ty error in remote machine

[ak24watch]

Describe the bug

ty error when kernel restarted in remote machine

Will you submit a PR?

• Yes

Environment

Details

{
  "marimo": "0.23.14",
  "editable": false,
  "location": "/home/AK/.cache/uv/archive-v0/U5FSDwdnffYjOX_ukgVrw/lib/python3.12/site-packages/marimo",
  "OS": "Linux",
  "OS Version": "6.17.0-35-generic",
  "Processor": "x86_64",
  "Python Version": "3.12.9",
  "Locale": "--",
  "Binaries": {
    "Browser": "--",
    "Node": "--",
    "uv": "0.8.4"
  },
  "Dependencies": {
    "click": "8.4.2",
    "docutils": "0.23",
    "itsdangerous": "2.2.0",
    "jedi": "0.19.2",
    "markdown": "3.10.2",
    "narwhals": "2.23.0",
    "packaging": "26.2",
    "psutil": "7.2.2",
    "pygments": "2.20.0",
    "pymdown-extensions": "10.21.3",
    "pyyaml": "6.0.3",
    "starlette": "1.3.1",
    "tomlkit": "0.15.0",
    "typing-extensions": "4.16.0",
    "uvicorn": "0.51.0",
    "websockets": "16.1"
  },
  "Optional Dependencies": {
    "loro": "1.13.1",
    "python-lsp-ruff": "2.3.1",
    "python-lsp-server": "1.14.0",
    "ruff": "0.15.21",
    "ty": "0.0.58"
  },
  "Experimental Flags": {}
}

Code to reproduce

No response

[ak24watch]

this is the ui error, i have installed it

[rehan243]

Oh, the ty error after a kernel restart, yeah, I've hit this before. From your environment, it looks like you're on ty==0.0.58. IIRC, there was a bug in some of the 0.0.x versions of ty where it didn't properly handle serialization/deserialization after a context switch, and that could blow up when the underlying kernel state resets. Not ideal.

What worked for me was bumping ty to at least 0.1.x. It's not a direct drop-in, though, you might need to tweak any type annotations or decorators if you're using them in your code. If upgrading isn't an option, you could also try explicitly reinitializing ty after the kernel restart, though that's a bit hacky and depends on how you're wiring it up.

Curious, are you seeing this issue consistently, or does it only happen under certain conditions (e.g., heavy workloads, specific configs)? Wondering if something else might be interfering.

[ak24watch]

0.59 is latest but you tell how 0.1.x?

[ak24watch]

Oh, the ty error after a kernel restart, yeah, I've hit this before. From your environment, it looks like you're on ty==0.0.58. IIRC, there was a bug in some of the 0.0.x versions of ty where it didn't properly handle serialization/deserialization after a context switch, and that could blow up when the underlying kernel state resets. Not ideal.

What worked for me was bumping ty to at least 0.1.x. It's not a direct drop-in, though, you might need to tweak any type annotations or decorators if you're using them in your code. If upgrading isn't an option, you could also try explicitly reinitializing ty after the kernel restart, though that's a bit hacky and depends on how you're wiring it up.

Curious, are you seeing this issue consistently, or does it only happen under certain conditions (e.g., heavy workloads, specific configs)? Wondering if something else might be interfering.

in remote it happens always

[mscolnick]

@ak24watch so ty works but when you restart the kernel, it no longer works? can you describe your bug a bit more so we can help?

[ak24watch]

@mscolnick when connected to remote server, for first time the ty error did not came when restarted the kernel i got error .

again ty full functionality like incremental type checking and lsp server is not working even in local no error is been thrown but still not works, also in remote

please do not use molab, i am talking about marimo runinng in your own computer or remote computer.