Commit Graph
15 Commits
Author SHA1 Message Date
Alex SouthwellandSAPHID 7b53af1c18 Fix OBS lifecycle cleanup and validate real headset capture (#219)
Co-authored-by: SAPHID <saphid@users.noreply.github.com>
2026-10-25 06:42:11 +00:00
Alex SouthwellandSAPHID e7bfa80e09 Add opt-in Open at Login with tray startup (#189)
* Add opt-in Open at Login with tray startup

* Show login window when Linux tray has no visible icon

---------

Co-authored-by: SAPHID <saphid@users.noreply.github.com>
2026-10-25 01:32:10 +00:00
SAPHIDandClaude Opus 5.5 7a7dca70de PC in the headset: Windows only, Experimental and off until turned on
host_view now picks the PC host on Windows only; Linux and macOS keep
the Mac view as on main, so #38's Linux adapter is never offered. The PC
host starts nothing, loads no native library and lists no windows until
the person ticks "Turn on PC in the headset" on the card (action
"enable" on /api/macview). Turning it off ends streams, viewers and the
helper. The choice is a client preference, like mvSeparate.

Tools lists "PC in the headset" on Windows only, beside the macOS-only
Mac link, with an Experimental tag and a short note on what hasn't been
tried. The Mac card's wording and Linux-portal tweaks from #38 go back
to main's, apart from the shared "your computer" hint.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-24 00:29:12 +00:00
saphidandClaude Opus 5.5 be9e7b5cda Merge feat/android-store-redesign (with main) into the Apps / Games / Tools / Settings structure
Keeps the four primary destinations and ports main's newer work into them:
the Home "Right now" panel (volume, cast, quick actions) replaces the old
device block, whose readings now live in the header's headset popover; the
install queue sits above the Apps views; Import from my Quest and My Meta
library sit with Browse; the pad, panel switcher, Mac view and media panels
take main's markup. Scrolling follows main's #scroll container. Settings
stays a tab, so main's header gear is not added. The old .home-status
toggle is gone (main's startup test caught it touching a missing node).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-20 19:49:13 +11:00
saphidandClaude Opus 5.5 fed2249776 Apps / Games / Tools / Settings, one Add / Install flow, and the whole-app review fixes
Primary tabs become Apps (Installed / Browse / Linux apps), Games (Installed /
Owned / Store), Tools and Settings, with shortcuts 1-4 and one Add / Install
chooser for catalogs, Steam, Linux apps, files and install links. Installed
rows show identity, one state and Open/Stop; the rest moves to app details.

Review fixes (#106): input, touch, approvals, artwork repair, keyboard helper
delivery and catalogue jobs stay bound to the headset they were meant for;
report preview follows the latest consent; failed Store reads are not an empty
library; keyboard access to device details and the Store sheet; the iOS
bundle ships the artwork renderer and OpenXR resources; typed addresses and
custom SSH ports survive setup; WebKit crashes recover; native preferences
survive relaunch; APK update rollback and repair stay retryable.

Refs #101, #104, #106

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-20 15:08:09 +11:00
saphidandClaude Opus 5.5 d773babaa3 Meta sign-in: allow Instagram, and say where a stopped redirect went
A live sign-in with a real Meta account goes auth.oculus.com -> auth.meta.com
-> www.facebook.com -> www.instagram.com -> auth.meta.com. The window only
allowed oculus.com, meta.com and facebook.com, so it stopped the Instagram
step and sat blank. instagram.com is now allowed (same HTTPS-only checks;
the cookie is still read only from oculus.com). A page change the window
stops is remembered by host only, so closing the stuck window says why; it
is forgotten once the window loads a page again.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 13:39:59 +11:00
saphidandClaude Opus 5.5 c7125bc3de My Meta library: Sign in with Meta closes two sign-out races
- cancel_signin marks an in-flight attempt and, independently, deletes the
  account if that attempt saved it, so a cancel landing after the save but
  before the connect's cleanup still removes the token.
- Sign-out runs one at a time (a second call joins it) and refuses new
  sign-ins while it runs; it goes round until no window, connect or cancel is
  left. The page keeps Sign in with Meta off from pressing Disconnect until
  the server has deleted the token.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 10:10:59 +11:00
saphidandClaude Opus 5.5 9c33711e6f My Meta library: Sign in with Meta cancels only after its connect settles
When the sign-in window is closed, times out or is signed out of while a
connect is under way, the page is answered at once, but cancel-signin is
posted only after that connect settles (saved, refused or failed), so the
server always sees the cancel after the connect and deletes the token if
that connect saved it. Nothing sent means no cancel. Sign-out, Disconnect's
first step, waits for any such cancel (also from a window closed earlier)
before the page posts its disconnect.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 10:00:36 +11:00
saphidandClaude Opus 5.5 9d92396d49 My Meta library: Sign in with Meta cancels by in-flight attempt and saved owner
- frame_meta keeps only in-flight sign-in attempts (removed when their connect
  finishes) instead of a bounded list of cancelled ids, so other cancels can't
  push one out, and records which attempt saved the account. Cancelling an
  attempt being checked stops its save; cancelling the attempt that just saved
  deletes that token; a later sign-in's or a pasted token's account is never
  touched; unknown attempts change nothing.
- will-redirect applies the per-frame rule: frames may be redirected to any
  https: page, the window only to Meta's sites.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:54:38 +11:00
saphidandClaude Opus 5.5 be2eda8304 My Meta library: Sign in with Meta lets frames load any HTTPS page
Meta's login can embed captcha and CDN frames from other sites (Arkose Labs,
hCaptcha, reCAPTCHA, fbcdn), so frames inside the sign-in window may now go to
any https: address (or about:blank/about:srcdoc); other schemes (http:, data:,
blob:, javascript:, file:...) are still stopped. The window itself stays on
the Meta/Oculus/Facebook allow-list. Frames can't reach the cookie read or IPC.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:44:32 +11:00
saphidandClaude Opus 5.5 f5bed08931 My Meta library: Sign in with Meta drops abandoned attempts, holds frames to the list
Review fixes for the sign-in window:
- Each sign-in has a random attempt id. Closing, timing out or signing out
  stops cookie intake first, then tells the server (new cancel-signin action)
  not to save a connect from that attempt, even one still waiting on Meta
  (frame_meta.connect/cancel_signin; 409 from the server). Disconnect signs
  out of the window before deleting the token.
- will-frame-navigate holds every frame to the allow-list; a frame going
  elsewhere is stopped, never opened outside.
- The window is destroyed, not closed, and will-prevent-unload is overridden,
  so beforeunload can't keep it alive; sign-out clears the session only after
  the window is gone.
- Permission checks refused as well as requests; no developer tools.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:39:25 +11:00
saphidandClaude Opus 5.5 291b8ec295 My Meta library: Sign in with Meta in a Frame Control window
The desktop app can now open Meta's own login page (https://secure.oculus.com,
as BSManager's "Log in with Meta") in a window with its own session
(persist:meta-signin: sandboxed, context isolation, no preload or Node,
permissions refused). Navigation stays on https oculus.com, meta.com and
facebook.com; other links open in the system browser and other redirects are
stopped. The oc_ac_at cookie is read only from that session and only for
oculus.com, sent once per value through the server's existing connect path
(check_token, who am I, 0600 file), and the window closes. The page only gets
the server's answer, never the token. Disconnect also signs out of Meta in
Frame Control by clearing that session. Pasting a token stays as the fallback.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-14 09:26:18 +11:00
saphidandClaude Opus 5.5 f5f3a1f9ca Releases: exact draft match and a fixed release page (review fixes)
- publish-release.sh: the fallback takes only an untagged-... draft titled
  exactly "Frame Control X.Y.Z" (optionally ": subtitle"); a pre-release's
  draft ("9.8.7-rc.1") or one bound to another tag is refused.
- updater.js: ignore the page in update.json/the API entirely and always
  link to releases/tag/v<version> built from the validated version, so a
  path like tag/..\..\other/repo can't reach shell.openExternal.
- Tests for both, including backslash and %2e%2e traversal.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-08 18:50:00 +11:00
saphidandClaude Opus 5.5 a41f635a7c Releases: publish drafts through REST and link update.json to the tag page
publish-release.sh took update.json's "page" from the draft's url, which is
releases/tag/untagged-... and dies on publishing; 0.4.0's manifest shipped
that dead link. It also looked the draft up with `gh release view <tag>`,
which reports "release not found" while a draft's tag_name still reads
untagged-... (and uses rate-limited GraphQL).

- publish-release.sh: REST only. Checks the tag exists, finds the release by
  tag_name or else the one untagged draft titled "Frame Control X.Y.Z",
  keeps the 8-installer digest check, replaces update.json via the uploads
  API, then PATCHes tag_name/draft/prerelease/make_latest. Adds --dry-run.
- updater.js: trust only this repo's releases/tag/<tag> pages (never
  untagged-...); otherwise link to releases/tag/v<version>.
- release.yml: one draft job before the matrix (no racing creates), with
  --verify-tag and tag_name set explicitly.
- Tests: tests/test_publish_release.py with a stand-in gh (tests/fakegh/gh),
  and an updater test for the page.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-08 18:41:37 +11:00
saphidandClaude Opus 5.5 9eeca79b5d Analytics, self-update and Report a problem
- Anonymous PostHog analytics (ui/frame_telemetry.py): usage on by default
  after a first-run notice; compatibility results and error details opt-in,
  offered together by the notice's "Share more to help fix problems" button.
  Random id, no person profiles or GeoIP, scrubbed text, an offline outbox,
  and "Show what's been sent" in the new Privacy panel. Inert without a
  project key, from a source checkout, or with DO_NOT_TRACK=1.
- APK installs now record install_failed when the APK itself won't install,
  and offer a 20-second test after installing. Opted-in reports reach the
  shared database through PostHog and `frame_compat_db.py sync`.
- The desktop app updates itself from published releases (app/updater.js):
  update.json from releases/latest/download, SHA-256 checked, no downgrades;
  macOS bundle swap, Windows NSIS, Linux AppImage, otherwise the release page.
  scripts/publish-release.sh publishes a tested draft with its manifest.
- Report a problem (header button, Privacy panel, Help menu) files a GitHub
  issue through the website's feedback API, with a previewed, scrubbed
  diagnostics snapshot; activity and logs only when asked for.

Reviewed by GPT-6 Astra (xhigh, read-only) three times; all findings fixed.
Docs: docs/privacy.md, docs/releasing.md.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-09-28 17:34:21 +10:00