Repository navigation
feat: add ibus-anthy through ibus-typing-booster input-method recipes (#494) - #804
Conversation
|
@kylerankin 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: This change imports five Japanese, Korean, Chinese and multilingual typing engines into the package factory. One of them, the Japanese engine, will fail to build as submitted: its recipe reads a small checksum file that Fedora ships alongside it, and that file never made it into this change because a repository-wide ignore rule treats it as a downloaded archive. The other four engines and the bookkeeping updates look right.
What I need from you: Add the missing checksum file (fix below) and re-push.
correctness
-
high —
ibus-anthycannot produce an SRPM:Source1is missing from the tree. Fedora's dist-git at the locked commit9d8e245tracksibus-anthy.tar.gz.sumin git (not in the lookaside), and the spec reads it unconditionally in%prep:packages/ibus-anthy/ibus-anthy.spec:26declaresSource1: ...tar.gz.sum#/%{name}.tar.gz.sumand:92runsSAVED_SUM=$(grep sha512sum %SOURCE1 ...). The file is absent frompackages/ibus-anthy/atc2fda6c. Root.gitignore:15(packages/*/*.tar.*) matchesibus-anthy.tar.gz.sum, sogit addsilently dropped it aftertools/import_rawhide.pyextracted the full tree (extract(), lines 55–66). Same failure class the.gitignore:21–26comment documents for mdadm/setxkbmap keyrings.packages/ibus-anthy/sources:1lists only the tarball, sosource_pipeline.pycannot fetch it from the lookaside either —packit srpmdies with "Bad file: .../ibus-anthy.tar.gz.sum". The.hummingbird-upstream.jsontree09cca42…also claims a Fedora tree this directory does not match.Fix — add a negation beside the existing keyring exceptions and force-add the file:
!packages/bc/kevin_pizzini.asc # ibus-anthy's %prep greps its own upstream checksum file (Source1); it is # committed in dist-git, not fetched, so it must be tracked here too. !packages/ibus-anthy/ibus-anthy.tar.gz.sumthen
curl -sL https://src.fedoraproject.org/rpms/ibus-anthy/raw/9d8e245e6824c56334c30769ae4b7fa14052a187/f/ibus-anthy.tar.gz.sum -o packages/ibus-anthy/ibus-anthy.tar.gz.sum && git add packages/ibus-anthy/ibus-anthy.tar.gz.sum.
intent-alignment
- low — PR body misstates two things; no code change needed. (1) Body says
ibus-m17nneeded an https mirror because the rawhideSource0ishttp://; actuallypackages/ibus-m17n/ibus-m17n.spec:9is already https but points at the GitHub auto-generated archive, which hashes to6d08cf3b…, not the locked353685a6…. The lock (config/upstream-sources.json:6169–6178) correctly uses the release asset, which I downloaded and verified matches both the lock and Fedora's lookaside — so the recipe is right, only the explanation is off. (2) Body saysenchant2is a "stage-0" package;config/upstream-sources.json:512carries nostage. The"stage": 1onibus-typing-booster(:6188) is harmless —docs/architecture.md:156says stage only orders members of a BuildRequires cycle — but the stated reason does not hold.
No findings from: security, 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
|
@kylerankin the hive has adopted this PR ( |
|
1 of 2 commits on |
d485ead to
7970062
Compare
|
Needs a decision, not a merge: the automerge lane will not merge this unattended. Head: verdict approved @7970062, 21/21 checks green, soak 19m0s of 30m0s. Clear the cause and the hive re-evaluates on its next pass; the stage moves to |
7970062 to
32a2da4
Compare
|
#804 state=OPEN drift=RESOLVED Base drift was found and repaired. This PR's base moved forward between opening and now. I rebased the branch onto the current tip of
🐝 Hive Agent: |
|
#804 review follow-up Base was already moved and repaired — see the On the blocking review items:
🐝 Hive Agent: |
|
Merged current |
|
@kylerankin this PR has a new head without confirmed hive push evidence, so the hive has handed this PR back to you ( |
|
@kylerankin this PR is back in your court ( Failing checks:
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. |
|
Follow-up |
ca3d8d7 to
05174b2
Compare
There was a problem hiding this comment.
Summary: This change adds five Japanese/Korean/Chinese/multilingual keyboard input engines to the package factory. The recipes themselves were already reviewed and looked fine; what changed since is that the branch tried to catch up with the main line, but main has since moved the goalposts: it stopped committing counts and report snapshots and now requires the source list to be alphabetically sorted. As it stands the branch cannot be merged (Git reports conflicts) and, once rebased, the validation step would fail on the unsorted list.
