neuwirth
/
pmacs
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

docs: two unresolved macOS signatures from #229 CI, as separate rows 

PR #229 first CI run went red on Test (macos-latest / lua54) with
a_pty_resize_blanks_the_host_before_repainting. The rerun turned that
selector green in 0.50s against 20.26s failing, and went red on a
DIFFERENT selector, ctrl_c_during_reconnect_sleep_yields_clean_exit.
Per this file matching rule that is a new incident, not the resize
signature occurring twice, so they are two rows.

U4 records the resize failure with its exact fragments. The diff is
excluded on two structural grounds that do not depend on a rerun: #229
touches no src/ at all and no test but gate_script_acceptance, and the
sibling luajit leg passed on the same commit. A deterministic platform
defect fails both flavours --- that is how #227 non-UTF-8 fixture
presented.

U5 records the Ctrl-C failure at deliberately WEAKER exclusion
strength, and says so in its own field. The changed gate suite ran
earlier in the same job and creates worktrees and directories. No
leaked child or persistent signal-state mutation was observed, but "the
diff touches no src/" is not the same argument here as for U4, because
cross-suite leaked state is a path reachability reasoning does not
close. Its control is to run m5_8_acceptance alone, without the gate
suite ahead of it, before attributing anything either way.

Neither row claims a mechanism. The exit status shows only that Ctrl-C
arrived as SIGINT rather than as the raw-mode key event the test
drives; whether injection preceded raw mode, raw mode was lost, or
something else happened is open, and the fragment does not separate
them.

Also flagged: worker-identity-stage1 independently defines its own U4
and U5. This lane merges first, so that branch must renumber on rebase
--- a conflict resolved textually without renumbering would leave two
different incidents sharing an id, which is the failure the matching
rule exists to prevent.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016bqGA6s9tTUFzYpbeW3tai
This commit is contained in:
Levi Neuwirth 2026-08-09 23:30:53 +02:00
parent 93d557a527
commit ae6a815435
No known key found for this signature in database
2 changed files with 51 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
docs/active-work.md
View File

@ -267,6 +267,21 @@ census.
## `scripts/gate --protocol` build step — IMPLEMENTED at `49bc141`, RE-OPENED by review, witness CLOSED at `677fd25`. No PR yet
**PR #229 OPEN** — https://github.com/levineuwirth/pmacs/pull/229,
opened at `93d557a`. **Held, not merged.** Its first CI run went red on
`Test (macos-latest / lua54)`; the rerun turned that selector green and
went red on a **different** one. Both are recorded as **U4** and **U5**
in `docs/ci-red-signatures.md`, as separate incidents per the matching
rule rather than one signature twice.
**Registry numbering conflict, expected at merge.**
`worker-identity-stage1` independently defines its own **U4** and **U5**
on its branch. This lane merges first, so on `main` the rows above are
U4/U5 and **worker identity must renumber its pair on rebase**. Flagged
here because a rebase that resolves the textual conflict without
renumbering produces two different incidents sharing an id, which is
exactly the failure the registry's matching rule exists to prevent.
**Written with the lane's first commit**, per the standing correction
from #171 and #215.

36
docs/ci-red-signatures.md
View File

@ -566,3 +566,39 @@ The #214 occurrence is the strongest available evidence that these are
not caused by the PRs they appeared on — that PR is **docs-only and its
tree is byte-identical to a green `main`**. It is not evidence that any
of them is harmless.
### U4 — `a_pty_resize_blanks_the_host_before_repainting`, macOS `lua54`, one occurrence
Surfaced on PR #229's CI. **The diff is excluded on two independent
structural grounds, neither of which is a rerun**: #229 touches no
`src/` at all and no test but `tests/gate_script_acceptance.rs`, and the
sibling `macos-latest / luajit` leg **passed on the same commit**. A
deterministic platform defect fails both flavours — that is how #227's
non-UTF-8 fixture presented.
| field | value |
|---|---|
| **selector** | `--test full_grid_resync_acceptance a_pty_resize_blanks_the_host_before_repainting` |
| **job / flavor** | GitHub Actions, `Test (macos-latest / lua54)`, `macos-26-arm64`. **NOT** seen on the `luajit` leg of the same commit |
| **required fragments** | `FG-INV: the post-resize resync must blank the host` · `no CSI 2 J appeared in the 25362 bytes emitted after the first painted frame` · `panicked at tests/full_grid_resync_acceptance.rs:126` |
| **status** | **one occurrence; INTERMITTENT — passed on rerun in 0.50s against 20.26s failing** |
| **what IS established** | the assertion saw ~25 KB of post-frame output with no `CSI 2 J` in it, and the suffix head shows ordinary repaint traffic (`ZQXMARKERQZ` rows with SGR + CUP), so the host was *painting* — it just never emitted the blank. The rerun's 0.50s versus 20.26s is a 40× spread, which is itself the signal |
| **what is NOT** | any mechanism. Whether the blank was never emitted, emitted before the capture window, or lost in transport is open. The green rerun establishes **intermittence only**, per this file's rerun rule |
| **discriminating control for the next occurrence** | capture the **full** post-frame byte stream rather than the 25 KB window, with timestamps, so "never emitted" separates from "emitted outside the window". The 40× duration spread suggests a readiness/timing interaction worth timing directly |
### U5 — `ctrl_c_during_reconnect_sleep_yields_clean_exit`, macOS `lua54`, one occurrence
Surfaced on the **rerun** of PR #229's failed job — a *different*
selector from U4, so by this file's matching rule it is a **new
incident, not U4 occurring twice**.
| field | value |
|---|---|
| **selector** | `--test m5_8_acceptance ctrl_c_during_reconnect_sleep_yields_clean_exit` |
| **job / flavor** | GitHub Actions, `Test (macos-latest / lua54)`, rerun attempt 2 |
| **required fragments** | `Ctrl-C during reconnect sleep should produce a clean exit` · `ExitStatus { code: 1, signal: Some("Interrupt: 2") }` · `panicked at tests/m5_8_acceptance.rs:546` |
| **status** | **one occurrence, unresolved** |
| **what IS established** | Ctrl-C reached the process **as `SIGINT`** rather than as the raw-mode key event the test drives. That is all the exit status shows |
| **what is NOT** | whether injection preceded raw mode, raw mode was lost, or something else. Three mechanisms remain open and the fragment does not separate them |
| **exclusion strength — WEAKER than U4's, deliberately** | #229's changed `gate_script_acceptance` ran **earlier in the same job**, and it creates worktrees and directories. No leaked child or persistent signal-state mutation was observed, but "the diff touches no `src/`" is **not** the same argument here as it is for U4, because cross-suite leaked state is a path that reasoning about reachability does not close. Recorded at this strength on purpose |
| **discriminating control for the next occurrence** | run `m5_8_acceptance` **alone** on macOS `lua54`, without the gate suite ahead of it, before attributing anything to either |