Repository navigation
fix(auto-version): keep oSPARC metadata versions semver under dev stamps (unblocks verify-image-build; #695) - #697
Merged
Conversation
The develop-channel stamp fans the full X.Y.Z.devN into every [bumpversion:file:*] entry, including the 8 .osparc/*/metadata.yml service metadata files - whose schema (MetadataConfig, validated by 'ooil assemble-spec' in verify-image-build) requires strict semver. Every PR branched from a stamped develop therefore fails verify-image-build with string_pattern_mismatch on version '1.6.3.devN' (first surfaced on #694; systemic, tracked as #695). - .bumpversion.cfg: flag the 8 metadata entries semver-only = true - auto_version.py: semver-only entries are stamped (and gate-checked) with the .devN-stripped BASE; a dev stamp also self-heals metadata files left carrying the dev suffix by the pre-fix flow - one-time repair: metadata.yml 1.6.3.dev2 -> 1.6.3 (x8) - SPEC V5 amended: semver-only entries carry the base version Verified locally: check-pr passes on the repaired tree; injecting a dev suffix into one metadata.yml reproduces the §V5 drift rejection; simulated 'apply --to 1.6.3.dev3' keeps every metadata.yml at 1.6.3 while compose tags / pyproject / current_version take the dev version, with cmd_apply's built-in check_fanout green; self-heal from a dev-stamped metadata.yml verified.
This was referenced Oct 6, 2026
Closed
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #697 +/- ##
============================================
+ Coverage 50.63% 85.50% +34.87%
============================================
Files 107 109 +2
Lines 5350 12978 +7628
Branches 605 1353 +748
============================================
+ Hits 2709 11097 +8388
+ Misses 2563 1770 -793
- Partials 78 111 +33
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
Contributor
There was a problem hiding this comment.
Copilot review overview
🟡 Changes recommended
Clean human base bumps fail in both the new helper logic and the documented bump2version workflow.
Review effort: Balanced
Findings: 2
Open (2)
What changed in this PR
Keeps oSPARC metadata on strict semver while retaining .devN versions elsewhere.
Changes:
- Adds
semver-onlyfanout handling. - Repairs eight metadata versions.
- Documents the revised versioning invariant.
| File | Description |
|---|---|
SPEC.md |
Documents semver-only fanout. |
scripts/auto_version.py |
Implements base-version fanout and validation. |
.bumpversion.cfg |
Marks metadata entries semver-only. |
.osparc/flaskapi/metadata.yml |
Restores strict semver. |
.osparc/node/metadata.yml |
Restores strict semver. |
.osparc/proxy-uq-read/metadata.yml |
Restores strict semver. |
.osparc/proxy-uq-write/metadata.yml |
Restores strict semver. |
.osparc/proxy-sumo-read/metadata.yml |
Restores strict semver. |
.osparc/proxy-sumo-write/metadata.yml |
Restores strict semver. |
.osparc/proxy-moga-read/metadata.yml |
Restores strict semver. |
.osparc/proxy-moga-write/metadata.yml |
Restores strict semver. |
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
JavierGOrdonnez
pushed a commit
that referenced
this pull request
Oct 7, 2026
Re-ports the frontend bits of ghost #665 (fork replay branch port/13-fixes-debt-sweep-665) onto the stacked slices 1-2: - InputVariableDist: log-flag clearing now anchors on the normal MEAN too (uniform min already covered) - the payload builders read `scale` verbatim, so a surviving flag shipped an invalid log-normal the backend V16 guard 400s (B24mk) - OutputVariableDist: manual scale toggle now merges functionally into the live per-uid map - a render-time snapshot could clobber an entry useAutoDetectQoiScale wrote for a sibling QoI (B25qr) - useAutoDetectQoiScale: the CV pair scores under the CURRENT input scales instead of an all-linear strawman (#663 audit); merged onto the slice-1 requestJson transport - SobolIndicesPlot/sobolIndices: log-flagged normal seeding checks support before proposing the mean+-3sigma box (B26st) - node SPEC: V12 amendment + T32mt row + B24mk/B25qr/B26st history Stacked on #697/#698 lineage (slices 1-2 of #70).
