Skip to content

fix(e2e): settle udev before bootc install to release the post-merge v1 gate - #5

Open
kylerankin wants to merge 247 commits into
mainfrom
fix/e2e-devsettle-v1gate
Open

kylerankin wants to merge 247 commits into
mainfrom
fix/e2e-devsettle-v1gate

Conversation

@kylerankin

Copy link
Copy Markdown
Owner

Fixes projectbluefin#1123

bootc 1.17 (Dakota since 2026-11-01) reads the new root partition's type from udev ~200 ms after sfdisk. On GitHub-hosted runners udev accumulates a backlog and has not always processed loopNp3 by then, so bootc install to-disk --via-loopback aborts with root partition 3 has type <none> before writing any deployment. The setup step then records no scenario, so the v1 gate holds the tag with "no failing scenario recorded" for the common, kde-smoke and smoke suites.

The retry added in projectbluefin#1125 settled udev after the loopback was released and before the new partition existed, so it never helped the partition bootc creates — 6 of the last 7 runs (run 42807521428 @ 514c893, all 3 attempts) still failed identically. This is a runner/environment failure, not a test-code regression: candidate and baseline fail the same way on the same image digest.

This change moves the settle to the top of each retry attempt, before bootc creates the partition:

# before bootc install to-disk --via-loopback
sudo udevadm trigger --subsystem-match=block --action=change
sudo udevadm settle --timeout=30 || true

and removes the now-redundant post-release settle. The matching skill note in docs/skills/ci-ops/e2e-workflow/SKILL.md documents the environment failure.

Verification note: the full VM gate cannot run locally. Evidence is the root-cause analysis above plus upstream CI once this lands.


🐝 Hive Agent: contributor | SHA: 88a74d06

castrojo and others added 30 commits September 16, 2026 19:39
Keep the stateful Firefox journey within one scenario, resolve live accessibility nodes for every launch and retry, and ignore the QEMU-only TuneD first-boot timeout.

Assisted-by: GPT-5.6 via GitHub Copilot

Co-authored-by: castrojo <castrojo@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Assisted-by: GPT-5.6 via GitHub Copilot

Co-authored-by: castrojo <castrojo@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
* fix(smoke): ignore Nvidia CDI failures in QEMU

Assisted-by: GPT-5.6 via GitHub Copilot

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix(e2e): scope metadata artifacts by image

Assisted-by: GPT-5.6 via GitHub Copilot

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: castrojo <castrojo@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…jectbluefin#781)

Bumps [fast-uri](https://github.com/fastify/fast-uri) from 3.1.5 to 3.1.7.
- [Release notes](https://github.com/fastify/fast-uri/releases)
- [Commits](fastify/fast-uri@v3.1.5...v3.1.7)

---
updated-dependencies:
- dependency-name: fast-uri
  dependency-version: 3.1.7
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…uefin#785)

* docs(scanner): remove stray malformed line in unit-test-module-stubs.md

PR projectbluefin#755 left a stray ' list):' line between the _FakeSSH implementation
and its example, breaking the Markdown structure and making the
copyable code example invalid. Remove the stray line.

Fixes projectbluefin#760

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>

* test: isolate unit tests from module stubs

Unit tests replace shared modules in sys.modules; pytest-xdist can leak those stubs across test modules and fail unrelated imports. Restore the module table after each test to keep CI parallelism deterministic.

Assisted-by: Claude Sonnet 5 via GitHub Copilot

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>

---------

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…rojectbluefin#788)

scripts/generate_skill_index.py (165 lines) is a required merge gate
(.github/workflows/docs-validate.yml runs it with --check) but had zero
unit tests. Add 36 tests covering:

- parse_frontmatter: YAML parsing and its two error paths
- collect: all nine front-matter validations (required fields per-field,
  category/status enums, entry_point self-match, id == name), doc_type
  extraction, description whitespace normalisation, and id-sorted output
- render_json / render_md: payload shape and generated document contents
- main: write mode, and every --check failure/success path, including a
  regression guard pinning that --check reuses the committed generated_at
  date rather than date.today() (switching to today's date would make the
  docs gate fail every day after the catalog is committed)
- a repository invariant test asserting the committed docs/skills/index.*
  files are in sync, mirroring the assertion docs-validate.yml makes

No production code changed.

Fixes projectbluefin#767

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
…docker digest to da027df (projectbluefin#863)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…rojectbluefin#862)

Clarify across AGENTS.md, docs/SKILL.md, contributing/SKILL.md, and
skill-improvement/SKILL.md that the skill-update gate is a mandatory
review gate before enqueuing, triggered when changes introduce or alter
a pattern, workaround, or contract. Pure refactorings, helper deduplication,
and added test coverage without new conventions are explicitly exempt.
Also clarify that when no existing skill page directly covers an area,
contributors should update the closest matching skill or create one under
docs/skills/<category>/<area>/SKILL.md.

Fixes projectbluefin#861

Assisted-by: Gemini 3.8 Flash via GitHub Copilot CLI

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ctbluefin#864)

Add an informational GNOME Shell version canary step in the vanilla-gnome
suite that queries and logs the org.gnome.Shell ShellVersion D-Bus property.

The step reads the property via gdbus get (via SSH in container or local
subprocess) and warns on error without gating the run. Documents the GNOME 51
readiness procedure in docs/skills/ci-ops/ops/references/fedora-version-targets.md.

Fixes projectbluefin#826

Assisted-by: Gemini 3.8 Flash via Copilot

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
… provider-list (projectbluefin#850)

* test(smoke): suppress session idle-lock in smoke setup to unblock GOA provider-list

- Configure org.gnome.desktop.session idle-delay 0 and org.gnome.desktop.screensaver lock-enabled false in before_all hook (both local subprocess and via SSH in container mode)
- Drop @pending tag from Online Accounts provider list scenario in tests/smoke/features/gnome_online_accounts.feature
- Document session idle-delay / lock-enabled suppression pattern in docs/skills/test-authoring/gnome/SKILL.md (two-output rule)
- Update coverage snapshot in docs/skills/test-authoring/suite-map/SKILL.md
- Add unit tests in tests/unit/test_smoke_environment.py

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>

(Conflict resolution by castrojo: the coverage-snapshot block in this
commit was regenerated with scripts/update_coverage_snapshot.py during a
rebase onto current main. Original authorship preserved.)