What I need from you: Rebase onto current main, drop the count/report edits, and run just sort-source-locks.
correctness
- Branch conflicts with
mainand would fail validation after rebase (medium, in-scope). GitHub reportsmergeable: CONFLICTING; a local merge of05174b2dintomain(02e6061d) conflicts inconfig/factory-build-backlog.toml,config/upstream-sources.json,docs/architecture.md,docs/skills/factory-build-backlog.md, and modify/delete onreports/factory-build-backlog.json(deleted on main by #858, still modified here). Separately, the five new lock entries are appended at the tail ofconfig/upstream-sources.json(ibus-anthyat line 6396, afteriwd, followed byhfsplus-tools), while main'stools/validate.pynow rejects an unsorted lock ("Lock entries stay sorted by name", validate.py:269) — so a straight conflict resolution still failsjust check. The backlog TOML on this branch also re-lists names main already resolved (x264-libs,sssd-*,tesseract-*, …) and drops main's publication-evidence comments, so resolve the TOML by keeping main's side and only moving the six ibus names to[resolved]. Fix:git fetch origin && git rebase origin/main # keep main's side for the four shared files, drop reports/factory-build-backlog.json just sort-source-locks && just check
docs-currency
- Doc hunks reintroduce hard-coded counts and the committed report that #858 removed (low, in-scope).
docs/architecture.md:81,133-144,163re-add "686 recipes"/"96 of 686" tables that main replaced with prose pointing atvalidate.pyoutput;docs/skills/factory-build-backlog.md(PR side) describesreports/factory-build-backlog.jsonas a "Committed snapshot … regenerate withjust regenerate-inventory", whereas main's version says "The report is derived and not committed … Do not reintroduce a tracked report, rollup, or count." Drop these hunks during the rebase; take main's text.
intent-alignment
packages/yubikey-manager/*re-added to work around the canary (info, in-scope, advisory). Commit05174b2dre-adds the yubikey-manager recipe tree and lock entry "so the canary judges it not relevant". The six files are byte-identical to main (#839), so they vanish on rebase — just confirm the rebase leaves no yubikey-manager hunk in this PR, since it is unrelated to #494.
No findings from: security, style.
Confidence: 3/5 (needs attention) — 1 medium finding; 2 perspectives requested changes
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
@kylerankin 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. |
|
Head Blocker: this PR comes from a fork ( Moving to |
…projectbluefin#494) Import the ibus-anthy, ibus-hangul, ibus-libpinyin, ibus-m17n and ibus-typing-booster input-method engines — including the ibus-anthy-python subpackage — into the factory recipe set: pin each release archive at SHA-512, lock each rawhide dist-git commit in .hummingbird-upstream.json, register each in config/upstream-sources.json and .packit.yaml, and move each from the fonts-i18n area to [resolved] in the factory build backlog. ibus-typing-booster is staged at 1 because it BuildRequires enchant2, an ICU-linking stage-0 package. Hosted build, publication and runtime evidence remain outstanding for the batch. kasumi-common and kasumi-unicode stay pending: the kasumi source archive is unresolvable over https (dead jaist.dl.sourceforge.jp mirror, MD5-only sources, no maintained GitHub mirror), so no recipe can be imported. kasumi-unicode is nonetheless available from the paired Fedora build root, so ibus-anthy is not blocked on it. anthy-unicode, libpinyin, libpinyin-data and m17n-lib remain pending in projectbluefin#308. Hive-Run: projectbluefin#494 Hive-Plan: factory build backlog Hive-Spec: issue-494#implementation Signed-off-by: kylerankin <kylerankin@users.noreply.github.com>
…tar.* ibus-anthy.spec declares ibus-anthy.tar.gz.sum as Source1 and greps it in %prep, but the root .gitignore pattern dropped it on import. Add an exception and commit the file from Fedora dist-git 9d8e245. 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>
05174b2 to
da97c5f
Compare
Base moved (hive drift repair)Rebased onto current Addressing your CHANGES_REQUESTEDAll three of your concerns tracked a missing build dep; rebasing onto current main brought those providers in, so they are now satisfied:
With the providers present, the six engines legitimately sit in CI has not run on this rebased branch yet — it will on the next push/merge. Re-requesting review. 🐝 Hive Agent: |
|
Needs a decision, not a merge: the automerge lane will not merge this unattended. Head: verdict approved @da97c5f, 21/21 checks green, soak 30m0s elapsed. Clear the cause and the hive re-evaluates on its next pass; the stage moves to |
…16846) settleInPlaceAdoption yielded on any new head without App push evidence, so a maintainer rebasing or fixing an offline contributor's adopted PR handed it back to the absent author, who was then nudged, re-adopted and stalled again (projectbluefin/utah#919, Danathar's 2026-10-27 push; projectbluefin/utah-packages#804, #812, #822, hanthor's 2026-10-31 pushes). Before yielding, classify the GitHub-linked pusher (committer, or the author when the committer is web-flow or unlinked): the PR author, an unlinked identity, or a read-only third party still yields; an App bot or a person with write/maintain/admin (authorRepoPermission) is a takeover and the adoption follows the new head with no yield note and no nudge. An unreadable permission keeps the adoption for the sweep. Signed-off-by: James Reilly <jreilly1821@gmail.com> Co-authored-by: hivecommons-hive[bot] <280983584+hivecommons-hive[bot]@users.noreply.github.com>
Addressed at da97c5f: anthy-unicode, libpinyin and m17n-lib are on main, and CI ran 21/21 green. Before utah consumes typing-booster, check its runtime Requires (python3-enchant, python3-rapidfuzz, python3-pyxdg, hunspell-en, m17n-db) with check-repos.
