All checks were successful
Mirror to GitHub / mirror (push) Successful in 3s
Boot-critical (each confirmed with a real boot/install cycle, not just code review): - Live ISO's liveuser shell is zsh, but Hyprland autostart was written to .bash_profile (never runs) — moved to .zprofile, and switched to start-hyprland (Hyprland's own watchdog wrapper; raw `exec Hyprland` is no longer the recommended launch method). - copytoram (self-enables on most real hardware: non-optical boot + image < 4GiB + enough free RAM) unmounts /run/archiso/bootmnt, which unpackfs.conf and bos-copy-kernel both hardcoded as their source — broke the installer outright on real hardware, confirmed by forcing copytoram=y. Added a resolver step for unpackfs, and switched the kernel copy to /usr/lib/modules/$(uname -r)/vmlinuz (part of the live squashfs itself, unaffected by copytoram). - BIOS installs got no bootloader — post-install.sh only ever ran the UEFI grub-install path despite BOS shipping bios.syslinux. Added a BIOS branch with disk auto-detection. - @snapshots/@log/@cache were never real: iso/partition.conf's btrfsSubvolumes key isn't part of this Calamares version's partition module schema at all (same class of bug as the userShell fix below — silently ignored). Calamares only natively creates @ and @home. post-install.sh now creates the three subvolumes by hand after unpackfs, migrates existing /var/log + /var/cache content into them before mounting over, and adds the fstab entries — verified end to end on real hardware, including grub-btrfs generating bootable snapshot menu entries. The unmount step in the existing snapper create-config dance also gained retry + lazy-unmount fallback after a real chroot run hit a transient busy-mount race. - Default shell was bash instead of zsh post-install: users.conf's top-level `userShell` key isn't part of this Calamares version's users module schema either — the real key is nested (user.shell). - graphical-session.target ships RefuseManualStart=yes (systemd convention), so the earlier attempt to activate it from hyprland.lua silently failed and breadclipd (WantedBy=graphical-session.target) never started. Starts breadclipd.service directly instead. - /etc/os-release was never set (showed "Arch Linux"); live boot never had quiet/splash/plymouth wired in (raw kernel scroll the whole time) despite BOS already shipping a complete bread-logo+spinner plymouth theme for the installed system. Also: generalized build-local.sh's per-service skel baking (previously only breadd.service was hand-committed; breadbox-sync/breadmill/ breadclipd never shipped), added the four new bakery packages to BREAD_BINS, removed the redundant cliphist/fzf clipboard pipeline in favor of breadclip, mirrored fastfetch's bread-logo config into skel, and fixed a stale bos-update comment.
142 lines
7.4 KiB
Bash
Executable file
142 lines
7.4 KiB
Bash
Executable file
#!/usr/bin/env bash
|
|
# Local BOS ISO build for hermes (native Arch — no container needed).
|
|
#
|
|
# Builds straight from the working tree in ./iso. Two speedups vs the hestia
|
|
# container build:
|
|
# * runs natively (hermes is Arch; hestia needed a dockerised Arch)
|
|
# * no 2 GB scp afterwards — the ISO lands here, where we test it
|
|
#
|
|
# FAST_BUILD=1 zstd squashfs instead of xz -9e: compresses many times
|
|
# faster at the cost of a slightly larger image. Dev only.
|
|
#
|
|
# Usage: sudo ./build-local.sh # release-quality xz
|
|
# sudo FAST_BUILD=1 ./build-local.sh # fast dev iteration
|
|
set -euo pipefail
|
|
|
|
REPO="$(cd "$(dirname "$0")" && pwd)"
|
|
# WORK defaults to /tmp, but on hermes /tmp is a 16 GB tmpfs — a full xz build
|
|
# (uncompressed rootfs + squashfs + work copies) can exhaust it mid-build. Allow
|
|
# pointing it at the NVMe instead: WORK=/home/.../bos-work sudo ./build-local.sh
|
|
WORK="${WORK:-/tmp/bos-work}"
|
|
OUT="${OUT:-$REPO/out}"
|
|
|
|
# Build against a throwaway copy of the profile so the working tree stays clean
|
|
# when FAST_BUILD / the registry rewrite mutate profile files.
|
|
STAGE=/tmp/bos-iso-stage
|
|
rm -rf "$STAGE" && cp -a "$REPO/iso" "$STAGE"