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix(smoke): restore Activities-overview guidance, unpin the idle-lock tests

Two narrow problems in this branch:

1. It relocated `## Overview search entry` into references/top-bar.md but
   deleted `## Activities overview (GNOME 50 QEMU)` outright. That section
   carries a live instruction ("do NOT assert Main.overview.visible ...
   scenarios must be quarantined") and projectbluefin#848's GNOME 51 audit cites it as a
   source location. Moved it into references/top-bar.md alongside its
   sibling rather than dropping it.

2. Both new tests wrapped `before_all` in `except Exception: pass` and then
   asserted two exact argv lists. They passed against a before_all that
   crashed immediately after those two calls, and would fail on a harmless
   change to how the command is built. Rewritten to drive before_all without
   swallowing (so a crash now fails the test) and to assert the observable
   contract - both settings applied locally, the same two forwarded over SSH
   in container mode - while tolerating argv shape and ordering.

   Mutation-checked: flipping lock-enabled false -> true in environment.py
   fails both tests; reverting passes them.

The idle-lock suppression itself and the GOA un-pending are untouched.

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: castrojo <jorge.castro@gmail.com>
…path (projectbluefin#811)

Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
…ce) (projectbluefin#818)

.coveragerc set 'source = tests', so the coverage run in
.github/workflows/unit-tests.yml measured only tests/. scripts/ was
invisible to 'coverage report --fail-under=75' and to the codecov patch
target, even though 8 of the 10 scripts already have unit tests under
tests/unit/. A coverage regression in a CI gate script — or deleting its
test file outright — would not move the gated number.

Adding scripts to the source list closes that hole. Measured with the
workflow's own command: scripts/ is at 81% and the repo-wide total is
84%, so the existing --fail-under=75 threshold is unchanged and still
passes. 'coverage xml' now also reports scripts/ to Codecov.

Closes projectbluefin#817

Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
…rojectbluefin#858)

Pagefind returns data-pagefind-meta values verbatim; they are derived from
behave run artifacts and ORAS manifest annotations, so HTML reaching a run
title, environment, or commit field executed in the dashboard origin via
the search results renderer. Escape title/status/environment/commit before
interpolation; leave the Pagefind excerpt as-is (escaped upstream, carries
<mark> highlighting by design).

Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
…luefin#859)

Bumps [devalue](https://github.com/sveltejs/devalue) from 5.9.0 to 5.9.2.
- [Release notes](https://github.com/sveltejs/devalue/releases)
- [Changelog](https://github.com/sveltejs/devalue/blob/main/CHANGELOG.md)
- [Commits](sveltejs/devalue@v5.9.0...v5.9.2)

---
updated-dependencies:
- dependency-name: devalue
  dependency-version: 5.9.2
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…n#176 to projectbluefin#847 (projectbluefin#849)

The 8 @future scenarios in the software suite (6x GNOME Software
navigation/regression/close in flatpak.feature, 1x flatpak install/uninstall
round-trip in flatpak_cli.feature, and 1x AT-SPI note in bazaar.feature) were
anchored to closed issue projectbluefin#176.

Issue projectbluefin#176 was closed when Bluefin adopted Bazaar layout coverage in
bazaar_ui.feature (projectbluefin#504). However, upstream GNOME Software coverage on
vanilla-gnome/GNOME OS remains unvalidated on GNOME 51 and cannot run on
Bluefin images that do not ship org.gnome.Software.

Re-anchor all @future comments, suite-map documentation, and coverage
snapshot references to fresh tracking blocker issue projectbluefin#847 targeting
GNOME 51 validation.

Closes projectbluefin#828

Assisted-by: Gemini 3.8 Flash via GitHub Copilot

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Jorge O. Castro <jorge.castro@gmail.com>
…gs (projectbluefin#865)

Add scripts/image_slug.py as the single owner of the image-slug grammar
for screenshot and dashboard naming. It strips registry hosts and orgs,
folds remaining separators (/, :, @) into hyphens, and sanitizes to the
OCI tag charset.

Consume the canonical slug in e2e.yml across all three derivation sites
(Capture boot time, upload-screenshot, and Write job summary) and ensure
scripts/image_slug.py is included in sparse-checkout.

Fixes projectbluefin#840

Assisted-by: Gemini 3.8 Flash via GitHub Copilot CLI

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…tbluefin#867)

* ci: restore migration-target and extra-tags inputs to e2e.yml

migration-test.yml passes migration-target and extra-tags to the reusable
e2e.yml, which stopped declaring them in projectbluefin#315, so every run fails at
workflow-startup (startup_failure, zero jobs) and the @migration lifecycle
lane has never executed.

Restore the two workflow_call input declarations, the MIGRATION_TARGET and
EXTRA_TAGS env entries, the EXTRA_TAGS tag-filter line, and the
MIGRATION_TARGET export on the local (lifecycle) behave branch. The KDE
container branch deliberately does not run @migration and is left alone.

Assisted-by: Ornith-1.5-35B-A3B-GGUF-Q6_K via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* docs(e2e): move migration-input notes into a reference to clear the 500-line cap

docs-validate failed on this branch with:
  FAIL: docs/skills/ci-ops/e2e-workflow/SKILL.md: SKILL.md exceeds 500 lines (525)

SKILL.md was already 498 lines on main; the 28 lines added here pushed it
over the cap that scripts/validate_docs.py enforces. Moved the migration
input table, the four-part wiring list, and the red flag into
references/migration-inputs.md (49 lines, cap 200) and left a one-sentence
pointer plus an entry in the reference list, matching how this skill already
splits installer-suite, kde-suites, and permissions.

No content dropped and no workflow change: .github/workflows/e2e.yml is
untouched by this commit. SKILL.md is now 499 lines and validate_docs passes.

Assisted-by: Claude Opus 4.5 via omp
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
Co-authored-by: castrojo <castrojo@users.noreply.github.com>
Co-authored-by: Jorge O. Castro <jorge.castro@gmail.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…e status blockers (projectbluefin#852)

* test(common): activate podman user socket scenario and track immutable status blockers (projectbluefin#833)

- Activate 'podman user socket is active' in common_portals.feature by
  enabling user lingering and starting podman.socket in before_scenario hook
- Re-anchor @pending comments on 'No RPM packages are layered on the base image'
  and 'bootc status shows a pinned image' in common_immutable.feature to
  dedicated tracking issue projectbluefin#838
- Update coverage snapshot in docs/skills/test-authoring/suite-map/SKILL.md and
  scripts/update_coverage_snapshot.py

Fixes projectbluefin#833

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>

(Conflict resolution by castrojo: the coverage-snapshot block in this
commit was regenerated with scripts/update_coverage_snapshot.py during a
rebase onto current main. Original authorship preserved.)

