Repository navigation
bootc: extend test-49 bridge to a >=1.16.4 stager (stages V1 directly) #19
Description
Follow-up to bootc#2248 and forge bootc#18. test-49 only pins bootc 1.16.0, which stages the V2 fallback. 1.16.4+ parse composefs.digest= and stage V1 directly (and seal a V1 digest under bare composefs=), which nothing tests.
Plan:
- Parametrize test-49 by the stager version and expected staged format: env BOOTC_bridge_stager_version (1.16.0 | 1.16.4) and BOOTC_bridge_expected_format (v2 | v1); assert-old-fixture checks the given version; boot1 asserts the expected format instead of hard-coded v2; boot2/3 follow from it.
- Add a second pin next to the 1.16.0 one in the Justfile: bootc-1.16.4-1.fc44 (Koji tags coreos-pool and f44-updates), with RPM and /usr/bin/bootc sha256; generalize just test-composefs-bridge to take the stager (e.g. test-composefs-bridge <seal_state> [stager=1160|1164]).
- The stager Dockerfile should mostly work unchanged (1.16.4's ukify also takes --rootfs); check its seal-uki differences and that the current client accepts a 1.16.4-sealed image (the bare composefs= V1/V2 mismatch reported on #2248).
- Add CI legs (sealed+unsealed) or fold into the existing bridge job matrix; +2 jobs.
Activity
cgwalters-bot commentedon Sep 26, 2026
The Composefs Stable board's copy of this item said:
cgwalters-bot commentedon Oct 17, 2026
Board scheduler: no plan. A triage run is asked whether this is done; it answers here. Delete this comment to ask again.
cgwalters-bot commentedon Oct 17, 2026
Not done. None of the 40 PRs merged into bootc-dev/bootc since this issue was opened (2026-09-26) touches test-49. On upstream
main,test-49-composefs-1-16-bridge.nustill assertsbootc 1.16.0and a V2 stage. The Justfile has notest-composefs-bridgetarget andci.ymlhas no bridge job. Its last change is still the #2248 commits (75c779c3, 2026-09-23).The fix is written but not merged. It is in cgwalters-forge/bootc#30, a draft that is stacked on draft #18. Neither has been promoted upstream, and neither has reviews or comments. #30 covers every part of the plan:
bootc-1.16.4-1.fc44next to 1.16.0, withjust test-composefs-bridge MODE [1.16.0|1.16.4].The PR reports devspace runs at 1ad219c1 (
bls 1.16.4,sealed 1.16.4andsealed 1.16.0passed). Fork CI is off, so there is no CI run to link.What's left is not agent work:
/promotethem to bootc-dev/bootc.mainand GitHub reports it as conflicting. Its files (ci.yml,Justfile,tmt/plans/integration.fmf,tmt/tests/tests.fmf) overlap later merges such as ci: Fetch grub-cc from Fedora Koji bootc-dev/bootc#2575.Why there is no job plan: an implement run would only redo #30. Also,
cgwalters-forge/bootcis not in thereposbounds ofdispatch.toml, so a hosted plan for it would be refused. The rebase can be done by the bot that owns #30 or by the operator when promoting. Once #30 lands upstream, this issue can be closed against it.cgwalters-bot commentedon Oct 18, 2026
✅ dispatch ended well. Result: #19 (comment)
(Run status from 2026-10-17T10:58:14Z, re-posted by the bot without the agent's own words. The original job-token comment was removed so this issue counts as collaborator-only input, tracker#738.)
Generated-by: https://github.com/cgwalters/#llms
cgwalters-bot commentedon Oct 22, 2026
✅ dispatch ended well. Result: #19 (comment)
cgwalters-bot commentedon Oct 22, 2026
Still valid, and the fix is already written. It needs the operator to decide how it lands (option 3: no job plan).
Upstream has not changed since the 17 Oct triage.
test-49-composefs-1-16-bridge.nuwas last changed by #2248 (75c779c3, 23 Sep). It still pins only 1.16.0 and expects a V2 stage. No upstream PR touches it.Where the fix is: draft cgwalters-forge/bootc#30. It covers all four steps of the plan: it parametrizes test-49, pins
bootc-1.16.4-1.fc44, addsjust test-composefs-bridge MODE [1.16.0|1.16.4]and adds CI legs for sealed 1.16.4 and BLS 1.16.4. It does not cover the UKI case where 1.16.4 stages onto a 1.16.0 install, which is bootc-dev/bootc#2334 (still open). It was rebased on 17 Oct onto upstream 3c5d235a and is mergeable, one docs-only commit behind upstream main. Its comment reports thatjust validateand thebls 1.16.4,sealed 1.16.4andsealed 1.16.0legs passed on a devspace at bd84fec9. Fork CI is off, so there is no run to link.What blocks it: #30 contains all three of #18's commits. One of them was rewritten (354c29be) to fix the 1.16 stager install, which upstream 0c2f4879 broke by dropping
prepare-root.conf. #18 itself is still at the broken ccc70d3c. Bot runs that tried to refresh #18 and #30 on 17 Oct both stopped because the series edits.github/workflows/ci.yml.cgwalters-forge/bootcis in dispatch'sreposbounds but not in[unprotected_files](dispatch.toml), so no agent run can deliver this series. An implement plan would hit the same wall, or just redo #30.Decision for
@cgwalters: choose one.(a) Review #30 and promote it upstream as one PR that supersedes #18, then close #18.
(b) Push 354c29be onto #18 by hand, then promote #18 and then #30 as a stack.
(c) Add
cgwalters-forge/bootcto[unprotected_files]. The org ruleset 25084478 already requires an approving review there. A fix run could then refresh both PRs.(a) is the least work. Once the change merges upstream, this issue can close against it. #36, stacked on #30 and now conflicting, is an extension and is outside this issue.
cgwalters-bot commentedon Oct 26, 2026
Question for the operator (from the audit of the needs-human queue): Answered by the tracker#7 choice: review forge bootc#30 as one PR superseding #18 (A), hand-push the fix onto #18 (B), or unprotect bootc workflows for bot runs (C)?
The board's Why for this item was: Follow-up from forge bootc#18: bridge only covered 1.16.0; >=1.16.4 stages V1 directly (#2248). Stacked on #18. Result: bridge 1.16.4 BLS, 1.16.4 sealed UKI, 1.16.0 sealed + validate + unit-tests passed on a 16-core devspace.
Generated-by: https://github.com/cgwalters/#llms
cgwalters-bot commentedon Oct 27, 2026
Duplicate of #7: forge bootc#30 already implements this plan, and it waits on the review question there.
Generated-by: https://github.com/cgwalters/#llms