fix: Charge time remaining, account for time passing between updates #419
Conversation
|
For this type of nuanced issue, do you think you can write a test so we don't get a regression? Basically, the idea is the test will fail if this fix's behavior is undone. |
|
Hmm, my first thought on how to test would be to put a delay of 1 minute in and make sure the state has not changed. I'd hate to hold up the all the tests for 1 minute though. I included this behavior in the Destination arrival time PR. So that could end up with 2 minute delays in the tests. Is there a way to programmatically change the system time? Or run the tests with delays in a separate thread? |
|
Yes, use libraries like https://pypi.org/project/time-machine/. I don't remember what HA has standardized on though. |
|
Moving to a draft for now. This might not be needed if the polling is increased to every 60 seconds while charging. The charge time completion will always be within 60 seconds. |
InTheDaylight14
force-pushed
the
charge-time
branch
from
402fcd4
to
1fc3c4e
Compare
| @@ -261,6 +261,15 @@ async def test_time_charge_complete_charging( | |||
| assert state.attributes.get(ATTR_DEVICE_CLASS) == SensorDeviceClass.TIMESTAMP | |||
| assert state.attributes.get(ATTR_STATE_CLASS) is None | |||
|
|
|||
| mock_timetravel = mock_now + timedelta(minutes=2) | |||
There was a problem hiding this comment.
Little out of my element here, I want to draw attention to this to make sure this works for testing. Trying to change the time 2 minutes in the future and make sure that the state has stayed the same.
There was a problem hiding this comment.
I'm not certain what the issue is. Is the test failing? I haven't mocked timestamps before so I'd just be reading the documentation like you are and reviewing the HA examples.
There was a problem hiding this comment.
No the test is passing just fine. Maybe @carleeno can review the mock_timetravel. I based it off of their recent PR to shore up time tests. I just want to make sure that the state is actually being evaluated two minutes later by using that monkeypatch.setattr...
There was a problem hiding this comment.
@InTheDaylight14 the mock looks correct, however you're not asserting anything afterwards so it's not actually testing anything.
Add an assert state.state == charge_complete_str at the end and see what happens, this will reveal if your logic is good.
There was a problem hiding this comment.
for an explanation of what the monkeypatch is doing is it's replacing the homeassistant.util.dt.utcnow that HA uses with our mocked time. So when dt.utcnow is called from sensor.py it will actually get your mocked value.
InTheDaylight14
force-pushed
the
charge-time
branch
from
d2d0918
to
53397c6
Compare
|
Oh and it looks like the tests are being skipped in the checks. Not sure what is going on there. |
Saves the time when the value changes, subtracts the time that has passed since the last update.
Without this the value grows by 1 minute for every minute since an api update. Then, on api call, the time jumps back down to the proper time.