Skip to content

omamobile-shell 0.2.52: bound widget output and stop stale work - #193

Merged
derekross merged 3 commits into
derekross:mainfrom
Gr-linkk:fix/widget-lifecycle
Oct 31, 2026
Merged

derekross merged 3 commits into
derekross:mainfrom
Gr-linkk:fix/widget-lifecycle

Conversation

@Gr-linkk

@Gr-linkk Gr-linkk commented Oct 30, 2026 •

Copy link
Copy Markdown
Contributor

Current-main follow-up (2026-10-31)

Merged current main without rewriting the submitted history and resolved the new OwlPhoneHost conflict. The host retains upstream's device/board metadata and returns the cancellation handles for both system and device sources. The new real-Quickshell regression cancels both sources and rejects late callbacks; it fails before the resolution and passes after it. Version is now 0.2.52, as requested. After the merge, tools/pr-check upstream/main, tests/all, and tests/packages omamobile-shell pass with the documented environment skips. Independent review found no implementation issue.

The phone evidence below belongs to the earlier submitted component test on the existing Oriole shell, not a full-package test of this revised merge. The updated complete package still awaits the maintainer's Pixel 7 Pro test; no phone was changed during this follow-up.

What and why

Widget command output was truncated only after the process exited, and HTTP responses were checked only after receipt. Large replies could therefore accumulate in memory before either limit applied. Both widget hosts now share a process component that detaches its collectors at 64 KiB; HTTP reads abort above 1 MiB while receiving.

OWL also kept work alive in several cases: an expanded widget could bypass the home's pause, old requests could update a replacement instance, and source updates scheduled writes even when persisted state had not changed. This change cancels pending source work when hidden, paused or reloaded, ignores stale completions, and saves only changed state. Frequent updates no longer keep postponing a save; pending state is flushed on reload and destruction.

Device and build

  • Phone: Pixel 6 (oriole), native Arch Linux.
  • Kernel release: 7.3.0-rc6-oriole-hsi1-late-dev0-g4bd35d34abb9-dirty.
  • Package changed: omamobile-shell 0.2.51 → 0.2.52 (version requested in review).
  • Phone runtime: Omarchy 4.0.4-2, shell 0.2.12-1, Quickshell 0.3.2.r7.g2b4501a-1, Qt 6.12.0-1, Hyprland 0.56.2-7.

Checks (no phone needed)

  • tools/pr-check upstream/main fix/widget-lifecycle passes.
  • tests/all passes.
  • tests/packages omamobile-shell passes; the lifecycle and host regressions also passed on the previously submitted sources on the phone.
  • Package version is bumped and included in the commit subject.
  • Commit has an area prefix, rationale and an Assisted-by: trailer.

The regression tests exercise real Qt/Quickshell: command collectors stop retaining output before process exit, split UTF-8 remains intact, oversized HTTP replies abort before EOF, cancellation terminates a child command, old callbacks cannot affect a new instance, unchanged state is not written, and busy history updates still save. They also cover state set synchronously by the host while an instance loads.

The local tests/all run skipped checks requiring an Arch Docker image, AOSP mkbootimg, dtc or a built root image. Quickshell was available locally and on the phone, so the new host tests ran in both places.

On the phone

  • Ran isolated regressions first on native ARM Qt/Quickshell: 17 lifecycle tests passed; bounded command/HTTP, cancellation and state tests passed.
  • Backed up current bytes, modes and ownership before the temporary component test.
  • Temporarily replaced OwlWidget.qml, OwlPhoneHost.qml, WidgetHost.qml and owl/phone.js, added BoundedProcess.qml, and added only the pause binding to the installed OwlCell.qml. The older Home's visibility wiring was preserved.
  • Restarted the shell, checked its IPC response, inspected the rendered home screen and ran phone-smoke. No QML errors appeared in the captured startup log.
  • Restored every original file with matching SHA-256, mode and owner, removed the new helper, restarted the shell and reran smoke. One Hyprland process and active user/Wayland sessions remained; shell IPC answered.
  • No writes outside userdata, storage-guard changes, register reads, package upgrades or bring-up changes.

