Get status once per device instead of per platform #51
Conversation
|
I am getting (sorry, too late to debug for me, going to bed...): |
|
And still I get occasional of these errors: |
postlund
force-pushed
the
single_connection
branch
from
55396d2 to
ee8e7b9
Compare
|
Some relic override of Maybe no connection problems was a lucky shot for me. |
|
I've done some research, and somebody says that this behavior might be caused by opening and closing the socket for each request: codetheweb/tuyapi#84 . |
|
PR #51 merged into master. |
I've sketched something, and it does seem to keep the connection more robustly... |
|
I agree, we should definitely move towards a persistent connection. I think I included that in my "future" issue (#15). Also, clean up and convert to asyncio. If you have something up and running worth looking at, open a PR and we can look at it together 😊 |
|
OK @postlund , but before we move to the persistent connection, I have something else to submit to your attention. In detail:
After we've fixed these 2, we can move on to persistent connection. |
|
Jeeez, I've struggled really hard to comply to the lint and pydocstyle remarks, don't you think some rules are a bit too strict? Edit: nevermind, I figured it out and now I have set up my environment for local tests, but still believe that some rules are really cumbersome.... |
|
As long as you don't override 88 characters in comments, running |
|
Well, things like "Words must be imperative, not in 3rd person", or "missing trailing period"... I found them a bit too much 😆 |
|
PS, what about the bug I reported? |
|
Yeah, ok, some of the pydoc stuff can be a bit much.... 😉 Yeah, you are right regarding the bug. Since we now have an external loop that updates everything, service calls can't trigger additional state changes. The trivial way would be to somehow trigger an update after changing a DP. This has a small cost of course, since we have to fetch all DPS again, but this only happens when changing a DP which is very seldom when amortized over time (so a drip in the ocean). But, since you have already made a more optimized version we can probably go that route as well. We just need to send out the updated status (full cache) to all entities. I can add some comments in your other PR. |
* Rework SubDevices Connection * Fix some stuff in payload_dict for sub devices `3.3` * Don't send warning disconnect dc is intended * Hotfix for `3.3` devices `detect_available_dps` * Add sensors for code `wsdcg` * Attempt Fix: some dps aren't detected. * Adjust `detect_available_dps` Function when gateway existed * int `t` since sub_devices payload `t` is int * revert last commit force int `"t"` always, need for some cmds * test: disable removing the [ gwID, devId and uid ] * Disconnect the sub devices if gateway dc'd * ensure that the gateway is ready before setup subdevices * Fix: Block status_updated if gateway is sub_device * Add gateway_gwId to discovered sub devices. and prevent non local subdevices * Store Gateway ID If found on mergeDevicesList * minor changes of connect on initialization * config_entry_by_device_id search for gateways * Minor fix: Rename climates.py in tuya devices data * Tuya Devices Data rename binary_sensor * Adjust Auto Configure category `wkf` * Fix automatic update `ip` and includes sub_devices * Except errors from pytuya. * Except error on config flow and set timeout for connect. * Fix: Entity_Category and convert get_gateway to async * Fix Mirgate: Force Int for config_flow values and fix reverse always off * Fix: Fake Gateway fails due to no DPS found on parent DPS. * reformat unload function. * Fix Reconfigure cloud step * Fix error if "Auto Configure" used without cloud. * Add platform support for humidifiers and Rename DPS, and DPS_CONF Functions (rospogrigio#47) * Add support for humidifier platform * Add Humidifer to Auto Configure feature. * Improve English translations to be more user friendly rospogrigio#45 by @codyc1515 * Adjust `en` translation for humidifer * mark set_humiditiy as optional * log the disconnect reason of exists * Adjust get_gateway function * adjust error msgs in auto configure * Refactor: entities initialization msg and remove dp_value_conf (rospogrigio#49) * Initialized msg to `common.py` and del: `dp_value_conf` * def should_pull to _attr_should_pull and refactor restore_on_reconnect * refactor entity category * Refactor: `device_info` and properties annotation * Improve entry initialization (rospogrigio#50) * Connecting `CloudAPI` runs on background without interrupt initialization. * Unload: pulls platforms from `TuyaDevice` data Explain: rospogrigio#50 * Improves config flow (rospogrigio#51) * DPS Data now will be pulled with devices data, data stored in `device_id` -> in `devices_list` with `dps_data` key. * Adding new device DPS Data will be pulled from stored data `except the value!` * Devices list in config flow now will be sorted: Starting from known devices first. * Refresh cloud devices data when configure opened. `previously reload was needed in order to pull new devices data.` * Handle the errors if refreshing token failed. * Enable manually enter template filename (rospogrigio#52) * Templates list field are now insertable and searchable, so if you add templates new template into `templates` directory you can manually enter there name without need to load them up with HA boot. * Templates names now will shows the same as filename with extension. * update pytuya version * revert auther name * Sort discovered devices by `ip` * Handle sorting discovered devices better * refactor: devices list sorting * adjust dc reason log * refactor: mergeDevicesList * new helper `get_gateway_by_id` refactor mergeDeivceList * typo * * Refactor codes a little bit to make it easier to maintain. (rospogrigio#53) * * Refactor codes a little bit to make it easier to maintain. * cloud_api now has async_connect function. * Localtuya HASS Data is now stored as namedtuple `HassLocalTuyaData` * Removed `TUYA_DEVICES`, `DATA_CLOUD` and `UNSUB LISTENERS` * Sotred all unsub_callback in tuya_devices.unsub_listeners * Reconnect if disconnected will be called after 2secounds. * Fix reconnect after 2 secounds. * Hide reconifgured device if no devices setuped (rospogrigio#54)
This is the one that needs some serious testing.