[brettheap]

Status: ready for review. The PR is still a draft; lane 3 marks it ready. The whole suite passed on Ubuntu 26.04 four times: in round 1, on ubuntu-26.04 by name, and in rounds 2 to 4, on ubuntu-latest, which gave out ubuntu-26.04 each time (below). Round 4 ran on the final head 9402c4d92, whose only change from round 3's head 9ae7622d7 is one comment in tests.yml (RV1097's S1), and its tests job is green, as are package, tests-git-bash and SonarCloud.

What changed

A follow-up outside v1.2.0 (Brett Heap, 2026-10-21T10:32:04Z). It lands only after the v1.2.0 tag, on a later merge word.

The problem (#1096): on the Ubuntu 26.04 that ubuntu-latest now gives out, the full suite filled /tmp under pytest's default basetemp. #1092's run of the tests job (job 119050702775, image ubuntu-26.04 20261012.163.1) ended 2436 failed, 11382 passed, 380 skipped, 5579 errors, with 28325 lines of Disk quota exceeded and then No space left on device, all under /tmp/pytest-of-runner/pytest-0. #1095 moved the job to ubuntu-24.04 to keep it green.

What round 1 measured: 26.04's /tmp is a tmpfs of 7.9G, and the suite writes about 25G (the runner's disk went from 58G used to 83G over the suite). So the issue's inference holds: the suite cannot fit in 26.04's /tmp, and fits on the runner's disk with room to spare (62G still free after it).

• Step 1, the temporary files (.github/workflows/tests.yml, the tests step):
  • pytest's basetemp is --basetemp="$RUNNER_TEMP/pytest", and the step sets TMPDIR: ${{ runner.temp }} for everything it starts.
  • Why both: tests/conftest.py (temporary_directory, a session autouse fixture) already points TMPDIR and tempfile.tempdir at <basetemp>/tmpdir for the whole session, and tests/test_adopt_temp_work_roots.py holds it there. So moving the basetemp moves everything the tests and their children write while the session runs. The step's own TMPDIR covers what is written before that fixture and after it, such as pytest's capture files, which tempfile opens at start-up.
  • Why runner.temp and not the workspace: the workspace is this checkout, a git work tree. A basetemp inside it would put every test's repositories inside this repository, where any git that looks upward from a folder that is not a repository of its own finds this checkout instead. runner.temp (/home/runner/work/_temp) is outside the checkout and on the runner's disk. Its path is as long as the old one (/tmp/pytest-of-runner/pytest-0 and /home/runner/work/_temp/pytest are both 30 characters), so no test meets a longer path than before.
  • The step prints df -hT /tmp "$RUNNER_TEMP" before the suite and after it, so every run's log shows the room the suite had and used, and it still ends with the suite's own exit status.
  • The step's comment gives round 1's measured sizes.
  • Hard-coded /tmp in tests/: none is written to. Every /tmp in tests/*.py and tests/conftest.py is text that no test writes to: a --upload-pack=touch /tmp/pwned that is refused before it runs, a plan path --output=/tmp/pwned that execute refuses as plan-unsafe-path (tests/test_adopt_e2e.py:573), a spelling value "/tmp/x y/z" (tests/test_bootstrap_spelling.py:751), remote URLs whose names the doctor reads and never fetches, PurePosixPath values, a jobserver string, and a /tmp/claude- prefix that a hygiene test looks for.
• Step 2, the proof: round 1 ran the tests job on runs-on: ubuntu-26.04 by name (commit b3b8d5ad8).
• Step 3, back to ubuntu-latest (commit ba1a23204): tests runs on ubuntu-latest, and both the proof run's comment and tests.yml and release.yml: the full suite and the Python 3.9 jobs run on ubuntu-24.04, since the Ubuntu 26.04 that ubuntu-latest now gives out has no 3.9 for setup-python and runs out of /tmp under the suite (#1094) #1095's 24.04 pin comment are gone. The package comment says that the other jobs keep their runners: tests asks for 3.x and release.yml's build for 3.12 on ubuntu-latest, both of which 26.04 has, and tests-git-bash runs on windows-latest.
• The tests (tests/test_release_workflow.py):
  • test_the_tests_job_runs_the_suite_on_ubuntu_24_04 is gone.
  • In its place, test_the_suite_keeps_its_temporary_files_on_the_runners_disk holds the step's env to {TMPDIR: ${{ runner.temp }}} and its commands to the five lines, among them the basetemp and the final exit "$status". It sits after the allow-list test, so the package test's docstring ("the next test holds every job of both workflows on ubuntu-latest") is true again.
  • setup_python_pairs() and the allow-list test are unchanged, and hold tests to 3.x or 3.12 on ubuntu-latest again.
• The doctor docstring (commit 9ae7622d7, added to this PR's scope by lane 3): tests/test_doctor_legs_members_next.py:37 said "tests.yml runs the suite on ubuntu-24.04 alone". It now says ubuntu-latest, which is true once tests is back there. Nothing else in the file changes.
• The step's comment (commit 9402c4d92, the fix round for RV1097's S1): the comment in tests.yml said tests.yml: the full suite keeps its temporary files under runner.temp, not /tmp, so it runs on Ubuntu 26.04 and the tests job goes back to ubuntu-latest (#1096) #1097's run on 26.04 (job 119180581846) passed, but that job concluded failure: its actionlint step failed on the unknown ubuntu-26.04 label, and only its tests step passed. The comment now cites round 3 instead (job 119263572144, ubuntu-latest, image ubuntu-26.04 20261012.163.1, the image shape-doctor.py: a leg's gitlink that no .gitmodules section registers, beside one whose sections register other paths or beside none, is named with its own way out, as #914 names one beside a .gitmodules with no section (#1080) #1092 failed on): /tmp at 38M used before and after, and the runner's disk at 57G used before the suite and 83G after. Nothing but that comment changes.
• RV1095e's NIT (the helper's matrix assertion fires for a job with a standard matrix and no setup-python step): I kept it fail-closed. No such job exists. The message names the job and its keys, so a maintainer who adds one sees why at once. A job that gets its Python some other way than actions/setup-python@ is also caught.

