Repository navigation
ci: boot the suites a pull request affects before it merges - #1142
Conversation
e2e.yml only runs when something calls it, and contributor agents may not dispatch workflows, so a VM-related change got its first VM run after merge (as a v1 hold) or when a maintainer started a manual run. On 2026-11-10 every VM question (#827, #838, #908, #1138, #1139) waited on a person starting a run. Add "PR VM check": on every pull request, plan the PR's own changes with the v1 gate's planner (vm_validation_plan.py + v1_gate.py) and, when a VM is needed, boot those suites through e2e.yml on dakota:testing pinned by digest, testing the merge commit. Docs-only and tooling PRs boot nothing; a change the gate cannot boot (the gnome-e2e composite action) gets a notice. Advisory, not required. This PR changes the workflow itself, so it boots common and smoke as its own proof. Assisted-by: Claude Opus 5.5 via Claude Code Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WcWm7qQa538Zjcuvv8jBtn Signed-off-by: Danathar <Danathar@users.noreply.github.com>
9d26e0c
ℹ️ Sentinel notice — informational; author is trusted (owner), merge is not blockedHive flagged this PR (author @Danathar, head
Hive did not add the |
|
@Danathar this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (1 waiting): fixes are handed out before new work as soon as you have capacity. |
|
Converted to draft: this PR's own run booted smoke on dakota:testing and found 6 pre-existing failures (power-status-color extension missing from the image; geoclue/nfs-idmapd/rpc-statd failing at boot). The v1 gate tolerates them via its v1 baseline, but a per-PR check would turn every smoke-touching PR red, and the hive won't merge with a red check. Waiting on #1143; this comes out of draft once smoke is green on dakota:testing. |
There was a problem hiding this comment.
Summary: The new guidance sends contributors to a manual test run when automatic testing cannot cover their change. That manual run cannot exercise the separate desktop-testing tool, so following the instruction can still leave the changed code untested.
What I need from you: Correct the manual-validation instructions for changes the automatic check cannot cover.
docs-currency
candidate-vm-validation.md:61–65: The new paragraph recommends the manual run below for hold, including changes to the gnome-e2e composite action. But manual.yml:36–49 only calls e2e.yml, which never executes that action (vm_validation_plan.py:22–26). Point those changes to a workflow that actually invokes the action:
-skipped there. Use a manual run (below) for another image, or for a `hold`.
+skipped there. Use a manual run (below) for another image. For a `hold`
+caused by the `gnome-e2e` composite action, use a workflow that calls that
+action at the PR ref. For a suite without a hosted lane, use its required
+hardware or dedicated validation workflow; the manual run below does not
+supply the missing coverage.No findings from: correctness, security, intent-alignment, style.
Confidence: 3/5 (needs attention) — 1 medium finding; 1 perspective requested changes
— hive: agent=reviewer backend=copilot model=gpt-6.1-sol copilot=1.0.88
The manual run calls e2e.yml, so it cannot cover a hold. Point composite_action changes (the gnome-e2e action and the extension validation gate it runs) at gnomeos-canary.yml on a branch at the PR's head, and no_hosted_lane suites at their own hardware or a hand check. Also drop "advisory": the hive merges no PR while any check is red. Addresses the hive reviewer's docs-currency finding on #1142. Assisted-by: Claude Opus 5.5 via Claude Code Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01WcWm7qQa538Zjcuvv8jBtn Signed-off-by: Danathar <Danathar@users.noreply.github.com>
|
Merged by hive contributor-lane auto-merge: CI green for ≥30m, reviewer verdict approve, @Danathar was auto-promoted after 5 clean merged PRs in this hive's repositories. |
Summary: adds "PR VM check", which boots the test suites a pull request affects before it merges. Today
e2e.ymlonly runs when something calls it, and contributor agents aren't allowed to start workflows. So a VM-related change gets its first VM run either after merge (as av1hold) or when a maintainer starts a manual run by hand. On 2026-11-10 every VM question (#827, #838, #908, #1138, #1139) waited on a person starting a run.How it works
v1gate uses (scripts/vm_validation_plan.py, thenscripts/v1_gate.py).move: docs, unit tests, repo tooling. Boots nothing; done in seconds.gate: boots the planned suites throughe2e.ymlondakota:testing, pinned by digest, withtest_ref= the merge commit.hold: a change the gate can't boot (for example, thegnome-e2ecomposite action). Posts a notice and boots nothing; the skill doc says how to validate it instead (gnomeos-canary.ymlfor the action, a hand check for suites with no hosted lane).Safety and cost
pull_requesttrigger (notpull_request_target),contents: readat the top. The boot job grantspackages: writeonly becausee2e.yml's jobs ask for it (a called workflow can't be given less), for the desktop-screenshot push.continue-on-error) is skipped there. No secrets are used.e2e.ymlchange (a release file), so it's left as a follow-up.concurrencyper PR).Red means this PR broke something
The check is strict: any failing scenario turns it red, and the hive won't merge a PR with a red check, required or not (
FailOnAnyFailedCheckin its merge gate). So it only works ifmainitself is green on the image:dakota:testing(e2e: smoke-b fails on projectbluefin/dakota:testing, also at v1: GNOME Settings smoke tests :: Power settings panel is accessible #1153, the Power panel) was fixed by fix(smoke): skip Power settings on virtual machines without power hardware #1158, merged 11-11 21:51Z. Its smoke run from the PR branch was green: https://github.com/projectbluefin/testsuite/actions/runs/43102230803v1gate files every failure it accepts as pre-existing as atriage/acceptedissue for the hive to skip in VMs, quarantine or fix.A PR that runs into a new image failure before the hive handles it will be red until it is. That trade was chosen over a baseline comparison here, because each
e2e.ymlVM job is its own check and a comparison job can't turn those green.Proof
This PR changes the workflow itself, which the planner maps to the full lane set. So its own run boots
commonandsmoke(lifecyclestays ungated, as at the gate). Planned locally:gate(common, smoke);move;tests/common/...feature→gate(common);.github/actions/gnome-e2e/action.yml→hold.actionlint is clean. zizmor's only note is the
./vs$/self-reference style; every other workflow here uses./.🤖 Generated with Claude Code
https://claude.ai/code/session_01WcWm7qQa538Zjcuvv8jBtn