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* test(common): regenerate coverage snapshot after projectbluefin#850

projectbluefin#850 also moved one scenario from @pending to active, so the snapshot
totals on this branch (407 active / 112 pending) were stale once it
landed. Regenerated with scripts/update_coverage_snapshot.py rather than
hand-edited: 408 active / 111 pending.

Assisted-by: Claude Opus 4.5 via omp
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: castrojo <castrojo@users.noreply.github.com>
Co-authored-by: Jorge O. Castro <jorge.castro@gmail.com>
…e.md (projectbluefin#848)

* test(smoke): audit GNOME 50 workarounds against GNOME 51, update gnome.md

Issue projectbluefin#827 asks every GNOME 50-specific workaround in the repo to be
validated against GNOME 51 (Wayland-only, dropped legacy NVIDIA, frozen
extension API) and classified as still required / broken on 51 / obsolete.

Live validation requires a gnomeos-51 image via manual.yml (lab infra, not
available here) and PR projectbluefin#821 (qecore pins) landing first, so this PR ships the
part of projectbluefin#827 that is self-contained and does not depend on either: the audit
itself.

- Add references/gnome-51-audit.md: the full per-workaround inventory with
  where each lives, the GNOME 51 fact that bears on it, and a deliberate
  UNVALIDATED status. Frozen-API workarounds are flagged least likely to
  break; AT-SPI role/name workarounds (panel showing=False, Nautilus,
  Ptyxis, filler) most likely.
- Add a GNOME 51 workarounds audit section to docs/skills/test-authoring/gnome/SKILL.md
  and link the reference; bump skill to v1.1 and regenerate index.json.
- Keep every GNOME 50 workaround in place (Bluefin ships 50) and never delete
  on an unvalidated guess.

Assisted-by: Ornith-1.5-35B-A3B-GGUF-Q6_K via pi
Co-authored-by: kylerankin <kylerankin@users.noreply.github.com>
Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* docs(skills): keep gnome/SKILL.md under the 500-line cap

scripts/validate_docs.py fails any SKILL.md over 500 lines; this branch
pushed it to 518, which is what turned docs-validate red. The audit
section duplicated the governing facts, ground rules and validation
procedure that references/gnome-51-audit.md already carries, so the
inline copy is reduced to a pointer. File is now 499 lines.

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
Co-authored-by: castrojo <jorge.castro@gmail.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…cs (projectbluefin#854)

* docs(ci): make GNOME version string neutral across e2e summary and docs

The e2e job summary heading, README, and fedora-version-targets reference
pinned "GNOME 50" strings that go stale when gnomeos-latest flips to 51.

- e2e.yml summary heading: use "GNOME" instead of "GNOME 50" (no canary
  step exists to interpolate the detected Shell version).
- README: drop the pinned release from the top-bar fallback row and describe
  dual-version coverage (50 = Bluefin, 51 = gnomeos).
- fedora-version-targets.md: document gnomeos as tracking the current GNOME
  release rather than pinning GNOME 50.
- e2e-workflow SKILL.md: note the summary heading is intentionally
  version-neutral.

The job name at e2e.yml:166 is left unchanged pending human sign-off (it may
be referenced by downstream required checks).

Assisted-by: deepseek/deepseek-v4-flash-0731 via goose
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: mendezr <mendezr@users.noreply.github.com>

* docs(ci): address review — drop desktop word from summary heading, durable README wording

Responds to review on projectbluefin#854.

- e2e.yml summary heading: drop the desktop/version word entirely
  (`E2E Results ({suite})`), matching the precedent already set in
  .github/actions/gnome-e2e/action.yml. The e2e job runs KDE suites too,
  so claiming "GNOME" mislabeled KDE runs; dropping the word makes the
  heading agree with the job name on every run.
- README: replace the version-pinned "(50 = Bluefin, 51 = gnomeos)" with
  durable "(Bluefin and gnomeos)" that never needs touching as versions flip.
- fedora-version-targets.md: drop the "e.g. GNOME 50→51" that would read
  oddly once GNOME 52 ships; "current Fedora / GNOME release" is durable.
- e2e-workflow SKILL.md: update the heading note to document the desktop-free
  heading convention.

No behavioral change to e2e.yml.

Assisted-by: deepseek/deepseek-v4-flash-0731 via goose
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: mendezr <mendezr@users.noreply.github.com>

* docs(skills): keep e2e-workflow/SKILL.md under the 500-line cap

scripts/validate_docs.py fails any SKILL.md over 500 lines; the added
note pushed this file to 506, which is what turned docs-validate red.
Same content, reflowed into one paragraph. File is now 500 lines.

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* docs(skills): move e2e summary heading note

* docs(skills): drop e2e summary pointer

---------

Signed-off-by: mendezr <mendezr@users.noreply.github.com>
Co-authored-by: mendezr <mendezr@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: castrojo <jorge.castro@gmail.com>
…ings round-trips (projectbluefin#851)

* test(vanilla-gnome): add GNOME 51 reduced-motion and focus-ring gsettings round-trips

Add tests/vanilla-gnome/features/gnome_accessibility_51.feature with two
gsettings round-trip scenarios for the org.gnome.desktop.a11y.interface
settings GNOME 51 added (reduced-motion, keyboard-focus-visible-timeout) and
pre-51 images lack. Each writes the key, reads it back, restores the shipped
default, and checks the accessibility journal is clean.

Gate with @requires_gnome_51; environment.py probes the running Shell version
in before_scenario and skips on GNOME <= 50 (a runtime version probe, not a
non-runnable tag). The scenarios use only steps already loaded by vanilla-gnome's
steps.py (qecore.common_steps for command output, gnome_shell_steps for the
output assertion) — no extra step-module import, which would double-register a
step and break the dry-run. after_scenario is guarded so a pre-sandbox skip is
clean instead of an AttributeError in a real run.