Out of scope: nothing. #1096 has no "Needs a ruling" or "Documented limit" section.

CI rounds

• Round 1 (head b3b8d5ad8, run 39587689309, tests job 119180581846, label ubuntu-26.04):
  • Image: ubuntu-26.04 version 20261018.168.1, Ubuntu 26.04.1 LTS. setup-python's 3.x gave Python 3.15.0.
  • The tests step passed: 19388 passed, 392 skipped in 4120.87s (1:08:40).
  • The df lines: /tmp is a tmpfs of 7.9G with 38M used, before the suite and after it. /dev/root (ext4, 145G) had 58G used and 88G free before, and 83G used and 62G free after.
  • The job failed at its last step, actionlint 1.7.12, on the proof label alone: .github/workflows/tests.yml:69:14: label "ubuntu-26.04" is unknown ... [runner-label]. actionlint's list has ubuntu-latest and ubuntu-24.04, so step 3's ubuntu-latest clears it. This came from the proof run's own label, not from the runner, and it needed no fix of its own.
  • No test failed, so there is no behaviour difference (git 2.55 or otherwise) to triage.
  • package (24.04), tests-git-bash and SonarCloud passed.
• Round 2 (head ba1a23204, run 39600385929, tests job 119223091927, label ubuntu-latest):
  • ubuntu-latest gave out image ubuntu-26.04 version 20261018.168.1, Ubuntu 26.04.1 LTS.
  • The job passed: 19419 passed, 392 skipped in 3961.03s (1:06:01). actionlint passed.
  • The df lines are the same as round 1's: /tmp, a 7.9G tmpfs, had 38M used before and after. /dev/root had 58G used and 88G free before, and 83G used and 62G free after.
  • package (24.04), tests-git-bash and SonarCloud passed.
• Round 3 (head 9ae7622d7, the docstring commit on top of round 2's head; run 39612168967, tests job 119263572144, label ubuntu-latest, 13:54:56Z to 15:21:51Z):
  • It ran on the pull request's merge ref, 4578d0cc, which is 9ae7622d7 merged into main 6da757cbe, so main's newer tests are in its count.
  • ubuntu-latest gave out image ubuntu-26.04 version 20261012.163.1, the image shape-doctor.py: a leg's gitlink that no .gitmodules section registers, beside one whose sections register other paths or beside none, is named with its own way out, as #914 names one beside a .gitmodules with no section (#1080) #1092's failing run had, and not round 1 and 2's 20261018.168.1. setup-python's 3.x gave Python 3.14.8 this time, where rounds 1 and 2 had 3.15.0. git was 2.55.0.
  • The job passed: 19574 passed, 393 skipped in 5111.69s (1:25:11). The tests step ran 13:55:05Z to 15:20:19Z. actionlint 1.7.12 passed (steps install actionlint and actionlint both succeeded, no output).
  • The df lines: /tmp, a 7.9G tmpfs, had 38M used before and after. /dev/root (ext4, 145G) had 57G used and 88G free before, and 83G used and 62G free after.
  • No test failed, so there is nothing to triage.
  • package (24.04), tests-git-bash and SonarCloud passed.
• Round 4 (head 9402c4d92, the fix round's comment commit on top of round 3's head; run 39632361496, tests job 119333706724, label ubuntu-latest, 16:10:20Z to 17:52:02Z):
  • It ran on the pull request's merge ref, ff84540, which is 9402c4d92 merged into main 4ed7c7bf5, so main's newer tests are in its count.
  • ubuntu-latest gave out image ubuntu-26.04 version 20261012.163.1, the image shape-doctor.py: a leg's gitlink that no .gitmodules section registers, beside one whose sections register other paths or beside none, is named with its own way out, as #914 names one beside a .gitmodules with no section (#1080) #1092's failing run had, as in round 3. setup-python's 3.x gave Python 3.14.8, and git was 2.55.0.
  • The job passed: 19642 passed, 393 skipped in 5966.75s (1:39:26). The tests step ran 16:10:29Z to 17:49:59Z. actionlint 1.7.12 passed (actionlint_1.7.12_linux_amd64.tar.gz: OK, steps install actionlint and actionlint both succeeded, no output).
  • The df lines: /tmp, a 7.9G tmpfs, had 38M used before and after. /dev/root (ext4, 145G) had 57G used and 88G free before, and 83G used and 62G free after.
  • No test failed, so there is nothing to triage.
  • package (24.04), tests-git-bash and SonarCloud passed.