|
|
|
|
# Rewrite the [breadway] pacman repo URL to the fastest reachable address.
|
|
# CI_BUILD=1 — container runs on hestia with --network=host; localhost:3002 is direct
|
|
# default — building on hermes; git.breadway.dev is flaky from there, use Tailscale
|
|
# Only ever rewrites the staged copy, never the committed pacman.conf.
|
|
if [ "${CI_BUILD:-0}" = "1" ]; then
|
|
sed -i 's#https://git.breadway.dev/api/packages/Breadway/arch/os#http://localhost:3002/api/packages/Breadway/arch/os#' "$STAGE/pacman.conf"
|
|
else
|
|
sed -i 's#https://git.breadway.dev/api/packages/Breadway/arch/os#http://100.66.238.26:3002/api/packages/Breadway/arch/os#' "$STAGE/pacman.conf"
|
|
fi
|
|
|
|
if [ "${FAST_BUILD:-0}" = "1" ]; then
|
|
echo "=== FAST_BUILD: squashfs -> zstd level 6 ==="
|
|
sed -i "s#^airootfs_image_tool_options=.*#airootfs_image_tool_options=('-comp' 'zstd' '-Xcompression-level' '6' '-b' '1M')#" "$STAGE/profiledef.sh"
|
|
fi
|
|
grep airootfs_image_tool_options "$STAGE/profiledef.sh"
|
|
|
|
# --- Bake this laptop's bakery-installed bread ecosystem into /etc/skel -------
|
|
# The bread apps are managed by bakery (which fetches release binaries from
|
|
# GitHub), not pacman. bakery needs DNS at install time, which the live/installed
|
|
# image doesn't have — so instead of running bakery on the target, we copy the
|
|
# exact binaries + bakery manifest this laptop already has into skel. Every user
|
|
# created from skel (the live user and the installed user) then gets the same
|
|
# versions `bakery list` reports here, fully offline. Copied at build time so the
|
|
# binaries never bloat the git repo and always track the current bakery state.
|
|
BREAD_BINS=(bakery bread breadd breadman breadbar breadbox breadbox-sync breadcrumbs breadpad breadpaper bread-theme breadmon breadsearch breadmill breadclip breadclipd breadshot)
|
|
LAPTOP_HOME="${LAPTOP_HOME:-$(getent passwd "${SUDO_USER:-$USER}" | cut -d: -f6)}"
|
|
BAKERY_BIN="$LAPTOP_HOME/.local/bin"
|
|
BAKERY_STATE="$LAPTOP_HOME/.local/state/bakery"
|
|
BAKERY_CACHE="$LAPTOP_HOME/.cache/bakery"
|
|
SKEL="$STAGE/airootfs/etc/skel"
|
|
echo "=== baking bakery bread ecosystem from $LAPTOP_HOME ==="
|
|
install -d -m 0755 "$SKEL/.local/bin" "$SKEL/.local/state/bakery" "$SKEL/.cache/bakery"
|
|
for b in "${BREAD_BINS[@]}"; do
|
|
install -m 0755 "$BAKERY_BIN/$b" "$SKEL/.local/bin/$b"
|
|
done
|
|
install -m 0644 "$BAKERY_STATE/installed.json" "$SKEL/.local/state/bakery/installed.json"
|
|
# bakery fetches its package index from dl.breadway.dev (then a GitHub fallback),
|
|
# but falls back to a cached index when both are unreachable. With no network/DNS
|
|
# in the live/installed image, even `bakery list` errors unless that cache exists,
|
|
# so bake it in too — then bakery works fully offline (list/info from cache;
|
|
# install/update still need network, as expected).
|
|
install -m 0644 "$BAKERY_CACHE/index.json" "$SKEL/.cache/bakery/index.json"
|
|
echo "baked: $(ls "$SKEL/.local/bin")"
