Repository navigation
feat: add multilingual text library - #856
hivecommons-hive[bot] merged 7 commits into
Conversation
|
@mrbobbytables this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (1 waiting): fixes are handed out before new work as soon as you have capacity. |
|
@mrbobbytables this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (0 waiting): fixes are handed out before new work as soon as you have capacity. |
7c40309 to
843ebc6
Compare
|
stage 080-awaiting-merge-admission → 050-review-pending: contributor-lane:no-verdict |
There was a problem hiding this comment.
Summary: This change adds the build recipe for a library that lets people type in many languages, copied from Fedora's recipe. The recipe matches Fedora, the source archive's checksum matches the real upstream file, and the counts and tracking records line up. Two small points: one paragraph in the maintainer guide still tells readers to import this library as a future step, and the bundled patch file was reformatted slightly from Fedora's copy (it still applies). Neither blocks merging.
Reviewed at head 843ebc6d (the PR moved past 64e97886 before review).
docs-currency
- low —
docs/skills/factory-build-backlog.md:244-246still says "Build and publish the providers first, then import and buildm17n-lib", but this PR performs the import (packages/m17n-lib/m17n-lib.spec:1,config/factory-build-backlog.toml:742-743). A reader following the skill would be told to import a recipe that already exists.integrating current main; union independent catalog records and regenerate inventory instead of restoring stale workflow or Packit pins. The providers were built and published first; `m17n-lib` was imported recipe-only in #856 and still needs hosted build, publication, and runtime proof (#495). Configuration checks establish neither hosted build success nor input-method runtime acceptance.
style
- info —
packages/m17n-lib/m17n-lib-1.8.0-multilib.patch:7,13,22: blank context lines lost their leading space compared to dist-git3e20ff11(trailing-whitespace hook excludespackages/per.pre-commit-config.yaml:20-21, so this was a manual/editor change). Verifiedgit apply --check -p1against the 1.8.7 tarball succeeds, so harmless; restoring the exact Fedora bytes keeps re-import diffs clean.
Verified: m17n-lib.spec differs from dist-git only by http→https on Source0 (line 15, matches repo convention in docs/skills/factory-build-backlog.md:466); sources, tests/*, .gitignore byte-identical; upstream tarball SHA-512 eb6bd219… matches config/upstream-sources.json and sources; tools/validate.py → 697 (691 rawhide) matching docs/architecture.md:81,133-148; all 18 #495 names now under [resolved].
No findings from: correctness, security, intent-alignment.
Confidence: 5/5 (safe)
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@mrbobbytables this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (2 waiting): fixes are handed out before new work as soon as you have capacity. |
843ebc6 to
4a44daa
Compare
|
stage 080-awaiting-merge-admission → 050-review-pending: contributor-lane:no-verdict |
|
Updated the backlog guidance: m17n-lib is recorded as imported recipe-only in #856 and still needs hosted build, publication, and runtime proof. Restored the three Fedora patch context-space prefixes to keep re-import diffs clean. 🐝 Hive Agent: |
|
stage 080-awaiting-merge-admission: no activity for 30m0s: merge-train:queued-behind#861:3/7:0/0:.packit.yaml,factory-build-backlog.toml,upstream-sources.json,architecture.md,+15 |
|
stage 080-awaiting-merge-admission → 060-reviewed: check-pending |
|
stage 080-awaiting-merge-admission → 050-review-pending: contributor-lane:no-verdict |
|
@mrbobbytables this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (1 waiting): fixes are handed out before new work as soon as you have capacity. |
There was a problem hiding this comment.
Summary: Since the last review, the small compatibility fix that this new text-input library needs was reformatted, and the reformatting broke it: the build tool that applies it now rejects the file as malformed, so the package cannot build at all. The earlier version applied cleanly. One character on one line needs changing. The other changes since last time (three download sources pointed at their project homepages, and two guidance pages) came in from the main branch and are sound.
What I need from you: Restore the + on the blank line in the multilib patch (fix below).
correctness
-
high —
packages/m17n-lib/m17n-lib-1.8.0-multilib.patch:22— the delta turned the+(added blank line) into(context blank line). The hunk header@@ -64,17 +64,15 @@(line 4) now disagrees with the body, and GNUpatch, which%autosetup -p1atpackages/m17n-lib/m17n-lib.spec:70runs, rejects it. Dry-run ofpatch -p1 --fuzz=0inside the pinnedm17n-lib-1.8.7tarball exits 2 with "malformed patch at line 25" (theecho "-I@includedir@"context line);git apply --checklikewise reports a corrupt patch at line 26. The version at the previous head4a44daacand Fedora's original both apply with exit 0 against the same tarball. The PR body says this is recipe-only with no hosted build yet, so nothing has caught it. Fix:- fi;; - + echo "${LIBNAME}";; - ++ --cflags)(line 22: a lone
+, matching the earlier head; Fedora's copy has+which is equivalent.)
Delta otherwise checked: fwupd, lcms2, taglib upstream URLs in config/upstream-sources.json all serve bytes matching the unchanged SHA-512s (downloaded and hashed), and the config/fedora-primary-sources.txt removals satisfy the ratchet in tools/validate.py:161-192; the docs/skills/lane-validation.md:83-86 wording matches pipeline_relevant/sources_touch_existing in tools/canary.py:87-157. Those three files arrived via the merge from main, not PR work.
No findings from: security, intent-alignment, style, docs-currency.
