Skip to content

Point Voxtype evdev hotkey users at the Hyprland dictation binds - #12015

Open
kurenn wants to merge 6 commits into
omacom:quattrofrom
kurenn:restore-input-group-for-voxtype-hotkey
Open

kurenn wants to merge 6 commits into
omacom:quattrofrom
kurenn:restore-input-group-for-voxtype-hotkey

Conversation

@kurenn

@kurenn kurenn commented Sep 15, 2026 •

Copy link
Copy Markdown

Fixes #12013.

migrations/1787865477.sh (from #9200) drops the blanket input group grant, which is the right call: membership gives raw read/write access to /dev/input/event*, so any process running as the user can capture keystrokes. Users who set Voxtype's [hotkey] enabled = true relied on that grant. Their daemon keeps running and logs Listening for hotkey, then a single ERROR Hotkey listener error: No keyboard device found in /dev/input/ line, and the hotkey silently stops working.

Approach

Notify instead of regranting. This PR first restored input for opted-in users, but since then quattro binds push-to-talk on a bare modifier in Hyprland (default/hypr/bindings/dictation.lua: hold Right Alt or F9, toggle with Super + Ctrl + X, from #14552). That path needs no group membership, so a sudo-backed input grant gated on user-writable config is no longer justified.

The migration finds users with Voxtype installed, hotkey.enabled resolving to true (read with RUST_LOG=off so tracing output on stdout can't break the comparison), and no input membership. It prints the fix in the update terminal and sends a critical notification so it outlives the scrollback. It changes no groups and edits no config.

This also addresses the earlier review points: there is no grant, so the $USER target, the user-writable gate, and "input outlives Voxtype" no longer apply. The acceptance security check and the stubbed migration test are dropped; per agents/skills/migrations.md, migrations don't get tests under test/.

A failed voxtype config get still fails closed and marks the migration done. That's fine for a notice: the worst case is a missed message, not a missed grant.

Testing

  • Ran the migration with stubbed id, voxtype and omarchy-notification-send: it notifies when hotkey.enabled is true and is silent when false.
  • Real voxtype 1.1.0: RUST_LOG=off voxtype config get hotkey.enabled prints exactly true for a config with enabled = true.
  • ./test/cli, config-test.sh, migrate-scope-test.sh and migrate-wrapper-test.sh pass.

🤖 Generated with Claude Code

@omarchybot omarchybot added the bug Something isn't working label Sep 27, 2026
@llstrk

llstrk commented Sep 27, 2026

Copy link
Copy Markdown

Automated AI review

Community review: Independent automated community review, unaffiliated with the Omarchy team, intended to help prepare PRs for their review.

Verified: the new migration restores input for users who opted into Voxtype's evdev hotkey and leaves Omarchy's default compositor-bound setup alone, for both already-affected machines and late updaters who run 1787865477 and 1789490499 in the same update. Gating on voxtype config get hotkey.enabled rather than grepping the TOML is the right source: a [hotkey] section without enabled resolves to Voxtype's built-in default true, which matches what the daemon does. One robustness gap remains with the Voxtype version Omarchy currently packages (1.1.0), plus a hardening point on the target account.

Real omarchy-migrate with stubbed group, sudo, pacman and state commands and real Voxtype 1.0.1 and 1.1.0 binaries:

Scenario End groups Result
Late updater, hotkey on wheel input -d from 1787865477, then -a from 1789490499
Late updater, Omarchy default config wheel removed, not restored
Late updater, ydotool installed, hotkey on wheel input kept, no churn
Already affected, hotkey on wheel input restored
Already affected, default config wheel untouched
Voxtype not installed / already in input unchanged untouched

Log lines on stdout can defeat the == "true" check

Voxtype sends its tracing output to stdout, and the condition compares the whole captured stdout with true:

[[ $(voxtype config get hotkey.enabled 2>/dev/null) == "true" ]]

Two cases put log lines in front of true:

  • Voxtype 1.1.0, hotkey on, one config section it cannot read (for example max_duration_secs = "sixty" in [audio]): 1.1.0 skips that section with two WARN lines on stdout, exits 0 and still resolves hotkey.enabled to true.
  • RUST_LOG set to debug or trace in the update's environment (1.0.1 and 1.1.0): a DEBUG Loading config from ... line is printed first.
stdout (1.1.0, bad [audio] value, hotkey on):
  WARN Config section 'audio' could not be read and is using defaults. ...
  WARN 1 config section(s) were skipped: audio
  true
condition: false  ->  no gpasswd call, migration marked complete

Impact: in these cases the migration records itself as done without restoring the group. With 1.1.0 the daemon keeps running on the salvaged config with the hotkey enabled (per Voxtype's source), so the user stays in the broken state this PR fixes, and the migration never runs again. This needs the hotkey opt-in plus one of the two conditions above, so it is an edge case.