Overlaps the GNOME 51 canary in projectbluefin#841 (projectbluefin#826) on the coverage/suite-map and gnome
skill files below; close projectbluefin#841 in favour of this PR.

Assisted-by: Ornith-1.5-35B-A3B-GGUF-Q6_K via lemonade
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* test(suite-map): regenerate coverage snapshot for vanilla-gnome a11y scenarios

Regenerate docs/skills/test-authoring/suite-map/SKILL.md after adding two
active vanilla-gnome scenarios. vanilla-gnome 13 -> 15 active; total 526 -> 528
across 73 feature files. Also corrects a stale common-row count (101 -> 99
active, 20 -> 22 pending) that the committed snapshot had drifted from.

Update scripts/update_coverage_snapshot.py SUITE_NOTES for vanilla-gnome.

Assisted-by: Ornith-1.5-35B-A3B-GGUF-Q6_K via lemonade
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* docs(gnome): document GNOME 51 a11y gsettings round-trips and @requires_gnome_51 gate

Add the reduced-motion / keyboard-focus-visible-timeout round-trips and the
@requires_gnome_51 runtime-skip gate to the gnome test-authoring skill: extend
"When to Use" and add a new "GNOME 51 accessibility gsettings round-trips"
section.

Assisted-by: Ornith-1.5-35B-A3B-GGUF-Q6_K via lemonade
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

---------

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Jorge O. Castro <jorge.castro@gmail.com>
Co-authored-by: castrojo <castrojo@users.noreply.github.com>
…docker digest to b924c4d (projectbluefin#868)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…tbluefin#787)

* refactor: fold duplicated _first_value into shared ssh_config

tests/common/features/environment.py and tests/kde-smoke/features/environment.py
each carried a byte-for-byte copy of the _first_value(*values) helper that
already lives in tests/shared/ssh_config.py. Three copies of the same
four-line function is exactly the copy-paste-over-shared-abstraction shape
called out in projectbluefin#763: nothing enforced that a fix or behaviour change to one
copy would land in the other two.

Both suites now import _first_value from tests.shared.ssh_config instead of
redefining it. No behaviour change — the three implementations were
identical.

Extend tests/unit/test_suite_environment_contract.py, which already parses
every tests/*/features/environment.py with ast to enforce the quarantine-gate
contract, with the same treatment for _first_value: a suite may not define its
own copy, and any suite that calls it must import it from
tests.shared.ssh_config. This closes the gap before it can reopen.

Ref projectbluefin#763.

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>

* docs(skills): record the shared _first_value contract

This PR does not just dedup a helper - it adds an AST guard in
tests/unit/test_suite_environment_contract.py that makes "no suite may
redefine _first_value" an enforced contract. AGENTS.md:146 wants the
matching docs/skills/** note when a change introduces a contract.

Placed in the unit-test reference rather than shared-ssh.md so this PR
stays disjoint from projectbluefin#790, which documents the ssh_argv contract there.

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Co-authored-by: v <a@l>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Bumps  and [js-yaml](https://github.com/nodeca/js-yaml). These dependencies needed to be updated together.

Updates `js-yaml` from 4.3.1 to 4.3.2
- [Changelog](https://github.com/nodeca/js-yaml/blob/4.3.2/CHANGELOG.md)
- [Commits](nodeca/js-yaml@4.3.1...4.3.2)

Updates `js-yaml` from 3.15.1 to 3.15.2
- [Changelog](https://github.com/nodeca/js-yaml/blob/4.3.2/CHANGELOG.md)
- [Commits](nodeca/js-yaml@4.3.1...4.3.2)

---
updated-dependencies:
- dependency-name: js-yaml
  dependency-version: 4.3.2
  dependency-type: indirect
- dependency-name: js-yaml
  dependency-version: 3.15.2
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…fin#801)

Bumps [svgo](https://github.com/svg/svgo) from 4.0.2 to 4.1.0.
- [Release notes](https://github.com/svg/svgo/releases)
- [Commits](svg/svgo@v4.0.2...v4.1.0)

---
updated-dependencies:
- dependency-name: svgo
  dependency-version: 4.1.0
  dependency-type: indirect
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…efin#802)

Bumps [astro](https://github.com/withastro/astro/tree/HEAD/packages/astro) from 7.2.6 to 7.2.8.
- [Release notes](https://github.com/withastro/astro/releases)
- [Changelog](https://github.com/withastro/astro/blob/main/packages/astro/CHANGELOG.md)
- [Commits](https://github.com/withastro/astro/commits/astro@7.2.8/packages/astro)

---
updated-dependencies:
- dependency-name: astro
  dependency-version: 7.2.8
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
…rojectbluefin#816)

* [quality] test: unit coverage for the tests/shared/screenshot.py capture path

tests/shared/screenshot.py sat at 50% line coverage: the existing
tests/unit/test_screenshot.py covers only the pure helpers
(_safe_fragment, _screenshot_path, _flatpak_installed), leaving the
capture path itself — the code every GNOME GUI suite relies on to
produce failure artifacts — entirely unexecuted.

Adds tests/unit/test_screenshot_capture.py covering:

  _ssh_run                   SSH argv from SSH_KEY/VM_IP/VM_USER/SSH_PORT
                             and the documented defaults
  _take_screenshot_via_ssh   grim -> gnome-screenshot -> gdbus fallback
                             chain, shlex quoting of paths, SetUnsafeMode
                             before Shell.Eval, all-methods-failed -> False
  _gdbus_screenshot          container (SSH) vs on-VM (local gdbus argv)
                             routing and the JSON GVariant path encoding
  take_screenshot            results-dir creation, stale-file removal
                             before capture, materialization wait loop
  take_app_screenshot        flatpak / PATH binary / gtk-launch selection
                             and process teardown on the error paths
  take_fastfetch_screenshot  terminal selection, per-terminal exception
                             recovery, and the no-terminal plain-desktop
                             fallback

No production code changes. Coverage of tests/shared/screenshot.py goes
from 50% to 94%; 25 new tests, full unit suite stays green.

Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>

* test(shared): assert screenshot capture behaviour, not argv shape

The behavioural half of this suite is good - fallback ordering, the
stale-PNG deletion, the every-method-fails path, and the fastfetch
terminal recovery all fail on a plausible bug. Several assertions did not,
and would have broken on a refactor that changes nothing observable:

- exact ssh option strings (`StrictHostKeyChecking=no`) - ssh_config owns
  those flags and may reorder them
- the hardcoded default key path `/home/bluefin-test/.ssh/id_ed25519` -
  a default value, not a behaviour
- `calls[0].startswith("source /tmp/session.env && ")` - pins the exact
  prefix rather than the requirement that the session env is sourced
- `len(calls) == 1` / `== 2` - pins the call count rather than the
  ordering it stands for
- `argv[-1] == '"/tmp/out.png"'` - pins exact quoting rather than that
  the path is quoted

Rewritten to assert those contracts through what a consumer observes: the
SSH target and command, that SSH_KEY/SSH_PORT overrides reach the
connection and the default port is used without them, that grim runs first
and short-circuits the rest, that gnome-screenshot picks the capture up
when grim is missing, that the session env is sourced, and that a path
with spaces stays quoted.

Mutation-checked twice: dropping the `source /tmp/session.env && ` prefix
fails the suite, and ignoring the SSH_PORT override fails it; reverting
each passes.

Also documents the capture fallback chain and the stale-file deletion in
docs/skills/test-authoring/gnome/SKILL.md, which AGENTS.md:146 asks for on
a tests/** change and which no page covered before.

This also stops the suite from breaking on projectbluefin#790, which rewrites _ssh_run
to build its argv through ssh_config.ssh_argv().

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: castrojo <jorge.castro@gmail.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ap (projectbluefin#869)

PR projectbluefin#816 appended two lines of screenshot-capture documentation to
docs/skills/test-authoring/gnome/SKILL.md, pushing it to 501 lines and
tripping the scripts/validate_docs.py 500-line SKILL.md cap. That check
runs in .github/workflows/docs-validate.yml on every pull_request and
merge_group with no path filter, so it now fails PR validation and the
merge queue for every PR, not just ones touching this file.

Remove one redundant duplicate blank line after the H1 heading (the
file has 19 such double-blank pairs; the others are untouched, this
is the minimal fix). No content change.

Assisted-by: Claude Sonnet 5 via omp

Signed-off-by: Danathar <Danathar@users.noreply.github.com>
Co-authored-by: Danathar <Danathar@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…ectbluefin#825)

* fix(shared): route AT-SPI tree dump through resolve_results_dir

The results-directory contract (userdata > TESTSUITE_RESULTS_DIR > /tmp/results)
was implemented once in tests/shared/results_dir.py, but the AT-SPI tree dump in
gnome_shell_steps.dump_atspi_tree and its reader in the smoke and vanilla-gnome
feature environments bypassed it by hardcoding /tmp/results. A redirected run
(-D results_dir=... or TESTSUITE_RESULTS_DIR) therefore split artifacts across
two directories: the collector uploads results/ while the AT-SPI dump always
landed in /tmp/results.

Route all three dump sites through resolve_results_dir so the AT-SPI tree lands
with the rest of the run when results_dir is overridden (identical to the default
/tmp/results when it is not).

Update test_dump_atspi_tree_writes_expected_content to give the mocked context a
config so resolve_results_dir resolves to a known path instead of the MagicMock
that a bare MagicMock context would return.

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* test(shared): restore default-path guard and cover AT-SPI after_all dumps

Address kubestellar-hive review on PR projectbluefin#825:

- Finding 1: the PR rewrote test_dump_atspi_tree_writes_expected_content to
  assert only the userdata branch, deleting the sole assertion for the default
  /tmp/results path — the branch every real run takes. Parametrise the test
  over userdata-set and unset-default, isolating the default from any ambient
  TESTSUITE_RESULTS_DIR so it genuinely exercises resolve_results_dir's
  fallback (and using a real empty config so the MagicMock default does not
  resolve to a truthy mock).

- Finding 2: the smoke and vanilla-gnome after_all dump sites were untested and
  wrapped in except Exception: pass. Add an AST-based contract guard in the
  existing no-VM suite asserting every after_all that dumps the AT-SPI tree
  routes through resolve_results_dir and hardcodes no /tmp/results.

* docs(skills): record the resolve_results_dir artefact contract

This PR adds an AST guard asserting no after_all carries a /tmp/results
literal, which makes the results-dir routing an enforced contract rather
than a convention. AGENTS.md:146 wants the matching docs/skills/** note;
no page documented resolve_results_dir before this.

Filed under e2e-workflow troubleshooting because the observable symptom is
a missing artefact, and to keep this PR disjoint from projectbluefin#790 (shared-ssh.md)
and projectbluefin#787 (unit-test-module-stubs.md).

Assisted-by: Claude Opus 4.5 via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

---------

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
Co-authored-by: v <a@l>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Jorge O. Castro <jorge.castro@gmail.com>
…in-VM install (projectbluefin#821)

tests/smoke/features/environment.py monkeypatches qecore.utility and calls
_get_uinput_device(), which only exists in qecore >= 4.19. The runner image
pinned qecore==4.16, so the patched keyboard_character_input /
keyboard_key_input raised AttributeError, swallowed it, and silently dropped
every keystroke — observed as dead keyboard + Firefox navigation failures in
the GNOME 50 smoke lane (bluefin run 34695427457), which feeds the
testing-lab gate currently red on every bluefin/bluefin-lts PR.

- container/Containerfile.runner: qecore==4.16 -> qecore==4.19.3
- .github/actions/gnome-e2e/action.yml: pin the in-VM pip install to the
  same versions (was unpinned — the drift vector)
- tests/unit/test_qecore_pin_parity.py: gate that runner and VM pins match

Closes projectbluefin#820

Signed-off-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
Co-authored-by: sec-check[bot] <sec-check[bot]@users.noreply.github.com>
mergeraptor Bot and others added 30 commits November 3, 2026 07:17
…bluefin#1095)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…bluefin#1102)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…in#1089)

* fix(e2e): fail gnome-e2e action when no scenario passed

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* fix(e2e): route assert_passed.py to action lane and refresh gate docs

Map scripts/assert_passed.py to the gnome-e2e action pseudo-lane now that
the composite action runs it, and update the assert_passed.py docstring,
the e2e.yml comment and the behave-rc output description, which still
claimed the action does not call the gate.