Tests, each file alone under a stub gh (exit 97), GH_TOKEN=bogus, an empty GH_CONFIG_DIR, GIT_SSH_COMMAND=false, GIT_CONFIG_GLOBAL=/dev/null, EMAIL set, one basetemp and -o tmp_path_retention_policy=failed.

At head 9402c4d92 (the comment commit; only a comment in tests.yml changed from 9ae7622d7):

• tests/test_release_workflow.py: 59 passed
• tests/test_workflow_concurrency.py: 14 passed
• tests/test_repo_hygiene.py: 100 passed

At head 9ae7622d7 (the docstring commit):

• tests/test_doctor_legs_members_next.py: 186 passed, 6 skipped. pyflakes gives the same 207 warnings as at ba1a23204.
• tests/test_release_workflow.py: 59 passed
• tests/test_repo_hygiene.py: 100 passed

At head ba1a23204:

• tests/test_release_workflow.py: 59 passed
• tests/test_workflow_concurrency.py: 14 passed
• tests/test_install_sh.py: 399 passed
• tests/test_repo_hygiene.py: 100 passed
• tests/test_cli_reference.py: 23 passed
• pyflakes tests/test_release_workflow.py: no output, rc 0, as at origin/main.

At b3b8d5ad8, whose tests step is the same:

• tests/test_adopt_temp_work_roots.py, which holds the session's TMPDIR to <basetemp>/tmpdir: 9 passed.
• The same file in this step's own layout (TMPDIR and RUNNER_TEMP at one empty folder, --basetemp at <that folder>/pytest): 9 passed, and afterwards the folder held only pytest.

Probes and mutants, at head ba1a23204, each against tests/test_release_workflow.py. After each one, the file was restored from git show HEAD: and checked with cmp, and git status was clean.

#1095's probes and mutants:

• probe A, an unquoted python-version: 3.9 on tests: the allow list fails on ('tests.yml', 'tests', 'ubuntu-latest', '3.9') (1 failed, 58 passed). Now that tests is back on ubuntu-latest, the allow list catches this alone, as it did before tests.yml and release.yml: the full suite and the Python 3.9 jobs run on ubuntu-24.04, since the Ubuntu 26.04 that ubuntu-latest now gives out has no 3.9 for setup-python and runs out of /tmp under the suite (#1094) #1095's round 3.
• probe B, release.yml's build asks for '3.9': the allow list fails on ('release.yml', 'build', 'ubuntu-latest', '3.9') (1 failed, 58 passed).
• probe C, tests on a standard matrix (os: [ubuntu-latest], python: ['3.9']): the helper's assertion fails with ('tests.yml', 'tests', ['os', 'python']) (1 failed, 58 passed).
• probe D, an include row with os and no python: the ${{ check fails on ('tests.yml', 'tests', 'ubuntu-latest', '${{ matrix.python }}') (1 failed, 58 passed).
• probe E, the include row {os: ubuntu-latest, python: '3.9'}: the allow list fails on ('tests.yml', 'tests', 'ubuntu-latest', '3.9') (1 failed, 58 passed).
• mutant 1, the release's 3.9 row back on ubuntu-latest: the install-check test and the allow list fail (2 failed, 57 passed).
• mutant 2, package back on ubuntu-latest: the package test and the allow list fail (2 failed, 57 passed).

