Repository navigation
feat(terminal): import ptyxis and vte291 recipes for Hummingbird - #225
sramkrishna wants to merge 8 commits into
Conversation
There was a problem hiding this comment.
style
- The two imports carry Fedora dist-git automation files that no other recipe in the tree keeps:
packages/ptyxis/.packit.yamlandpackages/vte291/.packit.yaml(both declaringkoji_build/bodhi_updatejobs againstfedora-stable, meaningless here),README.packit, andmonitoring.toml(only 1 of the existing 352 recipes has one; none have the other two — comparepackages/gtk4/andpackages/tecla/, which strip them). Recommend dropping them so the imports match the established recipe shape and a nested.packit.yamlcan never shadow the rendered top-level config.
Verified independently at head 0b2a71c: both source locks' SHA-512/SHA-256 match the gnome.org tarballs and .sha256sum files; stage placement (vte291=5 > gtk4=4; ptyxis=7 > libadwaita=6, libportal/vte291=5) matches config/upstream-sources.json; inventory counts 354/353 match the 354 packages/ dirs; vte291's runtime deps (gnutls, libicu, pcre2, systemd-libs from Hummingbird; gtk3, pango, fribidi from the factory) and ptyxis's hicolor-icon-theme (factory recipe) resolve; vte's fmt/simdutf/fast_float BuildRequires come from the Fedora 44 buildroot per docs/targeting-hummingbird.md.
No findings from: correctness, security, intent-alignment, docs-currency.
Confidence: 3/5 (needs attention) — 1 perspective requested changes
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
There was a problem hiding this comment.
correctness
- (low)
packages/vte291/vte291.spec:32,45unconditionally BuildRequirespkgconfig(fmt) >= 11.0.0andpkgconfig(simdutf) >= 7.2.1, and%prepdeletessubprojects/fmt/andsubprojects/simdutf/even in the bundled-fast_float (rhel) branch (vte291.spec:148-153), so both must resolve from Hummingbird or the stage-5 build fails. Neither is a factory recipe (config/upstream-sources.jsonhas nofmt/simdutfentry), and the PR body's dependency audit covers only vte291's runtime Requires — build deps (fmt,simdutf,gi-docgen,vala, and ptyxis'sitstool/appstream-util) were not measured against the Hummingbird repo the way the runtime set was. A failed stage build will surface it, but the PR's own "measure, do not assume" rule wasn't applied here.
Verified: both specs are byte-identical to the declared Fedora dist-git commits (f8738e9, 1c4e5b9); SHA-512 locks match dist-git sources and the upstream .sha256sum values match independently downloaded tarballs; stage math (gtk4=4, libportal=5, libadwaita=6) matches config/upstream-sources.json; .packit.yaml entries match the tools/render_packit_config.py output shape; the four inventory assertions move consistently (354/353); the skill doc is indexed in docs/SKILL.md.
No findings from: security, intent-alignment, style, docs-currency.
Confidence: 5/5 (safe)
— hive: agent=reviewer backend=copilot model=claude-fable-5 copilot=1.0.78
hanthor
left a comment
There was a problem hiding this comment.
Requesting changes: unaddressed style finding — strip nested .packit.yaml/README.packit/monitoring.toml. No CI or human review yet.