Hive-Fix-Attempt: 1/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* docs(e2e): note gnome-e2e action runs assert_passed.py in e2e-workflow skill

Hive-Fix-Attempt: 2/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…rojectbluefin#1057)

* fix(lifecycle): reject skipped required upgrade checks

Use the existing targeted-image input to enforce required lifecycle scenario and step completion at the final Behave verdict. Preserve untargeted exploratory skips and cover runtime and tag-filtered skips with real Behave exit-code tests.

Hive-Run: projectbluefin#1023
Hive-Plan: required-lifecycle-verdict
Hive-Spec: issue-1023#skipped-required-scenario
Assisted-by: openai-codex/gpt-6.1-sol via pi
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: mendezr <mendezr@users.noreply.github.com>

* fix(lifecycle): do not require parked upgrade scenarios

Scenarios tagged @pending, @future, @quarantine or @hardware_blocked
never run (tests/shared/quarantine.py skips them), so counting them as
required made every targeted run fail on the parked greenboot rollback
scenario in update_failure.feature. Exclude them from the required set
and cover each tag, plus a parked-only suite, in the unit tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SnzFs53Qcer1xDySyxv1Vj
Signed-off-by: Danathar <berths-purger-0g@icloud.com>

---------

Signed-off-by: mendezr <mendezr@users.noreply.github.com>
Signed-off-by: Danathar <berths-purger-0g@icloud.com>
Co-authored-by: mendezr <mendezr@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Danathar <berths-purger-0g@icloud.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
projectbluefin#1097)

* Let callers run hardware-oriented e2e suites and report them when blocked

Summary:
The e2e workflow now accepts `hardware`, `nvidia` and `security` as
documented `suites` values. When a selected suite's every scenario is
`@hardware_blocked` (no matching CI VM hardware), the run reports an
explicit blocked result and fails the job instead of showing a green
pass with zero scenarios run.

---

Previously only `smoke`, `developer`, `dx`, `software`, `vanilla-gnome`,
`bazzite`, `common`, `lifecycle`, `installer` and `kde-smoke` were
documented as `suites` values. The `security` group (issue projectbluefin#1096) and the
folded `hardware`/`nvidia` groups were referenced by the feature files but
never selectable. Selecting one made behave run zero scenarios and still
exit 0 -- a misleading green pass, because `e2e.yml` filters
`--tags ~hardware_blocked`.

This makes the three hardware-oriented groups selectable (input
description, suite-map docs) and adds a structural guard,
`scripts/suite_blocked.py`, that answers one question before behave runs:
does the suite contain at least one scenario that is not
`@hardware_blocked`? It scans the suite's feature files for the tag, so the
guard covers nvidia now and any future all-blocked group rather than one
hard-coded name.

`e2e.yml` calls it after the existing KDE skip block. On BLOCKED it writes a
synthetic skipped scenario carrying the #706 reason, sets
`suite_blocked=true`, and a new "Fail blocked hardware-only suite" backstop
fails the job. This mirrors the KDE "suite skipped" / "passed > 0" backstop
so a hardware gap is loud, never a silent green.

Tests: `tests/unit/test_suite_blocked.py` (18 cases) covers the feature-file
tag parser, `is_all_blocked` against the real hardware/nvidia directories,
the synthetic report shape, and the CLI.

Closes projectbluefin#1096

Hive-Run: pytest tests/unit/test_suite_blocked.py && ruff check tests/ scripts/ --select E,F,W --ignore E501 && python3 scripts/update_coverage_snapshot.py --check
Hive-Plan: 1. Add scripts/suite_blocked.py to detect an all-@hardware_blocked suite.
2. Wire it into .github/workflows/e2e.yml after the KDE skip block + a failing backstop.
3. Document hardware, nvidia, security as selectable suites.
4. Add 18 unit tests; run pytest + ruff.
Hive-Spec: tests/unit/test_suite_blocked.py

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* docs: keep SKILL.md files under the 500-line cap

docs-validate failed because two SKILL.md files exceeded the factory
500-line cap: e2e-workflow (528, bloated by projectbluefin#1096's added section) and
gnome (506, pre-existing).

- e2e-workflow: move the 'Hardware-only suite blocked backstop' detail
  into a new references/hardware-suite-blocked.md and keep a tight pointer
  in the SKILL.md (+ add it to the on-demand references list). Shorten the
  duplicated Known limitations bullet to point at the same reference and
  trim tangential sparse-checkout padding.
- gnome: condense the 'Session readiness across a GDM restart' section.

Both files now sit at/under 500; docs-validate is green.

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>

* fix(e2e): address review on the hardware-suite blocked guard

- suite_blocked.py: an Examples: table header is no longer parsed as an
  untagged scenario, so an all-blocked Scenario Outline suite reads as
  BLOCKED (new parser test).
- test_blocked_report_scenario_is_not_a_pass now calls blocked_report
  instead of asserting on a hand-built dict.
- e2e.yml: drop the sparse-checkout entry for scripts/assert_kde_passed.py,
  which does not exist.
- hardware-suite-blocked.md: the run step, not the script, writes the
  suite_blocked output and the ::error:: annotation.

Reporting the blocked scenarios of a partially blocked suite (projectbluefin#1096
criterion 2) is deferred; the PR now refs projectbluefin#1096 instead of closing it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SnzFs53Qcer1xDySyxv1Vj
Signed-off-by: Danathar <berths-purger-0g@icloud.com>

---------

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
Signed-off-by: Danathar <berths-purger-0g@icloud.com>
Co-authored-by: Danathar <berths-purger-0g@icloud.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…bluefin#1104)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…docker digest to a0fa256 (projectbluefin#1105)

* chore(deps): update registry.fedoraproject.org/fedora-minimal:latest docker digest to a0fa256

* fix(kde-runner): bump lxml to 6.1.3 for Python 3.15 wheels

The new fedora-minimal digest is Fedora 45 with Python 3.15. lxml 6.1.1
has no cp315 wheel, so pip fell back to the sdist and failed without
libxml2/libxslt headers. lxml 6.1.3 ships cp315 manylinux wheels.

Hive-Fix-Attempt: 1/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…docker digest to fb62feb (projectbluefin#1106)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
Run the input daemon under the user service manager so it survives SSH session cleanup.