New mutants for this PR's lines. Each one fails test_the_suite_keeps_its_temporary_files_on_the_runners_disk (1 failed, 58 passed):

• T1, the step's TMPDIR set to /tmp.
• T2, --basetemp="$RUNNER_TEMP/pytest" dropped.
• T3, the final exit "$status" dropped, which would let the step pass over a failed suite.

T1 to T3 also failed the same way at b3b8d5ad8.

Merges

• git merge-tree --write-tree of head 9402c4d92 with origin/main 4ed7c7bf5 (read at 16:11Z, after the fix round's push): clean, rc 0, tree 971dcb7b62ad. Main's one commit since 823624d93 changes none of .github/, tests/test_release_workflow.py, tests/test_doctor_legs_members_next.py, tests/test_workflow_concurrency.py, tests/test_repo_hygiene.py, tests/conftest.py or pyproject.toml. The branch has not been merged with main or rebased: the landing after the v1.2.0 tag updates it.
• git merge-tree --write-tree of head 9ae7622d7 with origin/main 823624d93 (read after round 3 ended): clean, rc 0, tree 649e0d86bcd8. Main's move from 6da757cbe to 823624d93 changes none of .github/, tests/test_release_workflow.py, tests/test_doctor_legs_members_next.py, tests/conftest.py or pyproject.toml.
• Before round 3, 9ae7622d7 with origin/main 6da757cbe: clean, rc 0, tree 7cac4bb9b4dd.
• Earlier heads: ba1a23204 with 16f9a0e81 was clean (tree 2854a6a6590c). b3b8d5ad8 with 2a119f281 and with 16f9a0e81 was clean.
• The live open-PR list, read at 16:11:36Z on 2026-10-21 (22 open PRs: update-shape.py: update-accept-local-missing-name asks a copy an earlier apply kept, read unchanged, as a module a written copy imports from and as an importer of a written one, so a plain apply that stops either on its import is refused (#1082) #1133, bump-leg.py: a tag a leg with no origin holds without its commit, or through a tag it lacks, is told so, a tag of a tree or a blob it lacks is refused by name, and rows hold YAML_NOT_TEXT_RE's exponents of two digits (#1103) #1131, scaffold-project.py: a push ssh refused the host's key for names how to check that key against GitHub's fingerprints, not the ruleset hint, and the hint on GitHub says what its re-run can still wait at (#1108) #1130, repo_shape.py and bootstrap.py: COMMIT_RE, SHA256_RE and COMMIT_40 end in \Z, not $, so a pin's commit: or tree_sha256: that ends in a line break is refused as not 40 or 64 hex by every validator and adopt-project.py (#1017) #1126, adopt-project.py: check and execute ask a plan's org: and id: right after its three repository names, so a bad one beside an unanswered entry, a moved source, a refused name or another repository is that one refusal in both, and a pins: entry that is no pin is said as _read_as says it (#1020) #1125, validate-family.py: gitmodules-path-twice and the check for values git dies on read the .gitmodules git reads, the index's where the work tree's is gone, and every section's lines end git add -- .gitmodules, so a plain git commit after them takes the section out of HEAD (#1077) #1124, README.md, adopt-project.py and the handbook: a kept directory entry makefile/ is tested, README says make at the root reads a kept GNUmakefile or makefile instead, and a docstring says the shape's Makefile alone holds CONTRACTS_DIR (#1061) #1123, adopt-project.py: an org/repo --source too long for the disk is refused as a path is, naming no GitHub repository and offering no url, so --source <300 a's>/Thing is no longer offered https://github.com/<300 a's>/Thing, which can never clone (#1003) #1122, adopt-project.py: execute's NEXT prints python3 scripts/bootstrap.py where GNU make at the root reads a kept GNUmakefile or makefile, and its NOTE names that file and what it takes for make bootstrap to run there (#1047) #1121, adopt-project.py: plan's source line names the source's branch as _shown names a path, so a bidi control, a line boundary or a byte that is not UTF-8 in it is spelled and no longer reaches the terminal raw (#1035) #1120, adopt-project.py: check asks the split's own clone of the source in the work directory, so a clone this git will not make is adopt-git-unasked before any leg, and _resolved and _answers make their repositories as check's other throwaways do (#1083) #1119, adopt-project.py, shape_materialize.py and tests: a named bad link left out of the copy and pin findings takes no other path's finding with it, and the docstrings and the materializer's refusal name the #979 exception and what helps it (#1089) #1118, family.py: the remove that both member-duplicate-id remedies name and the add that bump's member-unknown names carry --family-root <holder>, quoted for the host, so each runs as printed (#1005) #1117, tests/test_path_guard_hygiene.py and shape-doctor.py: a bare is_file/is_dir/exists guard in a shipped tool fails by file and line unless its _is_a or a reasoned allowlist covers it, and a project.yaml linked into an unsearchable folder is yaml-unreadable (#430) #1111, shape-doctor.py: a gitlink beside a .gitmodules the leg's index tracks and this disk lacks is named from the index's copy, which git reads, with its own way out that restores the file first (git restore .gitmodules), as #1080 names one beside other sections (#1093) #1109, tests: the bump-leg.py and family.py reachability rows tag a descendant of the commit --to names, so a tag above it reads as one that reaches it and a tag below it keeps the plain words, and --points-at for --contains no longer survives (#1019) #1107, adopt-project.py: under a one-byte locale such as Latin-1, check reads a plan written in UTF-8 whose source is there, asking the plan's UTF-8 bytes where the spelling as written names nothing, rather than refusing plan-source-gone (#1078) #1106, shape_materialize.py and the pinned AGENTS-shape.md: where the root keeps a GNUmakefile or a makefile, which GNU make reads first, the Makefile collision follow-up no longer says the shape's make targets arrive with the merge, and the pinned file says make reads the first of the three and the block reaches it only once that file holds it (#1048) #1105, shape_materialize.py: env_commit's docstring says the split commit's message spells a path byte that is not UTF-8 as _shown does, an escape, and no longer says it names it with a ? in its place, which #900 replaced (#1018) #1104, family.py: a failed remove's by-hand lines spell a leg mounted at a path that is not UTF-8 as a printf word of octal escapes ending in a slash, so git -C <member>/<leg> and submodule update -- <leg> typed in any POSIX shell name that leg (#1086) #1098, tests.yml: the full suite keeps its temporary files under runner.temp, not /tmp, so it runs on Ubuntu 26.04 and the tests job goes back to ubuntu-latest (#1096) #1097 and the family root's bootstrap.py: credential_advice and ended_on say a member's failed clone ends git's update with 1 before any checkout but a leg's lets it go on, and killed_said says git's update was itself killed, not that it went on (#1085) #1088): none but this one touches .github/, tests/conftest.py, tests/test_release_workflow.py, tests/test_workflow_concurrency.py, tests/test_repo_hygiene.py, tests/test_doctor_legs_members_next.py or pyproject.toml.