Suggested change: silence tracing for the query:

[[ $(RUST_LOG=off voxtype config get hotkey.enabled 2>/dev/null) == "true" ]]

With this change, both cases above restore the group, and every other scenario in the table keeps the same outcome. voxtype -q covers the WARN case but not RUST_LOG=debug, because the environment filter takes precedence over -q. The PR's test stub still matches the arguments.

Group target comes from $USER

id -nG "$USER" and sudo gpasswd -a "$USER" input take the account name from the environment. A different USER value makes the sudo-backed grant add input to another account. Open PR #8534 ("Take privileged usernames from id -un, not USER") makes the same argument for sibling privileged commands but does not cover a new migration. 1787865477 uses the same pattern, but for a removal rather than a grant.

Impact: the privileged grant's target is not tied to the account running the update.

Suggested change: resolve the account once with user=$(id -un) and use it in both calls. The test's id stub would then also need to answer -un.

Design point: the grant is gated on user-writable state

For maintainers to weigh. 1787865477 keeps input only on root-owned package state (pacman -Qq xpadneo-dkms or ydotool). This migration grants input through sudo based on ~/.config/voxtype/config.toml, or on VOXTYPE_HOTKEY_ENABLED=1 (or true) if that is set in the update's environment (it yields true even with Omarchy's default config). Any process running as the user can set either before the next update. It is also a first-time grant for users who never had the group (installs made from 4.0.2 until this migration ships, where the hotkey was enabled in the meantime). Impact: a sudo-backed input grant can be triggered by user-writable config or environment state. The window is a single migration run and requires an existing same-user compromise, and the opt-in does reflect user intent.

Optional approaches: keep the automatic grant as is, or notify affected users and print the one command instead of granting automatically.

Optional test improvement: the added test kills five of six diagnostic mutants of the migration, but not a fail-open comparison (!= "false" in place of == "true"), which grants input whenever voxtype config get fails or prints nothing. A fifth case with a voxtype stub that exits non-zero with empty stdout, asserting no gpasswd call, separates the two (original: no grant; mutant: grant).

Comment accuracy: migrations/1789490499.sh lines 13-14 say a bare modifier "cannot be bound in Hyprland at all". #7302 reports that press binds on a lone modifier fire and only the release bind never fires (hyprwm/Hyprland#6946), so hold-to-talk is what is impossible. Suggested wording: "push-to-talk on a bare modifier can't be done with Hyprland binds (hyprwm/Hyprland#6946)".

Note: Voxtype before 1.0.0 has no config get (0.7.5 exits 2 with "unrecognized subcommand"), so the migration skips silently and marks itself done. This is unlikely in practice, because omarchy update upgrades packages before running migrations and Omarchy's voxtype-bin PKGBUILD moved to 1.0.0 on 2026-09-01.


Review information

Test scope: Source review of head d194305 against base 2fbac0c and Voxtype source at v0.7.5, v1.0.1 and v1.1.0. The PR test, six diagnostic mutants and real omarchy-migrate ran in an isolated sandbox with released Voxtype 0.7.5, 1.0.1 and 1.1.0 binaries and stubbed id, sudo, gpasswd, pacman and omarchy-state. No real group, sudo or account changes were made, and the hotkey itself was not exercised on a live session.

AI process: Opus 5.5 Medium coordination and synthesis, Opus 5.5 Xhigh technical review and final fact check, GPT 6 Sol Xhigh search for related issues, Opus 5.5 Medium editorial check.

Opt out: To stop receiving these reviews, reply to this comment saying so.

@llstrk

llstrk commented Sep 29, 2026

Copy link
Copy Markdown

Automated AI review

Community review: Independent automated community review, unaffiliated with the Omarchy team, intended to help prepare PRs for their review.

Outcome: One minor new issue: the acceptance security check still rejects the input membership this migration grants.

Earlier findings at 1c88d053: stdout log lines, the fail-open test gap and the Hyprland comment are resolved. The $USER target and the user-writable gate are unchanged.