Confidence: 3/5 (needs attention) — 1 high finding
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
c3300e4 to
5b16c7a
Compare
|
stage 080-awaiting-merge-admission → 050-review-pending: contributor-lane:no-verdict |
There was a problem hiding this comment.
Summary: Re-review of the rebased revision. The build recipe still carries a broken fix-up file (the "multilib patch") that the build applies to the upstream source before compiling; because two lines lost their leading characters, the patch tool rejects it and the package cannot be built. That was the only blocker last time and it is still present. One newly reworded guidance paragraph also claims the prerequisite libraries were already published, which the project's own tracking data does not record.
What I need from you: Restore the patch byte-for-byte from Fedora and soften the "published" claim.
correctness — still present from previous review (head c3300e4)
packages/m17n-lib/m17n-lib-1.8.0-multilib.patch:22 and :27 differ from Fedora dist-git 3e20ff11 (the commit .hummingbird-upstream.json:5 pins). Line 22 is + (an added line of two spaces) upstream but a bare context line here; line 27 lost its leading space. Hunk 1 therefore declares -64,17 but contains 18 old-side lines. Verified against the pinned 1.8.7 tarball: git apply --check → error: corrupt patch at line 26; Fedora's copy applies cleanly. %autosetup -p1 (m17n-lib.spec:72) will fail on the first hosted build. .pre-commit-config.yaml:19-21 already excludes packages/ from whitespace hooks, so this was a hand edit — re-copy the file verbatim:
curl -sL https://src.fedoraproject.org/rpms/m17n-lib/raw/3e20ff11bb678840764a50c792df15eddcc1326e/f/m17n-lib-1.8.0-multilib.patch \
-o packages/m17n-lib/m17n-lib-1.8.0-multilib.patch
# expect sha256 1320d30037cbb878499d00ea90936558d6df94443ae666b72f500db72d283eb0
docs-currency — new in delta
docs/skills/factory-build-backlog.md:246 now states "The providers were built and published first", but config/factory-build-backlog.toml:340 records anthy-unicode from #788 with no published_run/published_nevra, which is how this file records publication for other entries (e.g. :339). The same paragraph (:248-249) tells readers to keep build, publication and runtime evidence separate; the sentence should follow its own rule:
inventory instead of restoring stale workflow or Packit pins. The provider
recipes were merged first; `m17n-lib` was imported recipe-only in #856 and
No findings from: security, intent-alignment, style.
Confidence: 3/5 (needs attention) — 1 high finding
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@mrbobbytables this PR is back in your court ( Review asks:
Reply here or push a new head to this branch (same sign-off); CI and the review swarm re-run automatically and the PR returns to the normal path. Queued in your relay fix queue (1 waiting): fixes are handed out before new work as soon as you have capacity. |
Import the Rawhide m17n-lib recipe and pin its verified upstream release. This is recipe-only work; factory build, publication, and runtime evidence remain open for the tracking issue. Hive-Run: projectbluefin#495 Hive-Plan: issue-495 implementation Hive-Spec: issue-495#implementation Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Move m17n-lib from pending to resolved recipe accounting and link the import PR. Hosted build, publication, and runtime acceptance remain open. Hive-Run: projectbluefin#495 Hive-Plan: issue-495 implementation Hive-Spec: issue-495#implementation Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Clarify that dependency blockers must be rechecked against the selected base before carrying a historical waiting note forward. Hive-Run: projectbluefin#495 Hive-Plan: issue-495 implementation Hive-Spec: issue-495#implementation Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
5b16c7a to
ee040f0
Compare
|
Hive base moved: rebased onto 🐝 Hive Agent: |
|
stage 080-awaiting-merge-admission → 050-review-pending: pending: head pushed 2m56s ago (< min_head_age) |
Restore the pinned Fedora patch bytes and avoid claiming provider builds are published. Assisted-by: Auto via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: mrbobbytables <mrbobbytables@users.noreply.github.com>
|
Restored the multilib patch byte-for-byte from the pinned Fedora dist-git source (SHA-256 🐝 Hive Agent: |
|
stage 080-awaiting-merge-admission → 050-review-pending: pending: head pushed 2m14s ago (< min_head_age) |
|
Merged by hive contributor-lane auto-merge: CI green for ≥30m, reviewer verdict approve, @mrbobbytables holds the trusted tier in this hive's contributor registry. |
Summary: This adds the build instructions for a text library that helps people enter text in many languages. It is the last package recipe in this part of the language-support set, now that its supporting libraries are available. The package has not yet been built or published, so the broader work remains open.
What I need from you: Review the recipe and source pin.
Change
Imports the Fedora Rawhide
m17n-librecipe from dist-git commit3e20ff11bb678840764a50c792df15eddcc1326e. The upstream release archive is fetched over HTTPS; its SHA-512 digest matches the imported Fedora source manifest. Fedora's copy is retained only as a fallback mirror.This is recipe-only work. The package has not yet been built in the pinned hosted workflow, published in a factory image, or exercised in a runtime transaction. Those checks remain necessary before the batch issue can close.
Validation
just check— passed.just test— 1408 passed, 2 skipped.python3 tools/validate.pyandpython3 tools/factory_build_backlog.py --check— passed.SC2129in.github/workflows/rebuild-rpms.yml:367.Refs #495 (partial: hosted build, publication, and runtime proof remain).
— hive: backend=copilot model=gpt-6-luna
🐝 Hive Agent:
contributor| SHA:c81a7aed