Closes #1096

Lane: openreposhape-3 (openRepoShape-3)

🤖 Generated with Claude Code

[sourcery-ai]

Reviewer's Guide

The PR addresses Ubuntu 26.04 disk exhaustion by placing pytest and process temporary files under the runner's dedicated temp directory, adds diagnostics and a fail-safe exit-status wrapper, runs the job on Ubuntu 26.04 for validation, and updates workflow regression tests to enforce the new configuration.

File-Level Changes

Change	Details	Files
Moves the full pytest suite's temporary storage off /tmp and validates the workflow's disk-placement and exit-status behavior.	Sets TMPDIR to ${{ runner.temp }} for the test step. Passes $RUNNER_TEMP/pytest as pytest's basetemp. Adds pre- and post-run filesystem usage logging while preserving the suite's exit code. Documents why runner temp is used instead of the checkout workspace.	.github/workflows/tests.yml
Temporarily runs the tests job explicitly on Ubuntu 26.04 to prove the storage fix against the affected image.	Changes the job runner from ubuntu-24.04 to ubuntu-26.04 for the proof run. Updates package compatibility commentary for Python versions available on Ubuntu 26.04.	.github/workflows/tests.yml
Replaces the obsolete runner-version regression test with assertions covering the new temporary-file configuration.	Removes the expectation that the tests job runs on Ubuntu 24.04. Verifies the step environment, exact commands, basetemp location, filesystem diagnostics, and preserved exit status.	tests/test_release_workflow.py

Assessment against linked issues

Issue	Objective	Addressed	Explanation
#1096	Direct pytest’s basetemp and the tests’ temporary-file environment to the runner’s disk rather than /tmp.	✅
#1096	Run the full suite on Ubuntu 26.04 and establish whether it passes or reveals other differences.	❌	The PR describes the Ubuntu 26.04 proof run as still running and provides no completed result, so this objective is not yet demonstrated.
#1096	After the 26.04 run succeeds, return the tests job to ubuntu-latest and remove the runner-version pin.	❌	The job is currently pinned to ubuntu-26.04 by name for the proof run; it has not been changed back to ubuntu-latest.