Acceptance security check rejects the restored group

verify_input_group in test/acceptance.d/security-test.sh allows input only with xpadneo-dkms or ydotool installed. Impact: where this migration restored input and neither package is installed, the check fails with not ok - user is not in the input group. Acceptance VMs lack Voxtype by default.

Suggested change: also accept [[ $(RUST_LOG=off voxtype config get hotkey.enabled 2>/dev/null) == "true" ]] there and update its comment.

Reproducer (fails at 1c88d05)

Self-contained script, run from an Omarchy checkout (bash acceptance-input-group-case.sh .). Everything is stubbed: id and gpasswd share a fake group list, pacman knows no packages, sudo -n fails, grim does nothing. No real groups, sudo or accounts are touched.

#!/bin/bash
set -euo pipefail

root=${1:?omarchy checkout}
test_dir=$(mktemp -d)
trap 'rm -rf "$test_dir"' EXIT

stub_bin="$test_dir/bin"
mkdir -p "$stub_bin"
printf 'wheel\n' >"$test_dir/groups"

cat >"$stub_bin/id" <<'STUB'
#!/bin/bash
[[ ${1:-} == "-nG" ]] || exit 2
cat "$STUB_GROUPS_FILE"
STUB
cat >"$stub_bin/gpasswd" <<'STUB'
#!/bin/bash
[[ $1 == "-a" ]] || exit 2
printf '%s %s\n' "$(cat "$STUB_GROUPS_FILE")" "$3" >"$STUB_GROUPS_FILE"
STUB
cat >"$stub_bin/sudo" <<'STUB'
#!/bin/bash
[[ ${1:-} == "-n" ]] && exit 1
exec "$@"
STUB
cat >"$stub_bin/pacman" <<'STUB'
#!/bin/bash
exit 1
STUB
cat >"$stub_bin/omarchy-cmd-present" <<'STUB'
#!/bin/bash
exit 0
STUB
cat >"$stub_bin/voxtype" <<'STUB'
#!/bin/bash
[[ $1 == "config" && $2 == "get" && $3 == "hotkey.enabled" ]] || exit 2
printf 'true\n'
STUB
cat >"$stub_bin/omarchy-state" <<'STUB'
#!/bin/bash
exit 0
STUB
cat >"$stub_bin/grim" <<'STUB'
#!/bin/bash
exit 0
STUB
chmod +x "$stub_bin"/*

export STUB_GROUPS_FILE="$test_dir/groups"

echo "== migrations/1789490499.sh (hotkey enabled, no xpadneo-dkms, no ydotool)"
USER=tester PATH="$stub_bin:$PATH" bash -euo pipefail "$root/migrations/1789490499.sh"
echo "groups now: $(cat "$STUB_GROUPS_FILE")"

echo "== test/acceptance.d/security-test.sh"
set +e
OMARCHY_ACCEPTANCE_DIR="$test_dir/artifacts" PATH="$stub_bin:$PATH" \
  bash "$root/test/acceptance.d/security-test.sh"
status=$?
set -e
echo "acceptance security-test exit status: $status"

Actual output at 1c88d05 (stdout, then stderr):

== migrations/1789490499.sh (hotkey enabled, no xpadneo-dkms, no ydotool)
Restore the input group for users who opted into Voxtype's evdev hotkey
Restored tester to the input group for the Voxtype hotkey. Log out and back in to apply.
groups now: wheel input
== test/acceptance.d/security-test.sh
acceptance security-test exit status: 1
no controller or ydotool support installed to justify it
not ok - user is not in the input group

The same output was produced with the PR's two files applied on top of current quattro.

With the suggested change to verify_input_group, the same script prints:

== migrations/1789490499.sh (hotkey enabled, no xpadneo-dkms, no ydotool)
Restore the input group for users who opted into Voxtype's evdev hotkey
Restored tester to the input group for the Voxtype hotkey. Log out and back in to apply.
groups now: wheel input
== test/acceptance.d/security-test.sh
ok - input group membership is backed by an opt-in feature
ok - asdcontrol sudoers check skipped: sudo needs a password
ok - sshd hardening exercise skipped: set OMARCHY_ACCEPTANCE_SUDO_PASSWORD to run it
acceptance security-test exit status: 0

Suggested change used for that run:

--- a/test/acceptance.d/security-test.sh
+++ b/test/acceptance.d/security-test.sh
@@ -15,11 +15,12 @@
 source "$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)/base-test.sh"
 