|
|
|
|
# --- Bake systemd user services for bakery-managed bread packages -----------
|
|
# Historically only breadd.service was hand-committed to skel; every other
|
|
# bakery package's service (breadbox-sync, breadmill, breadclipd, ...) was
|
|
# silently left out, so those daemons never start on a fresh install/live
|
|
# boot until the user re-runs `bakery install` (which needs network).
|
|
# Generalize from the same source of truth as the binary bake above: read
|
|
# the services this laptop's bakery actually installed, copy each unit file
|
|
# into skel with ExecStart rewritten from this laptop's literal home path to
|
|
# the portable `%h` specifier, and recreate whichever *.target.wants enable
|
|
# symlink bakery created locally. Units already committed by hand (breadd.service
|
|
# carries a RuntimeDirectoryPreserve=yes fix not yet upstreamed — see bread-release-build
|
|
# notes) are left alone rather than overwritten.
|
|
echo "=== baking bakery service units into skel ==="
|
|
SYSTEMD_USER_DIR="$LAPTOP_HOME/.config/systemd/user"
|
|
SKEL_SYSTEMD="$SKEL/.config/systemd/user"
|
|
mapfile -t SERVICE_UNITS < <(python3 -c "
|
|
import json
|
|
with open('$BAKERY_STATE/installed.json') as f:
|
|
d = json.load(f)
|
|
for pkg in d.get('packages', d).values():
|
|
for s in pkg.get('services', []):
|
|
print(s)
|
|
")
|
|
for unit in "${SERVICE_UNITS[@]}"; do
|
|
if [[ -f "$SKEL_SYSTEMD/$unit" ]]; then
|
|
echo " $unit already committed in skel, leaving as-is"
|
|
continue
|
|
fi
|
|
src="$SYSTEMD_USER_DIR/$unit"
|
|
if [[ ! -f "$src" ]]; then
|
|
echo " warning: $unit not found at $src, skipping"
|
|
continue
|
|
fi
|
|
install -d -m 0755 "$SKEL_SYSTEMD"
|
|
sed "s#ExecStart=$LAPTOP_HOME/.local/bin/#ExecStart=%h/.local/bin/#" "$src" > "$SKEL_SYSTEMD/$unit"
|
|
for wants_dir in "$SYSTEMD_USER_DIR"/*.target.wants; do
|
|
[[ -L "$wants_dir/$unit" ]] || continue
|
|
target_name="$(basename "$wants_dir")"
|
|
install -d -m 0755 "$SKEL_SYSTEMD/$target_name"
|
|
ln -sf "../$unit" "$SKEL_SYSTEMD/$target_name/$unit"
|
|
done
|
|
echo " baked $unit"
|
|
done
|
|
|
|
# mkarchiso resets every airootfs file to 0644, so executables must be declared
|
|
# in profiledef.sh's file_permissions array or they ship non-executable and the
|
|
# exec-once launches fail with "permission denied". Inject a 0755 entry for each
|
|
# baked binary right after the array opener (keeps the binary list in one place).
|
|
perm_file="$(mktemp)"
|
|
for b in "${BREAD_BINS[@]}"; do
|
|
printf ' ["/etc/skel/.local/bin/%s"]="0:0:755"\n' "$b" >>"$perm_file"
|
|
done
|
|
sed -i "/^file_permissions=(/r $perm_file" "$STAGE/profiledef.sh"
|
|
rm -f "$perm_file"
|
|
echo "=== file_permissions after injection ==="; grep -A14 '^file_permissions=(' "$STAGE/profiledef.sh"
|
|
|
|
# Pin one timestamp for the whole build. Without this, mkarchiso derives the
|
|
# boot-config UUID (%ARCHISO_UUID%) when it starts and the iso9660 volume UUID
|
|
# when xorriso writes the image at the end — on a slow build these diverge by
|
|
# the build duration, so the initramfs searches /dev/disk/by-uuid/<wrong-uuid>,
|
|
# never finds the medium, and drops to a recovery shell. Fixing the epoch makes
|
|
# both derive from the same instant (and makes builds reproducible).
|
|
export SOURCE_DATE_EPOCH="${SOURCE_DATE_EPOCH:-$(date +%s)}"
|
|
echo "SOURCE_DATE_EPOCH=$SOURCE_DATE_EPOCH ($(date -u -d "@$SOURCE_DATE_EPOCH" +%Y-%m-%d-%H-%M-%S-00))"
|
|
|
|
echo "=== running mkarchiso ==="
|
|
rm -rf "$WORK" && mkdir -p "$OUT"
|
|
mkarchiso -v -w "$WORK" -o "$OUT" "$STAGE"
|
|
|
|
echo "=== RESULT ==="
|
|
if ls -lh "$OUT"/*.iso 2>/dev/null; then echo "ISO BUILT OK -> $OUT"; else echo "ISO BUILD FAILED"; exit 1; fi
|