Utah's payload image ships no terminal emulator: ptyxis, gnome-console, kgx and vte291 are in none of the repositories the image enables, and Hummingbird's public repository carries no VTE library at all (20,351 packages checked, zero vte* names). A terminal recipe alone therefore builds but installs nowhere, so this imports the VTE stack underneath it in the same change. - Import ptyxis 50.1 and vte291 0.84.1 from Fedora Rawhide dist-git, with provenance recorded in .hummingbird-upstream.json. - Lock both to their download.gnome.org release tarballs: SHA-512 pinned against the dist-git sources manifests, SHA-256 against the upstream .sha256sum files, Fedora lookaside carried as fallback_urls only. - Stage vte291 at 5 (above gtk4) and ptyxis at 7 (above libadwaita, libportal and vte291), because stage N resolves only against stages below it. - Regenerate .packit.yaml, and move the inventory count assertions from 352 to 354 (351 to 353 for the Hummingbird-supplied skip count). - Record the measurement and the resulting ordering in docs/skills/terminal-emulator-recipes.md, indexed in docs/SKILL.md. Resolves projectbluefin#224 Assisted-by: deepseek/deepseek-v4.1-flash via Claude Code Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> Signed-off-by: sramkrishna <sramkrishna@users.noreply.github.com>
The imported recipes carried Fedora dist-git's own automation config: .packit.yaml declaring koji_build/bodhi_update jobs against fedora-stable, README.packit, and monitoring.toml. None of that runs in this factory -- Packit here is driven by the rendered top-level .packit.yaml -- so the files are dead config that invites confusion about which one is live. Strip them so packages/ptyxis and packages/vte291 match the shape every other import already has (packages/gtk4, packages/tecla). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
8eda3a5 to
537b599
Compare
|
Rebased onto current main ( |
hanthor
left a comment
There was a problem hiding this comment.
Thanks for stripping the nested packit and monitoring files. Stage placement is right: vte291 at 5 (gtk4 is at 4) and ptyxis at 7 (libadwaita at 6, libportal at 5). Two things are still open:
- Build deps aren't proven:
packages/vte291/vte291.spec:32,45BuildRequirespkgconfig(fmt) >= 11andpkgconfig(simdutf) >= 7.2.1unconditionally, plusfast_float-devel(:56),gi-docgen(:48) andvala(:53). None of these is a factory recipe, and the dependency audit in the description only covers runtime Requires. Please showrpmspec --buildrequiresplusdnf repoqueryin the build root, or link a green srpm/build run. - Counts: the four count tests expect 375, but
mainis at 386. For merge order in this cluster, #239 goes first, then this PR, then #246, then #233. Please rebase after #239 lands.
Nit: packages/ptyxis/README.md is a two-line stub.
|
Verification & Status:
Next action: Author (@sramkrishna) must rebase onto |
# Conflicts: # tests/test_package_inventory.py # tests/test_packit_srpm.py # tests/test_render_packit_config.py # tests/test_source_inventory.py
Assisted-by: Gemini via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
castrojo
left a comment
There was a problem hiding this comment.
Rebased and merged origin/main onto fork branch sramkrishna:feat/ptyxis-terminal (commit c2ce056), resolving mechanical count conflicts across tests/test_*.py and updating docs/architecture.md to 402 packages. All factory contract checks and unit tests pass locally. Ready for @hanthor to re-review and dismiss or update the changes-requested status.
Danathar
left a comment
There was a problem hiding this comment.
I re-checked @hanthor's 09-25 points against c2ce056, which is a merge of main plus a docs count update.
- Counts: fixed. The four count tests now say 402 (401 for the Hummingbird-supplied skip), and
just validateprintsvalidated 402 source RPMs. #239 merged on 09-25, so the merge order he asked for holds. Locally,just factory-check validate workflow-quoting runtime-contractall pass, andpytest testsgives 654 passed. - Build deps: not fixed. The thread has no
rpmspec --buildrequiresor repoquery output, and there is no build run. Canary skips fork PRs, so this head has no CI at all. I ran the queries myself against the build roottools/mock_config.pysets up (Fedora 44 + updates + Hummingbird). The build side looks fine.fmt-devel-11.2.0,simdutf-devel-7.2.1andfast_float-devel-8.3.0come from Fedora 44; Hummingbird also hasfmt-devel-12.1.0. gi-docgen, vala, itstool, appstream-util, gperf and liblz4 are already BuildRequires of recipes on main. - New, from the same check: the runtime side has a gap. Fedora 44's own
vte291requireslibsimdutf.so.24(dnf5 repoquery --releasever=44 --requires vte291). Hummingbird has no simdutf package at all (dnf5 repoquery --repofrompath=hum,https://packages.redhat.com/api/pulp-content/public-hummingbird/x86_64/ 'simdutf*'returns nothing), and this repo has nopackages/simdutf. Every other soname that vte291, vte291-gtk4 and ptyxis need is either in Hummingbird or has a recipe here. My theory, not tested with a build: vte291-gtk4 built here will linklibsimdutf.so.24and fail the Hummingbird-only install check, and ptyxis will fail with it, until simdutf gets a recipe below stage 5. A green srpm/build run would settle it. - Nit: not changed.