This was a temporary component check on the existing 0.2.12 installation. I did not install the complete 0.2.46 package, upgrade its dependencies, test on cheetah, or measure CPU/battery savings. The existing NFC/backlight service failures came from earlier low-battery guards; neither was cleared or retried. Sound, sensors and camera checks were skipped as shown below.

tools/phone-smoke with the temporary components loaded

phone-smoke: oriole

Kernel 7.3.0-rc6-oriole-hsi1-late-dev0-g4bd35d34abb9-dirty, tools at d8980e4, 2026-10-30T20:08Z.

Check Result Detail
boot (required) ok kernel 7.3.0-rc6-oriole-hsi1-late-dev0-g4bd35d34abb9-dirty, slot b, up 2793s
storage guard (required) ok 49 partitions; only userdata and slot B's boot images writable
root (required) ok ext4 on userdata, 97G free
kernel health (required) ok no oops or BUG; 0 WARN, 50 error lines in dmesg
systemd units not ok failed: omamobile-oriole-backlight.service omamobile-oriole-nfc.service
USB gadget (required) ok bound to 11110000.usb
display ok card0-Unknown-1 connected, 1080x2400
GPU ok render node on panfrost
session (required) ok Hyprland and the Omarchy shell running
lock ok PAM file present (the shell didn't answer lock status)
touch ok ST FTM5 touchscreen (oriole staged)
buttons ok gpio-keys
Wi-Fi ok wlP1p1s0 on brcmfmac, link up
Bluetooth ok hci0, powered true
battery ok 23% Charging, 32 °C
sound cards skip none
sensors skip no sensor proxy
cameras skip no libcamera tools
thermal ok hottest zone 32 °C

Packages: hyprland 0.56.2-7, libcamera 0.7.2-4, mesa 1:26.2.4-1, omacam 0.1.9-1, omamobile-apps 0.2.0-1, omamobile-backlight-oriole 0.1.0-8, omamobile-battery-counters-oriole 0.2.0-1, omamobile-bluetooth-oriole 0.1.2-2, omamobile-bluez 5.87.omamobile1-1, omamobile-charger-cache-oriole 0.1.0-8, omamobile-device-oriole 0.1.9-2, omamobile-firstboot 0.1.1-1, omamobile-gpu-oriole 0.1.0-3, omamobile-keyring 20261013.oriole1-1, omamobile-mirror 0.2.1-1, omamobile-nfc-oriole 0.1.0-9, omamobile-omarchy 0.1.6-1, omamobile-owl-widgets 0.1.1-1, omamobile-shell 0.2.12-1, omamobile-torch-oriole 0.1.1-2, omamobile-wifi-oriole 0.1.8-3, omarchy 4.0.4-2, quickshell 0.3.2.r7.g2b4501a-1

An expanded OWL screen could keep sources and every triggers running
while home was paused. Source completions could also update a replacement
instance, and unchanged source values scheduled state writes on every
refresh. A busy stream could postpone the restarted save timer forever.

Honor the global pause, cancel pending source reads, and bind callbacks to
their instance and subscription. Save only changed persisted data with a
fixed coalescing interval, flushing pending state before reload.

Share bounded command collection between OWL and QML widgets. Discard
output beyond 64 KiB per pipe while allowing commands to finish, preserve
split UTF-8, and abort HTTP responses as received bytes exceed 1 MiB.

Verified with failing-before/passing-after lifecycle tests and real
offscreen Qt/Quickshell command and HTTP fixtures. tests/packages
omamobile-shell passes. No phone smoke or battery measurement yet.

Assisted-by: Codex <noreply@openai.com>
@Gr-linkk Gr-linkk closed this Oct 30, 2026
Main now ships 0.2.45. Merge the current base and use 0.2.46 so the
widget lifecycle and output fixes get a distinct package version.
Preserve main's new mouse-handling tests alongside the widget tests.
The reviewed widget implementation is unchanged.

