Skip to content

Commit 0289fe1

Browse files
committed
text dump checkpt in changelog
1 parent 8a821df commit 0289fe1

1 file changed

Lines changed: 17 additions & 0 deletions

File tree

‎CHANGELOG.md‎

Lines changed: 17 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -11,6 +11,23 @@ Work on the current development branch. Entries move to a versioned section on
1111
release.
1212

1313
### Added
14+
- Linux promoted from experimental to hardware-verified. Verification run
15+
2026-09-19 on Debian 12 (bookworm, kernel 6.12.95, Dell Latitude 7420)
16+
with `hackrf` 2022.09.1 and firmware 2024.02.1: full test suite including
17+
all hardware tests, 227/227, cross-checked the same day on Windows with
18+
hardware at 227/227 on identical bytes. Per the platform policy, Linux
19+
re-enters the CI matrix under the same standards as Windows (85% coverage
20+
gate, hardware evidence for core changes), which also returns the Linux
21+
`pdeathsig` dead-man path to regression testing. The verification itself
22+
surfaced the two process-lifecycle bugs fixed below (frozen-writer USB
23+
claim leak; open-retry race) -- the strongest possible argument for
24+
requiring it. `SETUP_LINUX.md` (Debian-family setup + the verification
25+
sequence) is now in the repo; platform language updated in the READMEs
26+
and CONTRIBUTING; macOS remains experimental pending its own board run.
27+
- Housekeeping: `hackrfpy/tests/fm_testdata/` outputs are regenerable and
28+
were never meant to be committed; the `.gitignore` entry pointed at the
29+
wrong path (`tests/fm_testdata/`), so three generated files were tracked.
30+
Path fixed, files untracked.
1431
- Type annotations across the entire shipped package, and `mypy` promoted to a
1532
blocking CI gate (`disallow_untyped_defs`). The package has always shipped a
1633
`py.typed` marker, which tells downstream type-checkers the inline annotations

0 commit comments

Comments
 (0)