tapo-py 0.3.1 code hangs on timeout

[mihai-dinculescu]

Two users (https://github.com/mihai-dinculescu/tapo/commit/c91ee35c6ca01c282bfafd0cb99fd5384093bb6a#commitcomment-144217052 & https://github.com/mihai-dinculescu/tapo/issues/228#issuecomment-2209587753) have reported that after the migration to pyo3 0.21, the code hangs when timeouts occur.

I have been able to reproduce the problem, and it seems that the hangs are caused by the isahc's send_async future that never completes. Using the non-async send function works as intended.

However, I'd like to explore migrating away from isahc, as it no longer seems to be well maintained.

[TTSKarlsson]

Is there a workaround for now, like choosing a version that works? If so how do I use it?

[avadiax]

Hi, I noticed you have released 0.3.2. The timeout issue still continue to persist and I have continued to use 0.3.0. FYI.

[mihai-dinculescu]

@TTSKarlsson v0.3.0 is the latest good version that doesn't exhibit this issue. @avadiax, the fix for this proved more difficult than I thought, and I'm still exploring what's the best way to address the issue. Please bear with me 🙏

[mihai-dinculescu]

The fix is on the main branch. Could you please give it a try @TTSKarlsson, @avadiax, @gsaviane?

[avadiax]

The fix is on the main branch. Could you please give it a try @TTSKarlsson, @avadiax, @gsaviane?

Hi Mihai, unfortunately I install via pip, and pip will not install from the main branch since it does not detect setup.py or pyproject.toml in it. :( Not sure if I miss anything.

[TTSKarlsson]

I'm not sure how to run this.

My steps:

1. clone repo
2. cd tapo-py
3. pip install poetry
4. poetry install
5. poetry shell -> now inside the venv
6. export TAPO_USERNAME=_________
7. export TAPO_PASSWORD=_________
8. export IP_ADDRESS=the local ip for a tapo plug
9. py ./examples/tapo_p100.py

Result:

Traceback (most recent call last):
  File "F:\rust\tapo\tapo-py\examples\tapo_p110.py", line 7, in <module>
    from tapo import ApiClient
ModuleNotFoundError: No module named 'tapo'

[mihai-dinculescu]

@TTSKarlsson After step 5, you need one more, maturin develop. @avadiax these steps should work for you as well.

CONTRIBUTING.md should be updated to better capture this.

[TTSKarlsson]

It does turn it on and then off, but then fails to catch an exception.

[TTSKarlsson]

This exception doesn't happen on tapo_generic_device_toggle example btw

[mihai-dinculescu]

That's a separate issue concerning the changes introduced in 0.3.2 in response to firmware updates. Is your P11X device on the latest firmware?

[TTSKarlsson]

fw_ver is '1.2.5 Build 240411 Rel.143808' , if that's the firmware.

I've got a few different plugs. I'm testing on a P100, and this response is the same using examples tapo_p100 and tapo_p110.

[mihai-dinculescu]

Is your P100 on the latest firmware? If so, can you please paste the response of get_device_info_json?

[TTSKarlsson]

{
  "avatar": "ceiling_lamp",
  "device_id": "8022C03D80DD64FC390D5CA114CA6C31209FE8B8",
  "device_on": false,
  "fw_id": "00000000000000000000000000000000",
  "fw_ver": "1.2.5 Build 240411 Rel.143808",
  "has_set_location_info": true,
  "hw_id": "4D44B233807D5BDA45BEC69DC18AAE09",
  "hw_ver": "2.0",
  "ip": "192.168.2.206",
  "lang": "en_US",
  "latitude": 0,
  "longitude": 0,
  "mac": "30-DE-4B-21-D6-4D",
  "model": "P100",
  "nickname": "Computer",
  "oem_id": "6A0E161B05B8FB93E53988B66CE51E30",
  "on_time": 0,
  "region": "Europe/Amsterdam",
  "rssi": -51,
  "signal_level": 2,
  "specs": "",
  "ssid": "Emsie",
  "time_diff": 60,
  "type": "SMART.TAPOPLUG"
}

[TTSKarlsson]

I don't know how to know if the firmware is the latest, it seems from the build name that it's from april this year

[TTSKarlsson]

I should add that I've had these for years and I don't use any apps or anything, which means they do manage to update themselves whenever they want to.

[mihai-dinculescu]

The fix for P100 is on main. Can you please try again?

[TTSKarlsson]

👍🏼 Yes now it works, and the test also returns the device info and device usage

[mihai-dinculescu]

What about the main issue described in this ticket? Did the hangs that you've experienced with v0.3.1 and v0.3.2 go away?

[TTSKarlsson]

Yes, this issue is gone in the main branch with P100 that I tested.

If it passes for you, you should probably release it as it's going to be a better experience than having to install an older version. My main use now is still 0.3.0 until this releases.

I'll open a new ticket if the new release wouldn't work for some reason.

[mihai-dinculescu]

The change is out now as part of 0.4.0. I just needed a little bit of time to get another improvement in.

[avadiax]

The change is out now as part of 0.4.0. I just needed a little bit of time to get another improvement in.

I can confirm 0.4.0 looks to be stable during the testing. Will report if issues pop up.

I take it that you replaced isahc with reqwest?

[TTSKarlsson]

0.4.0 works here too.

[mihai-dinculescu]

Great, thanks everyone for trying it out, and apologies for the delay in fixing it.

Indeed, isahc was part of the issue. From what I gather, it has its own internal async runtime that doesn’t seem to play well with pyo3. Replacing it with another library that supports an external runtime allowed me to configure tokio in a way that works.