 # Membership of `input` gives raw access to /dev/input/event*: any process
-# running as the user could log keystrokes. Only the opt-in controller and
-# ydotool features may grant it.
+# running as the user could log keystrokes. Only the opt-in controller, ydotool
+# and Voxtype evdev hotkey features may grant it.
 verify_input_group() {
   if id -nG | grep -qw input; then
-    if pacman -Q xpadneo-dkms &>/dev/null || pacman -Q ydotool &>/dev/null; then
+    if pacman -Q xpadneo-dkms &>/dev/null || pacman -Q ydotool &>/dev/null ||
+      [[ $(RUST_LOG=off voxtype config get hotkey.enabled 2>/dev/null) == "true" ]]; then
       pass "input group membership is backed by an opt-in feature"
     else
       fail "user is not in the input group" "no controller or ydotool support installed to justify it"
Details

Verified at 1c88d05:

  • RUST_LOG=off voxtype config get hotkey.enabled printed exactly true, false or nothing in all 90 sandbox runs across Voxtype 1.0.1, 1.1.0 (the packaged version) and 1.1.1-rc4, with RUST_LOG unset, debug or trace. The old form failed the check in 43 runs whose value was true. It also silences a new stdout warning that 1.1.1-rc4 prints for a misplaced config section.
  • The real omarchy-migrate (with stubbed id, sudo, gpasswd, pacman and omarchy-state) restores input at 1c88d05 with RUST_LOG=debug or RUST_LOG=trace, including trace combined with a bad [audio] value. That combined setup at d194305 does not.
  • The new fail-closed case fails against a != "false" mutant, and the PR test passes 5 of 5 at the head.
  • The out-of-order timestamp is fine: omarchy-migrate runs every unmarked migration, and fresh installs pre-mark all of them.

Design point, new detail: on Voxtype 1.1.0 and 1.1.1-rc4, a wrongly typed value in [hotkey] (for example key = 7 next to enabled = false, or a quoted enabled = "false") makes Voxtype drop the whole section and use the built-in default. config get then prints true, and the migration grants input. The daemon does run the built-in Scroll Lock evdev hotkey in that case, but the user wrote the hotkey off. On 1.0.1 the whole file is rejected and the migration skips.

Unchanged: the $USER point (open #8534 still covers only bin/ commands). Voxtype before 1.0 has no config get, which needs no action because packages are upgraded before migrations run and Omarchy ships Voxtype 1.1.0.

Optional notes

Optional test improvement: the PR test passes even with RUST_LOG=off removed, because the voxtype stub never prints a log line. The case below fails with the d194305 migration and passes at 1c88d05:

--- a/test/shell.d/voxtype-input-group-migration-test.sh
+++ b/test/shell.d/voxtype-input-group-migration-test.sh
@@ -23,6 +23,8 @@
 #!/bin/bash
 [[ $1 == "config" && $2 == "get" && $3 == "hotkey.enabled" ]] || exit 2
 [[ ${STUB_VOXTYPE_FAILS:-0} == "0" ]] || exit 1
+# Voxtype writes its tracing output to stdout, and RUST_LOG overrides its level.
+[[ ${RUST_LOG:-off} == "off" ]] || printf '%s\n' ' DEBUG Loading config from "/home/tester/.config/voxtype/config.toml"'
 printf '%s\n' "${STUB_HOTKEY_ENABLED:-false}"
 STUB
 cat >"$stub_bin/sudo" <<'STUB'
@@ -73,3 +75,7 @@
 [[ ! -e $gpasswd_calls ]] || fail "migration does not grant input when voxtype config get fails"
 [[ ! -e $state_calls ]] || fail "a failed voxtype query does not flag a reboot"
 pass "migration fails closed when the Voxtype config query fails"
+
+RUST_LOG=debug run_migration "wheel" true >/dev/null
+grep -qsxF -- "-a tester input" "$gpasswd_calls" || fail "migration restores input when RUST_LOG is set in the update environment"
+pass "migration ignores Voxtype log output when reading hotkey.enabled"

With the migration from d194305, the new case prints not ok - migration restores input when RUST_LOG is set in the update environment.


Review information

Test scope: Source and sandbox tests at 1c88d053 with Voxtype binaries and stubbed group and sudo commands; no live session.

AI process: Opus 5.5 Medium coordination and synthesis, Opus 5.5 Xhigh technical review and final fact check, GPT 6 Sol Xhigh search for related issues, Opus 5.5 Medium editorial check.