…Copilot #697) Both review findings share one root: after the semver-only split the fanout files no longer share a single searchable token, so the module must own every bump path. - apply: a semver-only entry's before-token is now whichever shape it legally holds - the full current version (pre-#695 dev-stamp, still self-healed) or the current BASE (the normal post-#695 state). The clean-base transition Copilot cited (cfg 1.6.3.dev2 + metadata 1.6.3, apply --to 1.6.4) no longer raises 'version drift?'. - new 'bump --part {patch|minor|major}': raises the base, drops the .devN counter, rewrites current_version + all 12 fanout entries via the same dual-shape logic, then re-checks the V5 fanout. 'make version-{patch|minor|major}' now runs it inside the same hermetic CI docker image (python 3.11, packaging 24.2, git 2.43, HOME=/tmp); bump2version retires - its single search = {current_version} cannot match split entries, which is exactly how the next required human base bump would have failed. SPEC: §I version cmd rewritten; §V5 names scripts/auto_version.py as the sole bump mutator (no bump2version). Verified in a sandboxed fixture repo (tags v1.6.1..v1.6.3): clean-base bump patch/minor/major -> 1.6.4/1.7.0/2.0.0, fanout green each time; self-heal + bot dev-bump unchanged; stale base (1.6.2 behind tag v1.6.3) refused; real drift (9.9.9) still raises, message naming both accepted shapes; the PRE-FIX script reproduces the drift failure on the clean state; the container recipe runs green end-to-end.
Collaborator
Author
|
Both Copilot findings addressed in 2927066 (replies on each thread). One root, two symptoms: with the semver-only split the fanout files no longer share a single searchable token, so every bump path moved into |
JavierGOrdonnez
pushed a commit
that referenced
this pull request
Oct 7, 2026
Re-ports the frontend bits of ghost #665 (fork replay branch port/13-fixes-debt-sweep-665) onto the stacked slices 1-2: - InputVariableDist: log-flag clearing now anchors on the normal MEAN too (uniform min already covered) - the payload builders read `scale` verbatim, so a surviving flag shipped an invalid log-normal the backend V16 guard 400s (B24mk) - OutputVariableDist: manual scale toggle now merges functionally into the live per-uid map - a render-time snapshot could clobber an entry useAutoDetectQoiScale wrote for a sibling QoI (B25qr) - useAutoDetectQoiScale: the CV pair scores under the CURRENT input scales instead of an all-linear strawman (#663 audit); merged onto the slice-1 requestJson transport - SobolIndicesPlot/sobolIndices: log-flagged normal seeding checks support before proposing the mean+-3sigma box (B26st) - node SPEC: V12 amendment + T32mt row + B24mk/B25qr/B26st history Stacked on #697/#698 lineage (slices 1-2 of #70).
JavierGOrdonnez
pushed a commit
that referenced
this pull request
Oct 7, 2026
Re-ports the frontend bits of ghost #665 (fork replay branch port/13-fixes-debt-sweep-665) onto the stacked slices 1-2: - InputVariableDist: log-flag clearing now anchors on the normal MEAN too (uniform min already covered) - the payload builders read `scale` verbatim, so a surviving flag shipped an invalid log-normal the backend V16 guard 400s (B24mk) - OutputVariableDist: manual scale toggle now merges functionally into the live per-uid map - a render-time snapshot could clobber an entry useAutoDetectQoiScale wrote for a sibling QoI (B25qr) - useAutoDetectQoiScale: the CV pair scores under the CURRENT input scales instead of an all-linear strawman (#663 audit); merged onto the slice-1 requestJson transport - SobolIndicesPlot/sobolIndices: log-flagged normal seeding checks support before proposing the mean+-3sigma box (B26st) - node SPEC: V12 amendment + T32mt row + B24mk/B25qr/B26st history Stacked on #697/#698 lineage (slices 1-2 of #70).