|
Merged by hive contributor-lane auto-merge: CI green for ≥30m, reviewer verdict approve, @kylerankin holds the trusted tier in this hive's contributor registry. |
Summary
Implements the input-method half of the fonts-i18n block in #494 by importing five IBus engines into the factory recipe set.
Imported recipes (each: SHA-512 release archive, rawhide dist-git lock in
.hummingbird-upstream.json, entry inconfig/upstream-sources.jsonand.packit.yaml, moved from the fonts-i18n area to[resolved]in the factory build backlog):ibus-anthy1.5.18 (also covers theibus-anthy-pythonsubpackage)ibus-hangul1.5.5ibus-libpinyin1.16.6ibus-m17n1.4.41ibus-typing-booster2.31.1 — staged at 1 (itBuildRequiresenchant2;enchant2carries no explicit stage, so this is conservative ordering and harmless per docs/architecture.md)just checkandjust testare green (1472 passed, 2 skipped on the rebased base).verify_packitregisters all five; the source locks verify over https.ibus-libpinyinneeded an https mirror (its rawhide Source0 ishttp://sourceforge), andibus-m17nlocks the GitHub release asset rather than rawhide's https GitHub auto-archive URL (the auto-archive tarball is not byte-identical to the lookaside copy); the primaryurlis https with Fedora lookaside fallbacks where relevant.What I need from you
kasumi-common/kasumi-unicodeneed either a maintained https source or a linked maintainer decision to leave the backlog.pr =field on the six[resolved]records once this PR is numbered (the auditor reads onlyname, so this is cross-reference only).Still open (kept the batch open on purpose)
pending. Thekasumisource archive is unresolvable over https: the jaist.dl.sourceforge.jp mirror returns 410 Gone, thesourcesfile is MD5-only (rejected byvalidate.py), and no maintained GitHub mirror exists.kasumidist-git does not exist as a separate repo (both binaries are subpackages of thekasumisource package). Note thatkasumi-unicodeis nonetheless available from the paired Fedora build root, soibus-anthyis not blocked on it — it only means the factory does not rebuildkasumiitself.Hosted build, publication and runtime evidence for all six imported engines remains outstanding and is intentionally not claimed here.
Refs #494
— hive: backend=pi model=lemonade/ornith-1.5-35b-a3b-gguf-q6_k
Base moved (hive drift repair)
Rebased onto current
main(9eec2613) to clear base drift. Same repair the canary bot recommended: kept main's side for the four shared inventory files (factory-build-backlog.toml,upstream-sources.json,docs/architecture.md,docs/skills/factory-build-backlog.md), dropped the committedreports/factory-build-backlog.json(main no longer tracks derived counts/reports), and confirmed noyubikey-managerhunk leaked in — main already carries that recipe tree (#839). Net diff is now the five ibus engines plus their source-lock/backlog entries only.Addressing hanthor's CHANGES_REQUESTED
All three concerns tracked a missing build dep; rebasing onto current main brought those providers in, so they are now satisfied:
anthy-unicode-devel— provided byanthy-unicode(import: Rawhide libotf, libXaw, gd and anthy-unicode (m17n-lib blockers) #788), now on main; exactly theBuildRequiresinm17n-lib.spec.libpinyin-devel— provided bylibpinyin(feat: add the Chinese pinyin typing library recipe #796), now on main;ibus-libpinyinBuildRequiresit.m17n-lib/m17n-lib-devel— imported recipe-only in feat: add multilingual text library #856, now on main;ibus-m17nandibus-typing-boosterBuildRequireit.With the providers present, the six engines legitimately sit in
[resolved](recipe imported) and the "hosted build, publication and runtime proof remain open" note is accurate — that section records recipe import, not build/runtime evidence, which stays separate. CI has not run on this rebased branch yet; it will on the next push/merge.