Possibly linked issues

• tests.yml: on an Ubuntu 26.04 runner the full suite fills /tmp under pytest's basetemp (Disk quota exceeded, 2436 failed and 5579 errors), so tests runs on ubuntu-24.04 until the suite can run there #1096: The PR implements the issue’s requested temporary-file relocation and Ubuntu 26.04 proof run.

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[brettheap]

(lane openRepoShape-3, 2026-10-21T15:33:07Z) RV1097, independent review of #1097 at head 9ae7622: READY.

MUST: none.

SHOULD:

• S1 .github/workflows/tests.yml:128-131: the comment says "tests.yml: the full suite keeps its temporary files under runner.temp, not /tmp, so it runs on Ubuntu 26.04 and the tests job goes back to ubuntu-latest (#1096) #1097's run on 26.04 (job 119180581846) passed", but that job concluded failure (step 9, actionlint, unknown label ubuntu-26.04). Only its tests step passed. Fix: cite round 3's job 119263572144 instead (success, ubuntu-latest, image ubuntu-26.04 20261012.163.1, the image shape-doctor.py: a leg's gitlink that no .gitmodules section registers, beside one whose sections register other paths or beside none, is named with its own way out, as #914 names one beside a .gitmodules with no section (#1080) #1092 failed on; df: /tmp 38M used before and after, /dev/root 57G used before and 83G after), or say "its tests step passed". Landing after the v1.2.0 tag needs an update-branch and a new run anyway, so this can ride along with it.

NIT (PR body only):

• N1 line 17: the list of /tmp strings leaves out tests/test_adopt_e2e.py:573 (--output=/tmp/pwned, refused as plan-unsafe-path) and tests/test_bootstrap_spelling.py:751 ("/tmp/x y/z", a spelling value). The claim still holds; add them or say "among them".
• N2 lines 11-19: Step 1, Step 2, Step 4, and no Step 3. Renumber, or name the steps after tests.yml: on an Ubuntu 26.04 runner the full suite fills /tmp under pytest's basetemp (Disk quota exceeded, 2436 failed and 5579 errors), so tests runs on ubuntu-24.04 until the suite can run there #1096's three asks.
• N3 line 91: the open-PR list at 13:54Z is stale. scaffold-project.py: the --reuse-empty-repo help, its docs and each remedy that sends a person into that re-run say ssh, a credential helper or askpass can still make it wait, and how to keep it from waiting (#915) #1101 and templates/assembly-root/project.yaml and templates/family-root/family.yaml: the comment above visibility: says it is the visibility the run asked for, and that a repository the run reuses keeps its own, where it said "this is all three repositories' visibility" (#913) #1102 are closed, and adopt-project.py: execute's NEXT prints python3 scripts/bootstrap.py where GNU make at the root reads a kept GNUmakefile or makefile, and its NOTE names that file and what it takes for make bootstrap to run there (#1047) #1121 to bump-leg.py: a tag a leg with no origin holds without its commit, or through a tag it lacks, is told so, a tag of a tree or a blob it lacks is refused by name, and rows hold YAML_NOT_TEXT_RE's exponents of two digits (#1103) #1131 are new. At 15:28:55Z there are 22 open PRs, and none but tests.yml: the full suite keeps its temporary files under runner.temp, not /tmp, so it runs on Ubuntu 26.04 and the tests job goes back to ubuntu-latest (#1096) #1097 touches .github/, test_release_workflow.py, test_doctor_legs_members_next.py, test_workflow_concurrency.py, test_repo_hygiene.py or conftest.py.

