qt5-feedback

Build times (auto-updated)

x86_64: 31s (2026-10-22 09:10 UTC, MAKEFLAGS=-j6)
i686: 45s (2026-10-22 09:10 UTC, MAKEFLAGS=-j6)
pentium4: 1m4s (2026-10-22 09:12 UTC, MAKEFLAGS=-j6)

Category: maintained (corrected 2026-10-22 from an initial, wrong adapted/ placement – see below). Flagged by https://buildmaster.archlinux32.org/checker/summary.html's consistency checker as one of 13 packages with a “missing state file” (i.e. dropped from official Arch’s extra/, listed as a candidate for AUR transfer). Not on AUR either (aur.archlinux.org/rpc/v5/info?arg=qt5-feedback -> resultcount: 0).

Corrected setup: full git history, maintained/ not adapted/ (2026-10-22)

First attempt just copied the final PKGBUILD as a flat file into adapted/qt5-feedback – wrong on two counts, caught by the user:

Lost the real git history. Arch’s own packaging repo (gitlab.archlinux.org/archlinux/packaging/packages/qt5-feedback, still reachable even though gone from the live extra/ tree) has the package’s actual commit history back to 2015 – 8 commits, every one of them Qt5 (never renamed/ported, so “the full history” and “the Qt5-era history” are the same thing here, nothing to trim). Matches the project’s existing precedent for a package dropped from its origin while its history is worth keeping ([[gsoap287]]/[[notion3]] – a submodule pointing at a preserved-history git remote), per memory/TOOLING_NOTES.md.
Wrong category. Since we’re the ones about to submit this fresh to AUR and become its real maintainer (not forking a still-alive upstream, which is what adapted/ is for), this belongs in maintained/, matching the gtksourceviewmm3/lunduke-paint-git precedent for “fresh package, we will be the AUR maintainer.”

Fixed: git cloned the full Arch packaging history (git log shows all 8 commits, authors/dates preserved untouched), placed it at maintained/qt5-feedback, and git remote set-url origin repointed at ssh://aur@aur.archlinux.org/qt5-feedback.git (not yet pushed – that repo doesn’t exist until the first push creates it, which is the user’s step, not mine, per the no-push policy). # Maintainer: comment updated to the user, with the two real prior contributors (Felix Yan, Evangelos Foutras) kept as # Contributor: lines rather than erased.

Bonus find from the real history: arch=() included i686 until a 2017 Arch-wide mass removal (adeedda, “Remove i686 from arch=()”, Evangelos Foutras) – confirms this package used to build fine on 32-bit and the only reason it stopped was Arch’s own i686 policy, not anything about this package.

What changed vs the recovered history

arch=('x86_64') -> arch=('x86_64' 'i686' 'pentium4') (restoring i686, adding pentium4).
pkgrel 5 -> 6 (plain integer bump, not the project’s internal X.Y “local deviation” scheme – this is a real AUR package now, not a silently-diverged local fork, so it just follows normal AUR versioning).
Build is a plain qmake-qt5 && make, depends only on qt5-declarative – confirmed available on x86_64/i686/pentium4 via pacman -Si before committing to the build.
Already built+published once under the old, wrong adapted/ setup (x86_64/i686/pentium4, 0.0.0.20150918-5.1) before being corrected – needs a rebuild+republish under the new maintained/ path/pkgrel 6 once the user has pushed to AUR, to replace that interim version.

Why it was dropped (per the user’s ask for a worth-maintaining assessment)

See the dropped-packages research the user asked for alongside this one – the general pattern for the other 12 flagged packages was either “superseded by a renamed/newer package” or “upstream project itself abandoned/archived”. qt5-feedback specifically: a niche Qt Mobility-era tactile-feedback addon, originally for mobile/Sailfish-style haptics, with no real desktop-Linux use case and no upstream activity since 2015 – low usage is the most likely reason nobody picked it up for AUR before now, not a build problem.