Hive-Run: projectbluefin#1059

Hive-Plan: issue 1059 approved workflow change

Hive-Spec: issue-1059#daemon-lifetime

Assisted-by: openai-codex/gpt-6.1-terra via omp

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…d fail on blocked actions (projectbluefin#1109)

* feat(e2e): boot the test VM with SELinux enforcing and gate on denials

Drop selinux=0 from the guest kernel arguments so the candidate runs
with its own policy, and make the run prove it:

- Label only the files the harness writes on the non-SELinux runner,
  using the candidate's own setfiles and file_contexts. Files already
  unlabeled before injection are candidate-owned and left alone so
  labeling defects stay product failures.
- Fail early when a guest with a policy store is not Enforcing, and
  again after the suite (covers lifecycle reboots). Images without a
  policy store are reported as unsupported.
- Fail on enforced AVC denials not named in selinux_allowlist.yaml
  (narrow type/permission/comm/tclass matching, reason required).
- Trigger a deliberate httpd_t denial after judging the scenario log
  and fail if it does not reach the gate.

Evidence goes to results/selinux/ in the results artifact.

Refs projectbluefin#1024

Hive-Run: projectbluefin#1024
Hive-Plan: issue-1024-selinux-enforcing
Hive-Spec: testsuite-e2e-selinux#1024
Assisted-by: Claude Opus 5.5 via Claude Code
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: codebymarshall <codebymarshall@users.noreply.github.com>

* docs(e2e-workflow): drop stale selinux=0 from KERNEL_ARGS summary

The e2e guest no longer boots with selinux=0; point the skill at the
SELinux enforcing reference instead.

Refs projectbluefin#1024

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: codebymarshall <codebymarshall@users.noreply.github.com>

* docs(e2e-workflow): list SELinux steps in SKILL.md; narrow labeling scope claim

Add the Verify SELinux enforcing and SELinux denial gate steps to the
SKILL.md step list and link references/selinux-enforcing.md from the
references index. State that only each deployment's etc and var are
relabeled: the fallback boot.N symlinks under /ostree/boot.* stay
unlabeled and are read only by unconfined ostree/bootc.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Signed-off-by: codebymarshall <codebymarshall@users.noreply.github.com>

* docs(uefi-boot): drop stale selinux=0 guidance

e2e.yml no longer passes selinux=0; point authors at the enforcing
reference instead of steering them to re-add the bypass.

Hive-Fix-Attempt: 2/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* feat(e2e): make the SELinux enforcing lane opt-in

The enforcing boot, harness labeling, enforcing check and AVC denial
gate have not run on a hosted VM yet, and e2e.yml is called by every
image repo. Gate them behind a selinux_enforcing input (default false,
also exposed by manual.yml): without it the guest keeps selinux=0 and
the SELinux steps are skipped, so existing runs are unchanged. A
manual dispatch from main with selinux_enforcing=true qualifies the
lane; the rollout plan in projectbluefin#1021 then flips the default.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SnzFs53Qcer1xDySyxv1Vj
Signed-off-by: Danathar <berths-purger-0g@icloud.com>

---------

Signed-off-by: codebymarshall <codebymarshall@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Signed-off-by: Danathar <berths-purger-0g@icloud.com>
Co-authored-by: codebymarshall <codebymarshall@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…wn (projectbluefin#1112)

* fix(e2e): start ponytail under a unit name the image cannot already own

projectbluefin#1108 starts gnome-ponytail-daemon with systemd-run --user
--unit=gnome-ponytail-daemon. On dakota:testing that fails before any test
runs: "Unit gnome-ponytail-daemon.service was already loaded or has a
fragment file." Every GUI and lifecycle shard on main now fails at
"Install gnome-ponytail-daemon" (run 42124147912: smoke-a, smoke-b,
smoke-firefox, lifecycle), which is what keeps v1 held (projectbluefin#1087).

Use the unit name testsuite-ponytail-daemon instead, and fall back to the
previous nohup start if the user manager refuses the transient unit for any
other reason. The bus-name wait that follows is unchanged.

Refs projectbluefin#1087, projectbluefin#1108

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SnzFs53Qcer1xDySyxv1Vj
Signed-off-by: Danathar <berths-purger-0g@icloud.com>

* fix(e2e): drop the nohup fallback for the ponytail daemon

The e2e skill doc requires setup to fail when the transient user unit
cannot start: a nohup'd daemon in the SSH session is killed by logind
when the ControlMaster connection expires, leaving a stale D-Bus owner.
Keep the unit rename and let systemd-run failures fail setup again.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SnzFs53Qcer1xDySyxv1Vj
Signed-off-by: Danathar <berths-purger-0g@icloud.com>

---------

Signed-off-by: Danathar <berths-purger-0g@icloud.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…docker digest to a4093b8 (projectbluefin#1113)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
* ci: pin downloads used by desktop tests

Hive-Run: projectbluefin#1091
Hive-Plan: issue-1091-hunks-1-and-2
Hive-Spec: issue-1091#download-pinning
Assisted-by: openai-codex/gpt-6.1-terra via omp
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>

* ci: isolate pinned fetches scan from gate checkout

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>

* ci: address pinned download review feedback

Fail immediately on clone errors, align the ORAS action pin, and document Debian digest updates.

Assisted-by: OpenAI Codex via Codex
Co-authored-by: OpenAI Codex <noreply@openai.com>
Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>

---------

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: OpenAI Codex <noreply@openai.com>
…g scenarios (projectbluefin#1111)

* fix(e2e): pass BASE_IMAGE to the GUI runner and gate common on passing scenarios

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* test(e2e): follow the suite expansion's env-passed inputs and Dakota common split

The matrix resolve step now reads suites and image through env vars and
keeps Dakota's common suite as a single shard. Match the new invocation,
pass an image to the expansion program, and cover the Dakota case.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* docs(e2e): align skill doc with Utah cache and common backstop; drop dead -D image flag

Hive-Fix-Attempt: 1/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
projectbluefin#1114)

The runner's stock podman parses each heredoc line as a build instruction,
so the GNOME OS canary harness has never built. Move the base/devel identity
check into container/gnomeos-harness-identity.py, stage it in the build
context, and unit-test it alongside probe()'s evidence and cleanup paths.

Signed-off-by: quality <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: quality <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Hive-Run: projectbluefin#838
Hive-Plan: approved-direction
Hive-Spec: issue-838#re-validate
Assisted-by: openai-codex/gpt-6.1-terra via omp
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>
Hive-Run: projectbluefin#1086
Hive-Plan: issue-1086-docstring
Hive-Spec: issue-1086#acceptance
Assisted-by: openai-codex/gpt-6.1-terra via omp

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
The immutability scenarios are active now, so the suite-map note for
common should only list flatpak model/state and dconf defaults as
pending on projectbluefin#838. Regenerated the coverage snapshot.