Checks:

• TMPDIR is /home/runner/work/_temp in the step env (round 3 log). The basetemp is $RUNNER_TEMP/pytest, and conftest's temporary_directory holds TMPDIR at <basetemp>/tmpdir. No test writes to a hard-coded /tmp: every /tmp in tests/ is text, every TMPDIR override points under tmp_path, and the env helpers lay their variables over os.environ. /tmp stayed at 38M in all three rounds.
• runner.temp is a sibling tree of the workspace /home/runner/work/openRepoShape/openRepoShape, so no git walk-up reaches the checkout.
• The step's exit status is pytest's: the extracted run text under bash -e with a fake pytest gave exits 0, 1, 2, 5 and 137 back unchanged. A failing df would end the step nonzero, so a failed suite is never hidden. No change asked.
• package (tests.yml:310) and the release 3.9 install row (release.yml:908) stay on ubuntu-24.04 with 3.9. release.yml is unchanged.
• Image variance: rounds 1 and 2 got 20261018.168.1 with CPython 3.15.0, and round 3 got 20261012.163.1 with CPython 3.14.8. All three passed, so the risk is low.
• The sentence at tests/test_doctor_legs_members_next.py:37 is true once this lands, and is exactly the pre-tests.yml and release.yml: the full suite and the Python 3.9 jobs run on ubuntu-24.04, since the Ubuntu 26.04 that ubuntu-latest now gives out has no 3.9 for setup-python and runs out of /tmp under the suite (#1094) #1095 wording.
• Round 3 (job 119263572144): 19574 passed, 393 skipped in 1:25:11. actionlint 1.7.12 passed. df before: /tmp tmpfs 7.9G with 38M used, /dev/root 57G used of 145G. df after: 38M on /tmp, /dev/root 83G used. tests, package, tests-git-bash and SonarCloud are success on the head.
• Body: draft, no milestone, closingIssuesReferences [1096], follow-up sentence present, ends Closes tests.yml: on an Ubuntu 26.04 runner the full suite fills /tmp under pytest's basetemp (Disk quota exceeded, 2436 failed and 5579 errors), so tests runs on ubuntu-24.04 until the suite can run there #1096 / Lane / robot, no other closing keyword, nothing above U+007F but the robot. The ready-for-review line is accurate.

Tests at the head (stub gh exit 97, GH_TOKEN=bogus, empty GH_CONFIG_DIR, GIT_SSH_COMMAND=false, GIT_CONFIG_GLOBAL=/dev/null, EMAIL set): test_release_workflow.py 59 passed; test_workflow_concurrency.py 14 passed; test_repo_hygiene.py 100 passed.

Mutants (file restored, checked with cmp, git diff empty): T1 TMPDIR: /tmp KILLED; T2 --basetemp dropped KILLED; T3 exit "$status" dropped KILLED. Each one failed only test_the_suite_keeps_its_temporary_files_on_the_runners_disk (1 failed, 58 passed).

Merges: merge-tree with origin/main 823624d is clean, tree 649e0d86bcd8. No other open PR touches .github/ or these test files.

Not re-run: the doctor test file (the writer ran it at this head, 186 passed, 6 skipped), probes A-E and mutants 1-2 (run at ba1a232, and the head commit adds no line they test), and the full suite.

Lane 3's plan: #1097 is READY at 9ae7622. S1 (the tests.yml comment cites round 3's job) and N1 to N3 (the body) go in one small fix round, made together with the update-branch that the landing after the v1.2.0 tag needs anyway, so they cost no extra CI round. RV1097 then checks that head. #1097 stays a draft with no milestone and lands after the v1.2.0 tag, on a later word.

Lane: openreposhape-3 (openRepoShape-3)

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[brettheap]

(lane openRepoShape-3, 2026-10-21T21:07:07Z) RV1097 re-check of #1097 at head 9402c4d (main 038dafb, 21:03Z): READY against main, but landing needs a ruling on #1142 first.

Fixed:

• S1: the comment at .github/workflows/tests.yml:128-131 now cites job 119263572144 (success, label ubuntu-latest, image ubuntu-26.04 20261012.163.1, the image shape-doctor.py: a leg's gitlink that no .gitmodules section registers, beside one whose sections register other paths or beside none, is named with its own way out, as #914 names one beside a .gitmodules with no section (#1080) #1092 failed on). Every figure matches its log: /tmp 38M before and after (lines 269 and 551), /dev/root 57G before and 83G after (lines 270 and 552). Only comment lines changed.
• N1: line 17 now lists --output=/tmp/pwned and "/tmp/x y/z".
• N2: steps numbered 1 to 3, and the round 1 bullet says "step 3's".
• N3: current at 16:11:36Z, stale again now (N4).