JavierGOrdonnez
pushed a commit
that referenced
this pull request
Oct 7, 2026
Re-ports the frontend bits of ghost #665 (fork replay branch port/13-fixes-debt-sweep-665) onto the stacked slices 1-2: - InputVariableDist: log-flag clearing now anchors on the normal MEAN too (uniform min already covered) - the payload builders read `scale` verbatim, so a surviving flag shipped an invalid log-normal the backend V16 guard 400s (B24mk) - OutputVariableDist: manual scale toggle now merges functionally into the live per-uid map - a render-time snapshot could clobber an entry useAutoDetectQoiScale wrote for a sibling QoI (B25qr) - useAutoDetectQoiScale: the CV pair scores under the CURRENT input scales instead of an all-linear strawman (#663 audit); merged onto the slice-1 requestJson transport - SobolIndicesPlot/sobolIndices: log-flagged normal seeding checks support before proposing the mean+-3sigma box (B26st) - node SPEC: V12 amendment + T32mt row + B24mk/B25qr/B26st history Stacked on #697/#698 lineage (slices 1-2 of #70).
JavierGOrdonnez
added a commit
that referenced
this pull request
Oct 8, 2026
fix(auto-version): keep oSPARC metadata versions semver under dev stamps (unblocks verify-image-build; #695)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
verify-image-buildis currently broken for every PR branched fromdevelop— independent of the PR's content. TheAssemble docker compose specstep (ooil,MetadataConfig) fails with:Root cause: the auto-tag bot's dev stamp fans the full
X.Y.Z.devNversion into all[bumpversion:file:*]entries — including the 8.osparc/*/metadata.ymlservice metadata files, whose schema requires strict semver. The compose image tags legitimately carry.devN; the oSPARC metadata does not survive it. Since195632ec(bump dev version to 1.6.3.dev2 [skip ci]) is an ancestor ofdevelop's tip, every new PR branch inherits the failing combination — this is what turns #694'sverify-image-buildred (not the a11 bump itself), and the same will bite #696 and anything after. Systemic, tracked as #695.What
.bumpversion.cfg: the 8 metadata entries gainsemver-only = true.scripts/auto_version.py:semver-onlyentries are stamped — and gate-checked bycheck_fanout— with the.devN-stripped base; everything else (compose tags,Makefile,pyproject) keeps the full version. A dev stamp additionally self-heals any metadata file left carrying a dev suffix by the pre-fix flow. Since the fanout no longer shares one searchable token, every bump path now lives here: a newbump --part {patch|minor|major}raises the base (dropping.devN) through the same dual-shape rewrite.Makefile:make version-{patch|minor|major}routes toauto_version.py bump --partin the same hermetic CI docker image — bump2version is retired (its singlesearch = {current_version}cannot match split entries; Copilot caught that this would have broken the next human base bump).metadata.ymlfiles1.6.3.dev2 → 1.6.3(required so this PR's version-check gate passes; also repairsdevelopfor in-flight PRs the moment it merges).Verification (all local, this tree)
check-pr --target developon the repaired tree:develop: 1.6.3.dev2 unchanged, ok✅1.6.3.dev2back into onemetadata.ymlreproduces a §V5 drift rejection (fields: ['1.6.3.dev2'], expected '1.6.3') ✅apply --to 1.6.3.dev3: all 8 metadata files stay1.6.3while compose/pyproject/current_versionmove to1.6.3.dev3;cmd_apply's built-incheck_fanoutgreen ✅apply --to 1.6.3.dev3→ metadata normalized to1.6.3✅v1.6.1..v1.6.3): clean-statebump --part patch|minor|major→1.6.4/1.7.0/2.0.0, fanout green each;apply --to 1.6.4clean-base transition (the flagged hole — pre-fix script reproduces itsversion drift?failure) ✅; stale base behind a tag refused; genuine drift still raises, message naming both accepted shapes; the exact container recipe runs green end-to-end ✅Landing order
Merging this first unblocks the
verify-image-buildleg of #694 and #696 (both only need a re-run afterwards — no rebase required, the fix is base-side). Closes part of #695; the remaining exposure there (gen-baselines.shplainuv syncre-stampinguv.lock) is unaffected by this PR.