packages/ptyxis/README.mdis still the two-line stub.
|
Checked 40308b0. The two new commits (05f038e, 40308b0) are merges of main. The only change to this PR's own diff is the docs/architecture.md line, which now follows main's "(399 rawhide, 3 upstream)" format. So my 09-26 points still stand:
Locally on 40308b0, python3 -m unittest discover -s tests passes. The branch merges cleanly with main 896ad14. It also merges with #198 except for a one-line docs/SKILL.md index conflict, and with both applied the #198 audit still reports 0 unclassified locks. |
|
Shepherding note: this now conflicts with main after the recent merges (#296 feeds rework touches nearby config). Needs a manual rebase before CI can validate it. |
There was a problem hiding this comment.
HUMAN DECISION NEEDED — @sramkrishna: confirm libsimdutf.so.* resolves on the Hummingbird bootc base (or add a simdutf recipe) before this stack is published; the factory has no gate that would catch it.
correctness
packages/vte291/vte291.spec:45,152— vte 0.84.1meson.build:672makessimdutfarequired: truedependency and the spec deletessubprojects/in the non-%rhelpath, solibvte-2.91*.solinkslibsimdutf.so.Ndynamically and the RPM gets an auto-generatedlibsimdutf.so.N()(64bit)Requires. Hummingbird's Pulp repository (queried today: 21,662 packages, 3,520 names) hasfmtbut zerosimdutf*, and there is nopackages/simdutf/recipe. Unless the bootc-os base RPM DB carries it (I could not inspect that image),vte291— and thereforeptyxis— hits exactly the "installs nowhere" failure #224 is about. Nothing catches this pre-publish:config/runtime-contract.toml[utah]lists neither package, so the Hummingbird-only transaction never resolves them. The claim atdocs/skills/terminal-emulator-recipes.md:54-56covers only the spec's explicitRequires:, not soname autodeps;fmtis fine (Pulp shipsfmt),simdutfis not.
docs-currency
docs/skills/terminal-emulator-recipes.md:60states "Stage N resolves against stages< Nonly" as the ordering rule.docs/architecture.md:163-167andtools/build_graph.py:4-7,19-23say the opposite: waves are solved from BuildRequires andstageis only a tie-breaker inside a BuildRequires cycle. A new procedure doc should not re-teach the retired model; the stage 5/7 values are harmless, the rationale is wrong.
intent-alignment
- PR body says inventory counts move 352 → 354 (351 → 353); the diff moves 400 → 402 / 399 → 401 (
tests/test_package_inventory.py:19,tests/test_source_inventory.py:71). Body is stale, diff is right.
No findings from: security, style.
Confidence: 2/5 (needs attention) — 1 high finding; 1 medium finding
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
Status at e02c630: nit and counts are fixed; the branch conflicts with main again. The Canary failure is the fork guard, not this diff. The open question is libsimdutf.so.24: Hummingbird has none and no recipe exists. Please add a simdutf recipe below stage 5 or show it resolves in the base. A maintainer needs to push the branch for canary. |
hanthor
left a comment
There was a problem hiding this comment.