Opt out: To stop receiving these reviews, reply to this comment saying so.

@greptile-apps

greptile-apps Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

RetriggerView in GreptileConfidence Score: 5/5

[Low impact] The PR appears safe to merge; no new actionable issues remain.

Summary

Adds a one-time notice for Voxtype users whose own hotkey depended on membership in input.

  • Voxtype users with an enabled hotkey get guidance for Hyprland’s dictation keys.

Reviews (3) · Last reviewed commit: "Check the group database and name quattr..." · Reviewed by Greptile

Comment thread migrations/1789490499.sh Resolved
Comment thread migrations/1789490499.sh Resolved
@omarchybot omarchybot added verified Omarchy Triage has verified that this issue is ready for final review ready Good to merge labels Oct 1, 2026
@omarchybot

Copy link
Copy Markdown
Collaborator

Reviewed at b126695 and verified on a disposable Omarchy VM with voxtype-bin 1.0.1, running the real omarchy-migrate against a user with [hotkey] enabled = true:

  • On quattro, and on this branch with 1789490499.sh removed, the user ends up outside input and voxtype setup check reports evdev hotkeys unavailable. This holds for a late updater, where 1787865477 removes the group in the same run, and for a machine 1787865477 already reached.
  • On this branch the same user is restored to input in both cases, and a fresh process can open /dev/input/event0. It is also restored with RUST_LOG=debug in the update's environment.
  • With Omarchy's default config (enabled = false), nobody is added to the group.

./test/cli passed. The new migration test passed 5/5 and input-group-migration-test.sh passed 3/3. The full test/shell.d run had four failing files: omarchy-kernel-migration, screenshot-sanity, ascii and branding-about-animation. All four fail the same way on the merge base 2fbac0c8 without this change.

I pushed b126695, which changes test/acceptance.d/security-test.sh. verify_input_group accepted input only alongside xpadneo-dkms or ydotool, so on a machine where this migration restored the group, the security acceptance check reported it as a leftover grant: not ok - user is not in the input group. It now also accepts an enabled Voxtype hotkey, read the same way the migration reads it. On the VM it passes for an opted-in user and still fails when the hotkey is off or Voxtype is not installed.

Codex at medium effort reviewed the change including that commit and found no defect. It agreed that a new migration is the proportionate fix, since 1787865477's marker already exists on the machines that were hit. That agreement is not guaranteed to be independent of this review. It also pointed out that the gate is Voxtype's effective setting rather than an explicit line in the file: a missing [hotkey] section counts as enabled because that is Voxtype's default, and that matches what the daemon actually does.

One question is left for the maintainer, and it is not a defect: whether a sudo-backed input grant should follow from user-writable Voxtype config. 1787865477 keeps the group only for root-owned package state. Waiting on the maintainer.

@jandrusk

Copy link
Copy Markdown

Looks right to me: restore input only when Voxtype's own [hotkey] enabled = true, fail closed on query errors, and keep the compositor-bound default out. Migration name 1789490499 is free on quattro. Related product tension with #12895 (narrowing other input-group grants) — worth a maintainer glance so the security tests stay consistent.

@kurenn
kurenn force-pushed the restore-input-group-for-voxtype-hotkey branch from b126695 to efbc7bc Compare October 25, 2026 15:59
@kurenn kurenn changed the title Restore the input group for Voxtype evdev hotkey users Point Voxtype evdev hotkey users at the Hyprland dictation binds Oct 25, 2026
@kurenn

kurenn commented Oct 25, 2026

Copy link
Copy Markdown
Author

Changed the approach at efbc7bc (rebased onto quattro): the migration now notifies instead of restoring input.

Since this PR was opened, #14552 added Hyprland push-to-talk on a bare modifier (default/hypr/bindings/dictation.lua: hold Right Alt or F9, toggle with Super + Ctrl + X). So the claim that drove the grant, that an evdev hotkey user has no compositor-side fallback, no longer holds. With that fallback in place, a sudo-backed input grant gated on user-writable config isn't justified.

What this means for the earlier review points:

  • $USER target, user-writable gate, input outlives Voxtype: no longer apply, because nothing is granted.
  • Acceptance check exception (b126695): reverted; verify_input_group is back to the quattro version.
  • Stubbed migration test: removed, per agents/skills/migrations.md ("Don't add a test for a migration under test/").
  • RUST_LOG=off on the query: kept, so tracing output on stdout can't hide a true.