Assisted-by: Codex <noreply@openai.com>
@Gr-linkk Gr-linkk changed the title omamobile-shell 0.2.45: bound widget output and stop stale work omamobile-shell 0.2.46: bound widget output and stop stale work Oct 30, 2026
@Gr-linkk Gr-linkk reopened this Oct 30, 2026
@derekross

Copy link
Copy Markdown
Owner

Welcome, and thanks for careful work. The tests that fail on the old code and the 64 KiB / 1 MiB bounds are exactly right. 0.2.46 to 0.2.51 are taken, so this will land as 0.2.52. Your phone check ran on an older shell with the files swapped in, so I'm installing the package on my Pixel 7 Pro first; I'll merge after that.

Merge current main into the widget lifecycle fix, keeping its new board
metadata and the cancellation handle returned by both system and device
sources. Without the handle, a source can survive a pause or reload; the
system branch also falls through to a second, null callback.

Use the version requested in review. Extend the real Quickshell host
regression to cancel both sources and reject any completion afterwards.
The regression fails before the conflict resolution and passes with it.

Assisted-by: OpenAI GPT-6 Astra
@Gr-linkk Gr-linkk changed the title omamobile-shell 0.2.46: bound widget output and stop stale work omamobile-shell 0.2.52: bound widget output and stop stale work Oct 31, 2026
@derekross

Copy link
Copy Markdown
Owner

Thanks for the careful merge and the late-callback test. I'm installing 0.2.52 on my Pixel 7 Pro next and will merge after. If #217 lands first, I'll renumber this to 0.2.53 myself.

derekross added a commit that referenced this pull request Oct 31, 2026
… rounds

L-348 to L-361: shared drivers judged against every int-line user (#150),
board pins against every stock user (#151), ownership flags under the
consumer's lock (#152), /chosen children and Google's panel constants
(#154), same-day notes PRs on one device (#208/#209), merge simulations for
DEVICES rows, ours included (#165), stacked PRs that merged main (#115/#117),
titles behind renumbered versions, a contributor merge that dropped the PR
(#99), smoke package lists as provenance and "owner-authorized" writes
(#200), promised fixes and named tags checked on the remote (#153/#157/#158),
and file swaps into old packages (#193). C-23 (usbc-deadman's device scope,
kernel #107) and C-24 (BNEP in a lynx-only fragment, #98).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@derekross
derekross merged commit fab1c9a into derekross:main Oct 31, 2026
2 checks passed
@derekross

Copy link
Copy Markdown
Owner

Installed on my Pixel 7 Pro from your head merged onto main: the home widgets render and update, and phone-smoke passes all 19. Merged as 0.2.52, thanks! (#217 also claims 0.2.52, so it'll take the next number.)

robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
… rounds

L-348 to L-361: shared drivers judged against every int-line user (derekross#150),
board pins against every stock user (derekross#151), ownership flags under the
consumer's lock (derekross#152), /chosen children and Google's panel constants
(derekross#154), same-day notes PRs on one device (derekross#208/derekross#209), merge simulations for
DEVICES rows, ours included (derekross#165), stacked PRs that merged main (derekross#115/derekross#117),
titles behind renumbered versions, a contributor merge that dropped the PR
(derekross#99), smoke package lists as provenance and "owner-authorized" writes
(derekross#200), promised fixes and named tags checked on the remote (derekross#153/derekross#157/derekross#158),
and file swaps into old packages (derekross#193). C-23 (usbc-deadman's device scope,
kernel derekross#107) and C-24 (BNEP in a lynx-only fragment, derekross#98).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
robertzas pushed a commit to robertzas/omarchy-mobile that referenced this pull request Nov 14, 2026
…kross#193)

OWL and command widgets read at most 64 KiB of a command's output and 1 MiB in all, through a new BoundedProcess, and a widget that is torn down or hidden stops its pending work instead of applying a late result. New tests (owl-lifecycle, widget-host) fail on the old code. Installed from the head merged onto main on the Pixel 7 Pro (2026-10-31): the home widgets render and update, one shell runs, and phone-smoke passes 19 of 19.
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