Skip to content

chore: bump out-of-workspace consumers to the published v9.1.0 wave - #857

Merged
fishloa merged 2 commits into
mainfrom
chore/post-publish-consumer-bumps
Jul 30, 2026
Merged

fishloa merged 2 commits into
mainfrom
chore/post-publish-consumer-bumps

Conversation

@fishloa

@fishloa fishloa commented Jul 30, 2026

Copy link
Copy Markdown
Owner

Follow-up to the v9.1.0 release wave, as flagged in docs/release-notes/v9.1.0.md under Post-publish follow-up.

acap-multimux, transmux-py and dvb-si-py sit outside the cargo workspace and resolve their dependencies from crates.io by version, not by path. They could not be bumped in the release PR itself — the versions did not exist yet, so cargo metadata could not resolve and their jobs failed by construction (six red checks on #854 before the bumps were pulled out).

Now that the wave is live:

consumer change
acap-multimux transmux 0.20 → 0.21
bindings/transmux-py transmux 0.20 → 0.21, crate 0.1.0 → 0.2.0
bindings/python (dvb-si-py) crate 0.1.0 → 0.2.0 (tracks lockstep 9.1.0)

Why this PR is the real publish verification

These are the only things in the repo that resolve against the registry rather than local paths. Every other gate builds from the workspace, which is exactly why the transmux-v0.21.0 publish failure (a dev-dep cycle, fixed in #856) got through CI untouched.

Green here means the published artifacts actually resolve and build for a consumer.

fishloa added 2 commits July 30, 2026 14:55
acap-multimux, transmux-py and dvb-si-py sit outside the cargo workspace and
resolve their dependencies from crates.io by version rather than by path, so
they could not be bumped in the release PR — the versions did not exist yet and
their CI was unsatisfiable by construction.

Now that the wave is live:

- acap-multimux: transmux 0.20 -> 0.21
- transmux-py:   transmux 0.20 -> 0.21, crate 0.1.0 -> 0.2.0
- dvb-si-py:     crate 0.1.0 -> 0.2.0 (tracks lockstep 9.1.0)

This PR's CI is the end-to-end verification of the publish: these are the only
things in the repo that resolve against the registry rather than local paths, so
green here means the published artifacts actually resolve and build.
acap-multimux keeps its own Cargo.lock and CI runs `cargo test --locked`, so
bumping the transmux requirement without regenerating the lock fails before it
compiles anything.

The regeneration pulls the whole wave in one resolution — transmux 0.20->0.21,
ll-hls-runtime 0.2->0.3, media-plane 0.1.0->0.1.1, multimux 0.5.0->0.5.1,
rtmp-runtime 0.2->0.3 — with no version conflict, which is the mixed-major
check from #824 verified against the registry rather than asserted in a doc.

Verified: `cargo test --locked` in acap-multimux, 18 passed.
@fishloa
fishloa merged commit 94ccca3 into main Jul 30, 2026
22 checks passed
@fishloa
fishloa deleted the chore/post-publish-consumer-bumps branch July 30, 2026 17:54
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant