Set buildAndTestSubdir + tauriBundleType so cargo-tauri.hook handles
build/install correctly, and drop the manual overrides:
- installPhase override (was needed because hook didn't know to pushd
into src-tauri without buildAndTestSubdir; hook's install script now
finds bundle/deb/.../data/usr/* on its own)
- preBuild with `bun run build` (hook runs beforeBuildCommand from
tauri.conf.json itself; bun2nix.hook has node_modules ready before
buildPhase)
- jq del(.build.beforeBuildCommand) in postPatch (no longer needed)
Net change: +3/-18. End-to-end verified on NixOS x86_64-linux:
GigaAM v3 ASR + Whisper large-v3-turbo on Vulkan both work.
* fix(macos): pass -parse-as-library to swiftc
Without this flag swiftc compiles a single-file input in script mode
and emits a synthetic `_main` into the object file. Packaged into
libapple_intelligence.a and linked alongside Rust's `_main`, Apple's
open-source ld64 (used by nixpkgs' Darwin stdenv) picks Swift's main,
leaving the app with a 5-instruction no-op that returns 0 immediately.
The binary looks complete — full Rust code, Metal, Swift runtime,
onnxruntime rpath — but launching it exits cleanly with code 0, no
output. Production CI masks the issue because Xcode's linker happens
to prefer Rust's `_main`.
`-parse-as-library` keeps swiftc in library mode so no `_main` is
emitted. The @_cdecl exports used by the Rust FFI are unaffected.
* fix(macos): respect SDKROOT/SWIFTC env vars for non-Xcode toolchains
xcrun is unavailable in non-Xcode setups (e.g. nixpkgs uses
apple-sdk_* plus a standalone swift compiler). Honor SDKROOT and
SWIFTC if set; fall back to xcrun otherwise so Apple-toolchain
behavior is unchanged.
Also invoke swiftc directly via the resolved path rather than via
`xcrun swiftc`.
Merge content from both files into AGENTS.md, making it the single
source of truth for all AI coding assistants. CLAUDE.md now references
AGENTS.md via `Read @AGENTS.md` directive.
Kept existing architecture/technical descriptions as-is to avoid
a separate review cycle.
Added sections: linting/formatting commands, i18n, code style, commit
guidelines, CLI parameters, debug mode, platform notes, troubleshooting
reference, and explicit links to CONTRIBUTING.md,
CONTRIBUTING_TRANSLATIONS.md, and PR template.
The previous check used `env::var(...).is_ok()`, which only tested for
the variable's presence and ignored its value. Setting the variable to
`0` or `false` counter-intuitively still disabled gtk-layer-shell.
Introduce `env_flag_enabled` helper that treats `0`, `false`, `no`,
`off`, and the empty string as falsy (case-insensitive).
Add a "Linux Startup Crashes or Instability" subsection to the
Troubleshooting section, with step-by-step instructions:
1. Install/reinstall gtk-layer-shell
2. Disable gtk-layer-shell via HANDY_NO_GTK_LAYER_SHELL env var
3. Disable WebKit DMA-BUF renderer
4. How to make workarounds permanent (.desktop file, shell profile)
Also add a cross-reference from the Linux Notes section.
* fix: surface paste errors as UI toast notification (#522)
When pasting the transcription fails (e.g. wtype/xdotool/dotool not
available or returns an error on Linux), the error was silently logged
to the backend console only. Users had no idea why their text was not
pasted.
Emit a new 'paste-error' Tauri event from the Rust side and listen for
it in the frontend App component, showing a sonner toast with the error
detail — mirroring the existing 'recording-error' pattern exactly.
Files changed:
- src-tauri/src/actions.rs: add PasteErrorEvent struct, emit event on paste failure
- src/App.tsx: add useEffect listener that shows toast.error on paste-error
- src/lib/types/events.ts: add PasteErrorEvent interface
- src/i18n/locales/en/translation.json: add pasteFailedTitle/pasteFailed keys
* chore: apply cargo fmt and prettier formatting
* add translationss
---------
Co-authored-by: CJ Pais <cj@cjpais.com>
* perf: add reasoning_effort passthrough to avoid thinking-mode latency
Reasoning models (Gemma 4, Qwen 3, etc.) default to thinking mode,
adding 10-40x latency for simple transcript cleanup. Pass through the
OpenAI-compatible reasoning_effort parameter so users can disable
thinking when speed matters more than deep reasoning.
- Add optional reasoning_effort field to ChatCompletionRequest
- Thread through send_chat_completion / send_chat_completion_with_schema
- New set_post_process_reasoning_effort Tauri command for persistence
- Dropdown in post-processing settings, Custom provider only
- i18n for en and ja
* refactor: simplify reasoning_effort to on/off toggle
Addresses review feedback on the reasoning_effort passthrough:
post-processing rarely benefits from reasoning, so a 5-way dropdown
(default/none/low/medium/high) is overkill. Collapse to a single
boolean toggle "Disable reasoning" that defaults to ON for Custom
providers (sends reasoning_effort: "none"). Cloud providers continue
to receive no reasoning_effort parameter.
- Replace post_process_reasoning_effort: Option<String> with
post_process_disable_reasoning: bool (default true) in settings
- Map bool to Option<String> at the request site (true -> "none",
false -> omit) so the API contract is unchanged
- Rename Tauri command to set_post_process_disable_reasoning(disable)
so parameter naming matches the setting and avoids inversion bugs
- Swap the Dropdown for a ToggleSwitch in the Custom provider section
- Collapse five i18n strings to one label + description, and add the
translation to all 18 additional locales so non-en/ja users don't
fall back to English for this control
* refactor: remove disable-reasoning toggle, always disable for custom provider
Per PR #1221 discussion with @cjpais: since this is an alpha feature,
simplify the UI by always sending reasoning_effort: "none" for custom
provider post-processing. If this causes issues with specific provider
setups, we can revisit.
* disable for openrouter too
---------
Co-authored-by: CJ Pais <cj@cjpais.com>
Fixes#1184.
The tray icon was created and refreshed without setting a tooltip, which caused Windows to show an empty hover tooltip.
This change sets the tooltip when building the tray icon and refreshes it during tray menu updates using the existing version label.
* fix: don't log cloud provider keys
* test: make sure keys are redacted
* change to newtype
* change secret struct to public
* test: test secretmap directly
* Update bindings.ts
---------
Co-authored-by: Shaan <shaankhosla@macbook-pro.mynetworksettings.com>
Co-authored-by: CJ Pais <cj@cjpais.com>
* audio_toolkit: remove long repeating words
I've noticed a couple of times that Paraket v3 may produce
repeats of words that are not folded by Handy so I propose
to lift up word length limitation.
* Update text.rs
---------
Co-authored-by: CJ Pais <cj@cjpais.com>
Add a visible feature freeze section near the top of CONTRIBUTING.md
so contributors see it before starting work. Fix "GATEHRED" typo in
the PR template.
* fix: require magic string in portable marker to prevent false portable mode
Scoop's extras bucket extracts the NSIS installer via #/dl.7z, which can
leave a stale empty portable file next to the exe. This caused
portable::init() to falsely trigger portable mode on non-portable scoop
installs, storing data in Data/ next to the exe (wiped on each update).
Changes:
- portable.rs: only enable portable mode when marker contains
"Handy Portable Mode"; extracted is_valid_portable_marker() with tests
- installer.nsi: write magic string when creating marker file
- installer.nsi: validate magic string in update auto-detect block
- installer.nsi: allow desktop shortcut creation from finish page
checkbox in portable mode (was silently doing nothing)
- commands/mod.rs + lib.rs: expose is_portable() as a Tauri command
- UpdateChecker.tsx: show manual-update popup for portable installs
instead of attempting auto-install (which would download MSI)
- en/translation.json: add i18n strings for portable update popup
Fixes#1124
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: add missing translation keys to all locales and fix prettier formatting
Add portableUpdateTitle, portableUpdateMessage, portableUpdateButton to
all 17 non-English locales (English fallback text) so check:translations
passes. Also run prettier on the PR's modified files to fix format:check.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix: migrate legacy empty portable markers from v0.8.0
v0.8.0 created an empty `portable` marker file. Users who manually
update (drop new exe in place) would silently lose portable mode since
the new magic-string check rejects empty files.
Migration: if the marker exists but has invalid/empty content AND a
`Data/` directory is present, treat it as a real portable install and
rewrite the marker with the magic string. This handles the v0.8.0 →
new upgrade path without regressing the scoop false-positive fix
(scoop does not create a Data/ directory).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* increase indicator width to fit German translation
* use text-end for proper alignment of Arabic translation
* use ms-4 instead of ml-4 to fix score bar alignment for Arabic
* use ms-auto to fix download button alignment for Arabic
Fixes#1109
* save recordings before transcription
* impart cj preferences
* cleanup
* re-transcribe anything
* format
* fix tests
* format
* ui cleanup
* format
* cleanup
* bunch of fixes, mainly for vm
* put comments back
* update transcription failure text
* alert -> toast
* spawn_blocking
---------
Co-authored-by: CJ Pais <cj@cjpais.com>
