg, no FDK

Fixes

• subverter: JSON nested past 1000 levels is refused alike on every machine
• ffman: a stopped job exits 128 + signum on macOS too, its zombie group refused by killpg
• ffman: a WebVTT transcript's character references decoded: A & B burns as A & B

Refactoring

• ffman: a trial's tags read through probe.py, narrowed as every ffprobe answer

Build

• nix: the image as an archive and a stream; each Python's types as itself
• hatchling 1.27.0 at least, the first writing PEP 639's metadata
• release: refuse an untagged current version and an ffman ahead of its libraries; push the release's tag alone
• hatchling's floor said exactly -- 2.4 is 1.27.0's default
• ffman: subverter 0.1.1 at least, whose markup.decoded ffman imports

Documentation

• ffman: what the VHS noise ceiling and chunk-word's fifth are

Tests

• ffman: the depth and GIF palette checks free of arm64's colour rounding
• the review's points -- a byte order read wrong shows, names and comments exact
• pytest: strictness as strict = true, which pytest 9.0 honours
• ffman: frames()' md5 declared as no security
• ffmeta's oracle tools held to ffman's by every name they carry

subverter 0.1.1

0.1.1 - 2026-10-09

Features

• subverter: WebVTT's character references: a reading says so, decoded decodes a run

Fixes

• subverter: JSON nested past 1000 levels is refused alike on every machine

Build

• hatchling 1.27.0 at least, the first writing PEP 639's metadata
• hatchling's floor said exactly -- 2.4 is 1.27.0's default

Tests

• the review's points -- a byte order read wrong shows, names and comments exact
• pytest: strictness as strict = true, which pytest 9.0 honours