Skip to content

chore(deps): termlens 0.11.2, and the report action with it - #103

Merged
vyncint merged 2 commits into
mainfrom
chore/termlens-0.11.2
Sep 19, 2026
Merged

vyncint merged 2 commits into
mainfrom
chore/termlens-0.11.2

Conversation

@vyncint

@vyncint vyncint commented Sep 19, 2026

Copy link
Copy Markdown
Owner

Two releases behind since the 09-11 migration. Both patches, so "0.11" is unchanged and only Cargo.lock moves.

The thing to check here was DASHBOARD_UNSUPPORTED. 0.11.2 makes unsupported() report more (OSC 4/10/11 sets, the xterm title stack) and re-spells a parameterless sequence ^[[s instead of ^[[0s — an exact pin is precisely what that breaks. It holds: the dashboard emits neither family, and ^[[59m carries a parameter so the re-spelling does not reach it. tests/emulation.rs and tests/report.rs pass unchanged.

Also arriving: the pending-wrap cursor moves off col == cols (#401).

Checked with what CI runs: cargo test --workspace (58 result lines, 0 failures), -p oxmera -p oxmera-runtime --no-default-features, and -p oxmera-cli --test termlens_cli -- --ignored (4/4). fmt and clippy clean.

Cut from main; release/0.5.2 is three commits ahead and untouched by this.

Locked at 0.11.0 since the 09-11 migration, two releases behind. Both are
patches, so the `"0.11"` requirement is unchanged and only the lock moves.

The one to check here was `DASHBOARD_UNSUPPORTED`. 0.11.2 makes
`unsupported()` report more — the set forms of OSC 4/10/11 (#392) and the
xterm title stack (#393) — and re-spells a parameterless sequence as
`^[[s` rather than `^[[0s` (#394), so an exact pin is exactly what such a
change breaks. This one holds: the dashboard emits neither family, and
`^[[59m` carries a parameter so the re-spelling does not reach it.
`tests/emulation.rs` and `tests/report.rs` pass unchanged, which is the
assertion doing its job rather than the pin being loose.

Also arriving: the pending-wrap cursor is reported on the last cell rather
than one column past it (#401), and the JSON reader no longer refuses a
screen the text parser would accept.

The two `report@` pins move from v0.11.0 to v0.11.2 — the action's own
script, which gained `render --out` in v0.11.1 so a failed render leaves
no truncated file in the step summary.

Verified with what CI runs: `cargo test --workspace`, `-p oxmera
-p oxmera-runtime --no-default-features`, and `-p oxmera-cli --test
termlens_cli -- --ignored` — the suite a plain `cargo test` skips. fmt and
clippy clean.

Signed-off-by: Vyncint Ng <115854244+vyncint@users.noreply.github.com>
`check-skill-version.sh` failed on the bump: `PIN_TERMLENS` is compared
with the *resolved* version in Cargo.lock, exactly rather than by
major.minor, because that is what the Monday pins watch compares with
crates.io. Left at 0.11.0 against a lock resolving 0.11.2, the watch would
have opened a drift issue for a bump that had already happened — which is
#68, the reason the check covers the whole list instead of one entry.

The two `cli-version:` literals go with it. The gate only compares those
by major.minor, so 0.11.0 passed and would have kept passing; but the
comment above each says it names "the version the lockfile names, so the
renderer and the harness are one release", and the lockfile now names
0.11.2. A literal that contradicts its own stated purpose is the drift
this check exists to stop, one release before the gate would notice.

The vendored skill and CONTRIBUTING's line are compared by major.minor and
are unaffected: a patch does not rewrite the skill.

`./.github/scripts/check-skill-version.sh` passes on all four.

Signed-off-by: Vyncint Ng <115854244+vyncint@users.noreply.github.com>
@vyncint
vyncint merged commit 6343325 into main Sep 19, 2026
15 checks passed
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