Thanks for sticking with this. It's still blocked on two things:
- simdutf.
packages/vte291/vte291.spec:45BuildRequirespkgconfig(simdutf) >= 7.2.1, and:150deletes the bundledsubprojects/simdutf/, so the builtlibvte-2.91*will Requirelibsimdutf.so.N()(64bit). Neither Hummingbird nor this factory provides that library (nopackages/simdutfon main). Please add asimdutfrecipe in a stage below vte291 (Source0 from upstream, SHA-512 locked), or show a repoquery proving it resolves on the Hummingbird base. - Rebase. The branch conflicts with main. After rebasing, update the recipe counts in
docs/architecture.mdand the four count tests to match main plus your additions (main is at 402).
A maintainer will push the branch for an in-repo canary after that, because fork PRs can't run it.
Evidence: mergeable=CONFLICTING; packages/vte291/vte291.spec:45 BuildRequires: pkgconfig(simdutf) >= 7.2.1, :150 rm -rf subprojects/simdutf/; no packages/simdutf on main (git ls-tree); no simdutf PR exists; Danathar 10-02 and Hive 09-29 reviews raise the same point
🤖 Generated with Claude Code
Assisted-by: GPT-6 via GitHub Copilot Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
|
Superseded by maintainer-owned #413, which retains the repaired recipe imports, adds the missing simdutf provider, and runs the native required Canary. The explicit three-package compilation is run 37572120300; merge remains conditional on passing checks. |
Utah's payload image ships no terminal emulator at all.
ptyxis,gnome-console,kgxandvte291resolve from none of the repositories theimage enables, so this imports the two recipes that close the gap.
Why two recipes and not one
Hummingbird's public repository carries no VTE library, so a terminal
recipe on its own builds but installs nowhere — the "installs nowhere" failure
mode
docs/targeting-hummingbird.mddescribes. Measured 2026-09-22 againstpackages.redhat.com/api/pulp-content/public-hummingbird/x86_64/: 20,351packages, and zero names matching
vte*,ptyxis,gnome-consoleorkgx.VTE is desktop stack, and Hummingbird is a base-OS overlay.
Both
ptyxisandgnome-consolelinklibvte-2.91-gtk4.so.0, which only thevte291-gtk4subpackage provides, sovte291lands first and the terminalgoes above it.
What changed
packages/ptyxis/f8738e9; provenance in.hummingbird-upstream.jsonpackages/vte291/1c4e5b9; same shapeconfig/upstream-sources.jsonstage5 and 7.packit.yamltools/render_packit_config.py --writetests/test_*.pydocs/skills/terminal-emulator-recipes.mddocs/SKILL.mdStage placement
Stage N resolves only against stages below it, so each recipe sits strictly
above its highest factory-built dependency:
vte291→ stage 5: needsgtk4(stage 4);gtk3,pango,fribidi,icu,gobject-introspectionare all stage 0.ptyxis→ stage 7: needslibadwaita(stage 6),libportaland the newvte291(stage 5),gtk4(stage 4).gnome-consolewould occupy ptyxis's slot; it is not imported here because theissue asks for ptyxis first for Bluefin parity.
Source verification
Neither lock points its
urlat Fedora infrastructure — Fedora is the fallbacktransport, never the primary source.
sources.sha256sumvte-0.84.1.tar.xzptyxis-50.1.tar.xzBoth
fallback_urlswere confirmed to answer200before being written.Validation
pre-commitis not installed in the authoring environment, so the hooks werereproduced by hand: JSON and YAML parse, no file added is over the size limit,
and the three local
language: systemhooks were run directly. Thepackages/exclusions for
trailing-whitespaceandend-of-file-fixermean the importedspec bytes are untouched, as intended.
What this does not do
The consumer side is deliberately out of scope here:
projectbluefin/utahmoves ptyxis into its install contract and drops the entry from
[unavailable]once these RPMs are published. Utah cannot pin a digest that does not exist
yet, and this repository does not edit
projectbluefin/utah.Fixes #224
— hive: backend=litellm model=deepseek/deepseek-v4.1-flash
🐝 Hive Agent:
contributor| SHA:0b2a71c