Skip to content

[Bug]: iOS 2.1.0 thread list misses days of threads from online environments; survives force-quit, fixed by toggling the environment #24997

Closed as not planned

Description

@goleary

Note

Claude Opus 5.5 posting on behalf of the reporter.

Before submitting

  • I searched existing issues and did not find a duplicate.
  • I included enough detail to reproduce or investigate the problem.

Area

apps/mobile

Steps to reproduce

Setup: the iOS app is paired to two environments (a MacBook and a Windows desktop), both online. The desktop app on the MacBook is also connected to the Windows environment.

  1. Use both machines normally for a few days, creating threads from desktop and from the phone.
  2. Open the iOS app's Threads list ("All environments", "All projects").
  3. Compare it with the desktop sidebar.

I don't have a reliable repro for how the phone gets into this state.

Expected behavior

The mobile thread list matches the desktop sidebar for the same environments.

Actual behavior

The mobile list was missing almost every thread created or updated in the previous ~4 days, across both environments. After the pinned threads, the list went from one thread updated a few minutes earlier straight to threads last active 4 days ago.

  • The only recent threads that appeared were pinned ones and one thread I had opened on the phone that day. Several missing threads had been created from the phone (creationSource: mobile).
  • The missing threads were not settled or snoozed. Desktop showed all of them.
  • Force-quitting and reopening the app did not fix it.
  • Turning the MacBook environment off and back on in the mobile app fixed it: the missing threads appeared right away.

So the phone seems to keep showing a saved copy of the thread list and does not replace it with the full snapshot it loads, until the environment is reset.

Impact

Major degradation or frequent failure: threads were silently missing with nothing showing the list was out of date.

Version or commit

iOS app 2.1.0. MacBook server 0.1.2 (T3 Code (Beta) desktop, macOS arm64).

Environment

iPhone (iOS 27) paired to a macOS arm64 laptop and a Windows desktop.

Logs or stack traces

Server traces from the MacBook while the phone was showing the incomplete list. The phone loaded the full shell snapshot successfully, then its subscription ended about 40–100 s later:

13:38:56 ShellStream.loadShellSnapshotParts      Success  69ms
13:38:56 http.orchestration.loadShellSnapshot    Success  69ms
13:38:56 ws.rpc.orchestration.subscribeShell     Interrupted after 98s
13:40:37 ShellStream.loadShellSnapshotParts      Success  78ms
13:40:37 http.orchestration.loadShellSnapshot    Success  79ms
13:40:37 ws.rpc.orchestration.subscribeShell     Interrupted after 39s

Over the same period the server's thread list (t3_thread_list) returned every missing thread as unsettled and not snoozed. I have no runtime logs from the phone.

Related: #24530 (cached threads from an offline machine shown as current; here the environment was online), #24344 (iOS 2.1.0 stale state for newly created threads, but that one clears on app restart), #24909 (iOS 2.1.0 short-lived sockets).

Activity

  1. juliusmarminge commented on Oct 26, 2026

    @juliusmarminge
    Member

    Not reproduced on main at 3de45c6254: iOS simulator dev client paired to two online isolated test environments. Added six explicitly unsettled, unsnoozed thread fixtures spanning today through five days ago while the first server was stopped. All six appeared after reconnecting, survived app termination/relaunch, and remained after toggling the environment off/on. This tests snapshot replacement and restart, but does not recreate several days of normal use.

    The missing piece is phone-side diagnostics while the list is incomplete, before toggling the environment. Can you provide those logs with the time of a reopen and one missing thread ID (or title), so we can correlate the loaded snapshot with the list the phone displayed? The server traces already provided show successful snapshot loading but do not show whether the phone applied it.

    All six dated fixtures after reconnect

    All six dated fixtures after relaunch

  2. tablepolar commented on Oct 31, 2026

    @tablepolar

    Note

    Written by Claude Opus 5.5 (claude-opus-5-5) on behalf of the reporter.

    Also seeing this on iPhone 17 (iOS 26.6.1, T3 Code app 2.2.0), with "All environments" and "All projects" selected. The original report was on 2.1.0, so this is still happening in 2.2.0.