SHOULD (ruling):

• S2: ci: support selected Linux runners in shape validation #1142 (ci/eagle-fleet-routing, head ce4dc66, not a draft, opened 20:27Z) conflicts with this PR: merge-tree rc 1, CONFLICT in .github/workflows/tests.yml and tests/test_release_workflow.py. It moves tests, package and release.yml's Linux jobs to the self-hosted label eagle-linux-ubuntu24 because of the GitHub-hosted billing cap, and it edits test_the_tests_job_runs_the_suite_on_ubuntu_24_04, which this PR deletes. This PR moves tests to GitHub-hosted ubuntu-latest. Rule on the order first.
• If ci: support selected Linux runners in shape validation #1142 lands first: rebase here, keep tests on eagle-linux-ubuntu24, keep the TMPDIR/basetemp step and its test, and reword tests.yml:121-131, the package comment at :301-308 and tests/test_doctor_legs_members_next.py:37.
• If this PR lands first: ci: support selected Linux runners in shape validation #1142 rebases over the ubuntu-latest line and drops its edit of the removed test.
• Either way, the second PR needs a new CI round.

NIT:

• N4 body line 105: the open-PR list from 16:11:36Z is stale, and "none but this one touches .github/" is no longer true. Live at 21:03:25Z: 24 open PRs; tests/test_path_guard_hygiene.py and shape-doctor.py: a bare is_file/is_dir/exists guard in a shipped tool fails by file and line unless its _is_a or a reasoned allowlist covers it, and a project.yaml linked into an unsearchable folder is yaml-unreadable (#430) #1111 closed; tests/test_adopt_root_symlinks.py: a row puts a directory the root keeps at a pin's file reached through a kept link, and holds that check renames the path the walk reaches, not the file kept in it, so rv1091's R6 no longer survives (#1116) #1140, adopt-project.py: a recorded source whose .git this user cannot read is refused source-unreadable, naming that .git, so check and execute no longer call a clone plan made with its .git at mode 000 no longer whole and clone it again (#1099) #1141 and ci: support selected Linux runners in shape validation #1142 are new; ci: support selected Linux runners in shape validation #1142 touches .github/actionlint.yaml, release.yml, tests.yml and tests/test_release_workflow.py.

Round 4 (run 39632361496, job 119333706724): success; merge ref ff84540 (9402c4d into 4ed7c7b); image ubuntu-26.04 20261012.163.1; CPython 3.14.8; git 2.55.0; 19642 passed, 393 skipped in 1:39:26; df before: /tmp tmpfs 7.9G with 38M used, /dev/root 57G used and 88G free; df after: /tmp 38M, /dev/root 83G used and 62G free; actionlint 1.7.12 OK. package, tests-git-bash and SonarCloud are success on the head. Main's 24.04 push job 119283526361 took 1:38:41, so the run time is not a 26.04 cost.

Body: draft, no milestone, closingIssuesReferences [1096], only closing keyword Closes #1096, ends Closes #1096 / Lane / robot, nothing above U+007F but the robot, no Sourcery block. The status line is accurate on CI but does not mention #1142.

Tests at 9402c4d (stub gh exit 97, GH_TOKEN=bogus, empty GH_CONFIG_DIR, GIT_SSH_COMMAND=false, GIT_CONFIG_GLOBAL=/dev/null, EMAIL set): test_release_workflow.py 59 passed; test_workflow_concurrency.py 14 passed; test_repo_hygiene.py 100 passed.

Merges: with origin/main 038dafb, clean, tree 53552f8085f5; main's two new commits touch none of .github/, the four test files, conftest.py or pyproject.toml. With #1142: conflict (S2). No other open PR touches these files.

Not re-run: the mutants (the step lines and the test file are unchanged since 9ae7622, where T1 to T3 were killed), the doctor test file and the full suite.

Lane 3's plan: #1142 is the operator's runner change for the GitHub-hosted billing cap, so #1142 lands first, and #1097 follows it, after the v1.2.0 tag as before. Lane 3 then merges main into this branch (a merge, no rebase and no force push), keeps tests on eagle-linux-ubuntu24, keeps the TMPDIR and --basetemp step and its test, rewords tests.yml:121-131, the package comment at :301-308 and tests/test_doctor_legs_members_next.py:37, and refreshes the open-PR list (N4). The four 26.04 rounds stay in the body as the record. What #1097 then keeps is the suite's temporary files on the runner's disk wherever it runs, and the body will say that #1142's routing supersedes its move to ubuntu-latest. If Brett Heap wants the other order, lane 3 follows that instead.

Lane: openreposhape-3 (openRepoShape-3)