Hive-Fix-Attempt: 1/6
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
…-timeout checks (projectbluefin#1117)

- Launch Files through the nautilus command; gio launch D-Bus-activates
  Nautilus on Dakota without mapping a window.
- Start the print socket listener with detached_launch_command so logind
  KillUserProcesses=yes does not kill it when the SSH session closes.
- stat -L sudoedit: it is a symlink to the setuid sudo binary on Dakota.
- Check Orca and speech-dispatcher with command -v instead of rpm, which
  Dakota does not ship.
- Ignore ublue-boot-timeout.service in the VM: bootctl refuses to run
  when not booted with UEFI.

Refs projectbluefin#1059

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

Co-authored-by: Danathar <6772335+Danathar@users.noreply.github.com>
…docker digest to ef221f0 (projectbluefin#1118)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…docker digest to 4aaf5db (projectbluefin#1120)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…ight code again (projectbluefin#1121)

* docs(gnome): point the GNOME 51 audit at functions, not stale line numbers

The GNOME 51 workaround inventory anchored every row to file:line. Six
weeks later those anchors were off by up to 244 lines, several rows named
one site of a workaround that lives in up to ten places, and three rows
described code that has since changed:

- row 9: smoke no longer quarantines on Main.overview.visible; it reads
  and sets the org.gnome.Shell OverviewActive D-Bus property
- row 10: the banner check reads the private Main.messageTray._banner
  field, so it is not covered by the frozen extension/notification API
- row 12: the keyring skip is in six environment.py files, not five
- row 8: no test step uses the ScreenSaver D-Bus lock; a suite run
  cannot validate it

Anchor each row by file + function/constant, list every site, add the
grep that enumerates the 62 GNOME 50 markers outside the twelve rows,
refresh the gnomeos-51 digest, and note that e2e.yml names every job
"GNOME 50" regardless of image. Every row stays UNVALIDATED: no live
gnomeos-51 run has been recorded yet.

Assisted-by: claude-opus-5-5 via omp
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Hive-Run: projectbluefin#827
Hive-Plan: issue-827 GNOME 51 workaround audit
Hive-Spec: projectbluefin#827#tasks-3
Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>

* docs(gnome): list gnome_files_steps Files app-name lookup in audit row 6

gnome_files_steps.py carries its own FILES_APP_NAMES, used by
_files_apps and _nautilus_app for the Files window and sidebar
checks; the gnome_apps_steps alias does not cover it.

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>

---------

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…docker digest to c0fcb89 (projectbluefin#1122)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…mmutable-validation

test: prepare system checks for a fresh image
…bluefin#1125)

bootc 1.17 (in Dakota since 2026-11-01) checks the new root partition's
type through udev about 200 ms after sfdisk. On GitHub-hosted runners
udev has not always seen loopNp3 by then, so bootc aborts with "root
partition 3 has type <none>" before writing anything (run 42611596358).
The install step then died on a bare "mount: wrong fs type".

Retry bootc install to-disk up to three times while no ostree
deployment has been written, probe the root partition for a filesystem
before mounting it, and fail with an annotation naming the image and
bootc's error line when every attempt fails. The late bootloader
failure ("bootupd is required for ostree-based installs") still
continues to direct kernel boot as before.

Closes projectbluefin#1124

Hive-Run: projectbluefin#1124
Hive-Plan: issue-1124 install-step retry
Hive-Spec: projectbluefin#1124#acceptance-criteria
Assisted-by: Claude Opus via omp

Signed-off-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Karibhou <Karibhou@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…in#1126)

Two Copyous stress scenarios errored on every Utah flavor in smoke-a
(run 42733699024):

- "generating rapid clipboard churn": wl-copy over SSH hung to the 30s
  timeout. wl-copy forks a child that keeps serving the selection, and
  the child inherited wl-copy's stdout/stderr, so ssh waited on it.
- "copying payload with embedded null bytes": ValueError, because the
  payload was put on argv (printf %s '<payload>' | wl-copy), and an
  argument cannot hold a NUL.

_clipboard_set now sends the payload on stdin (ssh_argv() in the runner
container, sh -c on the host) and runs
`wl-copy >/dev/null 2>"$err"`, printing wl-copy's stderr only when it
fails. Both directions source /tmp/session.env for the Wayland session.
_shell_quote is gone.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…docker digest to 7fcd241 (projectbluefin#1128)

Co-authored-by: mergeraptor[bot] <267480593+mergeraptor[bot]@users.noreply.github.com>
…bluefin#1119)

* feat(e2e): boot guest with SELinux enforcing by default

Flip selinux_enforcing default to true in e2e.yml and manual.yml so the
GHA VM is qualified with the candidate's own SELinux mode instead of
selinux=0. Callers can still pass false for the legacy boot.

Closes projectbluefin#1024

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

* test(selinux): expect enforcing-by-default e2e boot

Update selinux_qualify unit tests for the flipped selinux_enforcing
default and reword the KERNEL_ARGS comment so the install step keeps a
single selinux=0 occurrence.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>

---------

Signed-off-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: scanner <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
…v1 gate

bootc 1.17 (Dakota since 2026-11-01) reads the new root partition's type
from udev ~200 ms after sfdisk. On GitHub-hosted runners udev has a backlog
and has not always processed loopNp3 by then, so bootc install to-disk
--via-loopback aborts with 'root partition 3 has type <none>' before writing
any deployment. The setup step then records no scenario, so the v1 gate
holds the tag with 'no failing scenario recorded' for common, kde-smoke and
smoke.

The retry added in projectbluefin#1125 settled udev after the loopback was released and
before the new partition existed, so it never helped the partition bootc
creates; 6 of the last 7 runs (run 42807521428 @ 514c893) still failed all
3 attempts. Move the settle to the top of each attempt, before bootc creates
the partition: udevadm trigger --subsystem-match=block --action=change then
udevadm settle --timeout=30.

This is a runner/environment failure, not a test-code regression: candidate
and baseline fail identically on the same image digest. The full VM gate
cannot run locally, so verification is reasoning plus upstream CI.

Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
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.

v1 tag held by the post-merge VM gate