    What happened (Oct 31, 2026):

    • I created a thread from the phone, sent a message with photos, and got a reply. The thread was finished and wasn't settled, snoozed, or archived.
    • As soon as I left the thread, it disappeared from the Home list. Its row should have been at the top, but the list jumped straight to older threads.
    • Searching from desktop showed the thread was fine on the server.
    • Force-quitting and reopening the app fixed it for me (unlike the original report).
    • Another time, I created two threads close together and only one showed up in the list.

    It happens sometimes, not every time. It's disrupting my daily workflow because I often can't get back to a thread I just made on the phone.

    Possibly related: #27617 (a missing snapshot during thread creation is treated as permanent) and #27894 (the iPad sidebar stops showing new threads until the app is relaunched).

  3. juliusmarminge commented on Nov 12, 2026

    @juliusmarminge
    Member

    @goleary Next time threads are missing, before you force-quit or toggle anything: does a missing thread appear if you set the list's environment filter to just that machine, or search for its title on the phone? Please also tell us roughly how long the app had been in the background, and whether the list header said 'Loading threads…' or 'Synchronizing'.

  4. radres commented on Nov 14, 2026

    @radres

    Having similar problems with threads disappearing and reappearing on t3 iphone app where I am connected to remote environment.

  5. MorielLabs commented on Nov 16, 2026

    @MorielLabs

    Note

    Written by Claude Opus 5.5 (claude-opus-5-5) on behalf of the reporter.

    Seeing this on Android too. This answers the questions from the Nov 12 comment: filtering to the one machine did not bring the threads back. Filtering to their project did.

    Setup: Samsung Galaxy S26 Ultra, Android 17, T3 Code Android app 2.3.0 (versionCode 784). The phone is paired to three environments. The missing threads are on a Windows x64 laptop running server 0.1.9, connected through T3 Connect. The other two are cloud VMs added by URL. All three showed as connected the whole time.

    What happened (Nov 16, 2026, UTC):

    1. Around 04:34 the user couldn't find a thread on the phone that showed on desktop. The laptop's server restarted at 04:50.

    2. Around 05:00 I opened the app on the phone. Before that, another app was in the foreground; I don't know how long T3 had been in the background. The list was set to "All environments" / "All projects". Two threads from the same laptop project were missing:

      • 2d7871cc-4a14-4410-9d1f-b3519885d3f1, last updated about 5 hours earlier.
      • 96e7fb78-fc57-45a2-a282-36444f427f8b, active 2 minutes earlier. It was the most recently updated thread on that machine.

      Neither was settled, snoozed or archived (t3_thread_list on the laptop confirmed this). Other threads from the same laptop and the same time range did show. The list header didn't say "Loading threads…" or "Synchronizing".

    3. Search on the phone: "clip" didn't show either thread in the first results, though one has "Clip" in its title. I didn't scroll the results, so this one is less certain.

    4. Environment filter set to just the laptop: both threads still missing. The top rows were unchanged, and the 5-hour-old thread should have been 4th.

    5. Project filter set to their project: both threads appeared right away, with "Settled (11)" under them.

    6. Project filter back to "All projects": both stayed, in the right place, and search found them. I didn't need to force-quit the app or toggle the environment.

    One detail that may or may not matter: the folder of the project with the missing threads (C:\Dev) is the parent folder of several other projects on the same machine.

    Server traces from the laptop over the same period show the pattern from the original report. The snapshot loads fine, and the shell subscription is interrupted after a few seconds to a minute or so:

    ShellStream.loadShellSnapshotParts       Success
    ws.rpc.orchestration.subscribeShell      Interrupted after 3.9s
    ws.rpc.orchestration.subscribeShell      Interrupted after 27s
    ws.rpc.orchestration.subscribeShell      Interrupted after 84s
    ws.rpc.orchestration.subscribeShell      Interrupted after 32s
    ws.rpc.orchestration.subscribeShell      Interrupted after 2.7s
    ws.rpc.orchestration.subscribeShell      Interrupted after 31s
    

    I couldn't tell which of these came from the phone. adb logcat for the app's process had nothing from the app, only Android rendering messages.

  6. juliusmarminge commented on Nov 19, 2026

    @juliusmarminge
    Member

    Closing: no reply in 7 days after we asked for more information. Comment with the missing details and we will reopen.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions