fvwm

• Path: arch/maintained/fvwm
• Build failure during the 2026-07-07 “build all packages for x86_64” private-repo release pass (see repo_release_logs/maintained_fvwm.log).
• Same systemic intra-tree dependency gap as memory/kdstatemachineeditor.md (not KDE5 this time): needs libstroke, which is a package in this tree (maintained/libstroke) but alphabetically after fvwm so it hadn’t been built+published yet when this run reached fvwm. The other missing deps listed (fribidi, libxpm, librsvg, libxinerama, libxcursor, python-xdg) are real official packages that would resolve fine on their own – only libstroke actually fails resolution.
• Should resolve on a retry once both (a) libstroke has been built+published, and (b) rebuilt via the new scripts/archlinuxaba-x86_64-build chroot flavor (see memory/TOOLING_NOTES.md’s update on the intra-tree dependency fix) – not retried yet as part of this pass.
• 2026-08-13: widened arch=() to i486 i686 pentium4 x86_64, pkgrel 6 -> 7 (packaging-only, no upstream change – see PKGREL_VS_PKGVER.md), as the test case for wiring archlinuxaba-<arch>-build up for the 32-bit arches (see memory/TOOLING_NOTES.md’s 2026-08-13 entry). Required widening libstroke (a depends) to the same four arches first – see memory/libstroke.md. Built and published cleanly for i486/i686/ pentium4 on the first pass; x86_64 initially failed with pacman refusing a locally cached libstroke-0.5.1-11-x86_64.pkg.tar.zst as “corrupted (invalid or corrupted package (checksum))” – a stale package-cache artifact left over from rebuilding/republishing libstroke twice during that same session (non-reproducible SOURCE_DATE_EPOCH between builds), not a fvwm-specific issue; a bare retry of just the x86_64 arch built and published fine. No fvwm PKGBUILD/source changes needed beyond arch=() and pkgrel.