The PR diff is now just migrations/1789490499.sh. @jandrusk this also removes the tension with #12895: no new input grant path.

Comment thread migrations/1789490499.sh Resolved
@omarchybot omarchybot removed verified Omarchy Triage has verified that this issue is ready for final review ready Good to merge bug Something isn't working labels Oct 26, 2026
@omarchybot

Copy link
Copy Markdown
Collaborator

Re-reviewed at efbc7bc, the notice-only version. I read it against current quattro (5df8f9e); nothing was run, because the classification is unsettled (below), and that stops this pass before a worker. Codex at medium effort and muse-wonder via pi at high thinking reviewed the same head. Both found the same two defects I did, and nothing else. Their agreement is not guaranteed to be independent of this review.

1. The notice is skipped for the users it is meant for (migrations/1789490499.sh:15). id -nG with no user prints the groups of the running process, not the account's membership. When 1787865477 and this migration run in the same omarchy update, gpasswd -d changes only the group database. The update terminal still holds input, so the check matches, no notice is printed or sent, and the migration exits 0 and is marked done. Once that user logs out, the hotkey breaks and the notice never comes. 1787865477 checks id -nG "$USER" (line 10), which reads the database and sees the removal straight away. The same form here fixes it:

  ! id -nG "$USER" | grep -qw input &&

This is the same point Greptile raised on line 15.

2. The keys it names are not quattro's defaults any more (comment on lines 9–10, message on line 17). #17188 landed after this branch's merge base (2fca7fe). It moved the defaults to hold Right Alt and toggle with Alt + Space (default/hypr/bindings/dictation.lua:15-18, and the dictation.hint string in default/i18n/en/cli.json:106). F9 and Super + Ctrl + X now exist only where migrations/1793477875.sh appends them to a personal bindings.lua. That happens only for a user who has a backend configured and already has that file, and that migration runs after this one. Telling the user what dictation.hint says is true for every Voxtype user: "Hold Right Alt to dictate, or toggle with Alt + Space."

Checked and found fine:

  • The voxtype config get gate sits inside an if, so the runner's bash -euo pipefail can't abort on it.
  • RUST_LOG=off keeps log output off stdout.
  • omarchy-notification-send -u critical matches that command's options, and || true keeps a missing session bus from stalling the queue.
  • The older timestamp still runs on machines past it, because omarchy-migrate keeps one marker per file.
  • Nothing is granted or edited, so 1787865477's security point stands.

Classification is unsettled. I classify this as a bug fix, because it repairs the silent loss of the hotkey that 1787865477 caused. The classifier's separate pass, Codex at medium effort, called it an enhancement: it now adds a notice instead of restoring the old behaviour. With no agreement, I've taken the bug label off for now and left the class to the maintainer. No fix was pushed. Both changes above are small, and the author can make them or the maintainer can settle the class first. It waits on the maintainer.

kurenn and others added 6 commits November 2, 2026 07:25
Voxtype writes tracing output to stdout, so RUST_LOG=debug (or a skipped
config section on 1.1.0) puts log lines in front of "true" and the
migration marks itself done without restoring the group.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…closed

Press binds on a bare modifier do fire; only the release never does, so
push-to-talk is what Hyprland can't do. The new test case makes sure a
failing voxtype config query never grants input.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…eptance check

verify_input_group still allowed input only for xpadneo-dkms or ydotool, so on a machine where the new migration restored the group for an opted-in Voxtype hotkey, the security acceptance test reported it as a leftover grant.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…nting input

Hyprland now binds hold Right Alt and F9 for push-to-talk, so the evdev
hotkey no longer needs a sudo-backed input grant gated on user config.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…type notice

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@kurenn
kurenn force-pushed the restore-input-group-for-voxtype-hotkey branch from efbc7bc to 0c14335 Compare November 2, 2026 13:26
@kurenn

kurenn commented Nov 2, 2026

Copy link
Copy Markdown
Author

Both points are fixed in 0c14335, rebased onto quattro:

  1. The check now reads id -nG "$USER" (the group database), the same form 1787865477 uses, so a terminal that still holds input can't skip the notice.
  2. The comment and the notice now name quattro's defaults, matching dictation.hint: "Hold Right Alt to dictate, or toggle with Alt + Space."

The PR diff is still just migrations/1789490499.sh.

@omarchybot omarchybot added the enhancement New feature or request label Nov 18, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement New feature or request

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Migration 1787865477 drops the input group, silently breaking Voxtype's evdev hotkey

4 participants