Skip to content

Clean up Ally shell writes, TDP wake, and dead packaging - #80

Merged
DanielJBurbridge merged 2 commits into
feat/ally-parity-integration-20261111from
chore/ally-cleanup-20261117
Nov 17, 2026
Merged

DanielJBurbridge merged 2 commits into
feat/ally-parity-integration-20261111from
chore/ally-cleanup-20261117

Conversation

@DanielJBurbridge

@DanielJBurbridge DanielJBurbridge commented Nov 17, 2026 •

Copy link
Copy Markdown
Owner

What changes

Cleanup from the integ audits. No Ally deploy, and shell/plugins/lock is untouched.

  • A7. omarchy-setup-rog-ally-keyboard no longer rewrites ~/.config/omarchy/shell.json. --remove, and a run that is not an RC71L, drop the bar button with omarchy-plugin-disable while the shell is up. If the shell is down, the layout is left alone. The command says to drop the button the next time the shell runs only when that button is already in shell.json (a read-only check) or this setup's launcher was installed. Otherwise it stays silent.
  • A8. migrations/1794830400.sh installs qt6-virtualkeyboard only when omarchy-hw-rog-ally succeeds. The stamp is still 1794830400. The package stays in install/omarchy-base.packages because etc/sddm.conf.d/10-wayland.conf sets GreeterEnvironment for every machine's greeter, so a fresh laptop install still needs it.
  • A9. Wake TDP reassert is only omarchy-ally-tdp-resume.service ExecStop. The shim no longer subscribes to logind PrepareForSleep.
  • A10. migrations/1795075200.sh is a no-op. 1794843321 is the catch-up. The hardware leaf and omarchy-setup-gaming-session remain the retry.
  • packaging/rog-ally-tdp/ is removed. The shim reads $OMARCHY_PATH/default/ally-tdp/, and the hardware leaf already runs omarchy-ally-tdp-install. packaging/ is gone. The gaming-session install scriptlet still runs omarchy-ally-tdp-install session as the packaged-install fallback. It does not install the deleted package.
  • omarchy-gaming-session is 1.0.0-6. sha256sums are the local files, and depends includes iproute2 (ss in default/gaming/steam-ready.sh). L155 is fixed. L200 and the handoff name the sudoers package as omarchy-gaming-session 1.0.0-5 or later.

L210–L219 is claimed and empty. Nothing from this cleanup was left open.

Model: grok-4.7

Verification

  • bash test/shell.d/hw-rog-ally-test.sh passed, including no shell.json write when the shell is down, a quiet removal when the button was never added, a note when the layout or this setup's launcher has it, no qt6-virtualkeyboard install off the Ally, and 1795075200 as a no-op.
  • bash test/shell.d/ally-tdp-test.sh passed, including the shim having no on-sleep reassert and the vendored interface living only in default/ally-tdp/.
  • bash test/shell.d/handheld-autologin-test.sh passed, including no SKIP in the gaming-session PKGBUILD, iproute2 in depends, and checksums matching the local files.
  • The desktop-keyboard DMI gate stays quiet on a non-Ally with no button installed.
  • Not run on an Ally.

Screenshots

No visible change: this is command, migration, and package cleanup. The bar button is still removed through the shell when it is running.

Open in Web Open in Cursor 

Stop the keyboard setup from rewriting shell.json, install the greeter keyboard package only on an Ally, and leave wake TDP reassert to the resume unit. Make the earlier power-button migration a no-op, drop the unused TDP package, and pin the gaming-session checksums.

Model: grok-4.7
Co-authored-by: Daniel Burbridge <DanielJBurbridge@users.noreply.github.com>

@DanielJBurbridge DanielJBurbridge left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

dev-red review: FAIL at 3f1222e

Blocker B1 (Tests are red, and it's a non-handheld regression): Actions test fails on test/shell.d/desktop-keyboard-test.sh. It reproduces locally. On a machine that isn't an Ally (rog-laptop case), omarchy-setup-rog-ally-keyboard now prints The shell is not running, so the Keyboard button stays on the bar. Drop it the next time the shell runs: omarchy plugin disable omarchy.desktop-keyboard (hardware_acceptance, test/helpers/desktop-keyboard-process.py:100). The old strip_keyboard_widget did nothing when there was no button. The new --remove / non-RC71L path tells every machine to remove a button that was never added. Say it only when the button is actually on the bar: a read-only jq check of shell.json is fine, because reading isn't writing. Or say it only when this setup's launcher was installed. Keep A7's "never write shell.json".

Everything else checks out:

  • A9: the shim no longer subscribes to PrepareForSleep, and pkexec is --set only. omarchy-ally-tdp-resume.service (ExecStop=... --reassert, WantedBy=sleep.target) is installed and enabled by install_hardware, so wake reassert still happens without the session.
  • A8: the 1794830400 hardware gate is fine, since the package stays in base for every greeter. A10: 1795075200 is a no-op, and 1794843321 is the one catch-up.
  • packaging/ is gone. The only references left are the test asserting it isn't shipped. The TdpLimit1 XML and NOTICE.steamos-manager are still in default/ally-tdp/, so the MIT attribution is kept.
  • The PKGBUILD's 7 sha256sums match the files at this head. iproute2 is in depends, so L155 is right.
  • There's a Model trailer (grok-4.7), and no follow-up lines are removed except edits to L155 and L190.
  • Low (fix with B1): L200 and HANDOFF still say omarchy-gaming-session 1.0.0-5. Make it "1.0.0-5 or later", so a 1.0.0-6 deploy plainly follows the one pacman -U rule.

A delta check is next: the B1 fix, the 1.0.0-5 wording and green Tests.

A stopped shell stays quiet unless the button is already in shell.json or this setup's launcher was installed. L200 and the handoff cover gaming-session 1.0.0-5 or later.

Model: grok-4.7
Co-authored-by: Daniel Burbridge <DanielJBurbridge@users.noreply.github.com>

@DanielJBurbridge DanielJBurbridge left a comment

Copy link
Copy Markdown
Owner Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

dev-red delta review: PASS at 84ebcb0

  • B1 is fixed. The shell-down note now only names the button when keyboard_button_in_shell (a read-only jq -e, where missing or invalid JSON counts as no button) finds it, or when this setup installed the launcher. Nothing writes shell.json. desktop-keyboard-test passes locally.
  • L200 and HANDOFF now say "1.0.0-5 or later".
  • Integ e742281 is an ancestor of this head. The new commit has a Model: grok-4.7 trailer.
  • Actions test and i18n-check are both green on this head.

Low follow-up (not blocking): keyboard_button_in_shell reads $HOME/.config and ignores XDG_CONFIG_HOME, which the removed helper honoured.

Cleared to merge at this sha.

@DanielJBurbridge
DanielJBurbridge merged commit a1fc4bb into feat/ally-parity-integration-20261111 Nov 17, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants