[manaflow-cmux-ffi-release]

Pins CCmuxAppFFI to cmux-app-ffi-2e0b6acc359319c06eddc608d50e84895de41cc7 (checksum 62eb4a41502fd55f15b35bfd3e8be2056bfdfa01c4fe21eeff38d9a481923eba), published by run https://github.com/manaflow-ai/cmux/actions/runs/41782688580, which checked it with check-app-ffi-pin.sh --verify-release. A newer publish updates this branch.

Changelog

none

---

^{Open in [code]smith · Need help on this PR? Tag @codesmith-bot with what you need.}

---

Summary by cubic

Updates CCmuxAppFFI to cmux-app-ffi-2e0b6acc359319c06eddc608d50e84895de41cc7, published from the latest FFI release run, and updates the matching checksum in the Swift package manifest.

^{Written for commit 0b3ba36. Summary will update on new commits.}

[github-actions]

cmux status _{(one comment, edited in place; each line is updated by the job that owns it)}

Review 0b3ba36: HQ feed card · gallery before/after runs · checks

App of this head (app.zip, kept 1 day)

gh workflow run reload-build.yml -R manaflow-ai/cmux --ref feat-cmux-next -f tag=pr34995 -f ref=app-ffi-repin -f platform=macos -f backend_mode=local

Shipped in:

• cmux NEXT nightly 1.0.0-nightly.4182499138501 (e0aa712b3b)
• cmux NEXT nightly 1.0.0-nightly.4184087390301 (08052d331e)
• cmux NEXT nightly 1.0.0-nightly.4184224384601 (a3abfc4160)
• cmux NEXT nightly 1.0.0-nightly.4184409870901 (78ee0e108e)

[lawrencecchen]

Review (independent subagent, correctness first): SHIP at 0b3ba36
perf claim: none
R9: pass (no limit, queue or deadline is touched)

Reviewed at 0b3ba36: git ls-remote and the PR agree, and the PR is open. It is one commit by manaflow-cmux-ffi-release[bot] (committer: GitHub) on parent da12a52, which is on feat-cmux-next. I compiled nothing on this machine.

Findings

None.

1. The pin and the release

• The diff: only the url and checksum of CCmuxAppFFI in Packages/macOS/CmuxNext/Package.swift, which move from cmux-app-ffi-1f0abddf9f2f to cmux-app-ffi-2e0b6acc359319c06eddc608d50e84895de41cc7 and checksum 62eb4a41502fd55f15b35bfd3e8be2056bfdfa01c4fe21eeff38d9a481923eba. Feat has not touched Package.swift after the parent.
• The release assets, downloaded anonymously (no token):
  • SOURCE_SHA is 2e0b6ac.
  • SHA256SUMS has exactly the line 62eb4a41…eba CCmuxAppFFI.xcframework.zip.
  • The archive itself (15,745,362 bytes) hashes to the pinned checksum.
  • The release is by manaflow-cmux-ffi-release[bot], and its tag points at commit 2e0b6ac.
  • The four C headers in the archive are byte-equal to the source's at 2e0b6ac.
• Release run 41782688580: all three attempts ran on head 2e0b6ac (push to feat).
  • build xcframework ran once, in attempt 1. Later attempts reused it, so there is one archive.
  • Attempt 1's publish uploaded the assets (06:19:01Z), then failed its read-after-write step "The release exists with this run's archive" with release … does not exist (a lag right after the create).
  • Attempt 2's publish passed that step against the same build's archive. Nothing was overwritten (the step never overwrites).
  • Attempt 3's re-pin printed app FFI pin: sources match 2e0b6acc… and app FFI pin: release built from 2e0b6acc… with checksum 62eb4a41…, then pushed 0b3ba36 to app-ffi-repin.

2. What the new binary changes

• The inputs: check-app-ffi-pin.sh --list-inputs gives 16 inputs. From 1f0abdd to 2e0b6ac they change in one commit, focus-core: a window whose focus waits for another tab is unsettled in the Rust world check (P0 INVERSE-KEY rule; A12 step A, R-INPUT S2) #34641: cmux-focus-core/src/invariants.rs and its test.
• The inputs at this PR's parent da12a52 and at the feat tip equal 2e0b6ac's, so the pin is current. R-INPUT S2a: the input router reduces each window's focus as the app's focus handle does (WindowFocus), Rust only #34929, which changes cmux-focus-core and the FFI crate, has not landed.
• No header changed between the two pins, so the C ABI is the same.
• The changed code is unreachable from the app: it is invariants::world and is_settled (the new awaits_other_tab), and its only route is the FFI op focus_input_world. No Swift code on this head or on the current feat tip calls focus_input_world; the Swift InputInvariants.world is still the pure-Swift check, until Swift calls cmux-focus-core's focus rules; the Swift copies go (A12 step B, carried from #31835; R-INPUT S2) #34685. So the app's behavior does not change with this pin.

3. CI on this head

• cmux-next app compile (build fleet) passed: fleet job 4d4696635b279362d67ad283 at this sha (log read) copied libcmux_app_ffi.a, linked cmux-next-exe ("Build complete! (73.30s)"), and its cmux-tui check is ok.
• Also passed: cmux-next fleet Release compile, cmux-next/compile and pinned assets exist in store.
• cmux-next checks fails on three steps, all base findings in files this PR does not touch:
  • check-scrollbars (Swift and CSS files over the ceiling, the known base red).
  • The fixture host-name test: webviews/src/agent-session/acpmux/subagents/SubagentTray.gallery.ts:157 names the fleet host glaeda-gh. The same line is at this head and at the feat tip fb9c908.
  • tests/test_test_reds.py: test_the_committed_contract_is_the_workflows errors in scripts/ci/base_red_excuse.py derive_rollups.
• Not verified, still pending when I posted: cmux-next fast tier, cmux-next swift test, cmux-next daemon tests, ios-tests and release-notary. The pin cannot change their outcome through the ABI or behavior (same headers, no Swift caller of the changed code). Linking is proven by the app compile.

[lawrencecchen]

REWRITE R-INPUT: this re-pin carries #34641 (A12 step A). Independent review: SHIP at 0b3ba36 (#34995 (comment)). The app compile linked the new libcmux_app_ffi.a and the fleet Release compile passed.

The red check on this head, cmux-next checks, has three base findings in files this PR does not touch (it changes only the pin lines of Packages/macOS/CmuxNext/Package.swift):

• check-scrollbars: 12 hits, over the ceiling of 10 (also red on focus-core: a window whose focus waits for another tab is unsettled in the Rust world check (P0 INVERSE-KEY rule; A12 step A, R-INPUT S2) #34641 and R-INPUT S2a: the input router reduces each window's focus as the app's focus handle does (WindowFocus), Rust only #34929).
• The fleet host name glaeda-gh in SubagentTray.gallery.ts:157: the same on the feat tip.
• test_test_reds.py (base_red_excuse / derive_rollups).

No Swift code on feat calls focus_input_world yet, so the app's behavior does not change before #34685.