junlinp

Obj decoder support Double-precision floating-point

1

I found that `ObjDecoder::ParseVertexPosition` parses the position using single-precision floating-point. can it support the double-precision floating-point?

It seems that the source code don't implement the case N = 4 when choose the null vector of M^TM? It is sufficient in practice?

two-factor authentication problem

1

first time I send a request with `node activate.js $email $password $ulf_file`, I will get a error `await input.uploadFile(alfPath)`. when My phone receive a auth code and I send request...

sometimes, watchdot thread will create multiple event thread.Then crash will occur since `pollNal` methods is not thread-safe. `pollNal` process frame data with three step: 1. query the frame length in...

Why queue two frame at most instread of read the latest frame?

1

https://github.com/alvr-org/alvr-visionos/blob/6005e8c20c8d82c39f05463c8a5b6f9abc1ac93d/ALVRClient/Renderer.swift#L598 should it read the latest one frame in the queue? two frame cache would cause 22 ms around latency when the render rate is 90Hz.

``` Tue Dec 17 2024 15:21:08.297 [Info] - Unable to load driver openvr_driver_test from D:\中文路径\openvr_driver_test\bin\win64\driver_openvr_driver_test.dll (126) Tue Dec 17 2024 15:21:08.297 [Error] - ASSERT: "DriverManager alvr_server load error 126u" at...

Explicitly set `weights_only=False` because, starting in torch 2.6.0, the default value of `weights_only` in **torch.load** changed from False to True.