dirrao commented on code in PR #39908:
URL: https://github.com/apache/airflow/pull/39908#discussion_r1672235407
##########
tests/_internals/forbidden_warnings.py:
##########
@@ -62,6 +62,11 @@ def pytest_itemcollected(self, item: pytest.Item):
# Add marker at the beginning of the markers list. In this case,
it does not conflict with
# filterwarnings markers, which are set explicitly in the test
suite.
item.add_marker(pytest.mark.filterwarnings(f"error::{fw}"),
append=False)
+ item.add_marker(
Review Comment:
> As I said on Slack, I don't see anything that changes my previous comment
from last month.
>
> > To add to that, even if we do standardize them, why break two (Datadog
and Otel) to match one (StatsD)? If we were going to do this, shouldn't we
change the one that doesn't match the other two?
It's not about breaking. It's about doing the right thing. If we go with the
milliseconds, then we have high precision and I am not sure with seconds. If
seconds is the right way, then we will go with it. Let's hear from @potiuk /
@kaxil.
>
> We get YACO (yet another config option) but we're still breaking two
backends.
This option is temporary and required during the transition and we can
remove it at a later point in time. If you have any better option in mind,
then please do share it with us.
> Without some usage numbers to see what percentage of users are going to be
affected here, I don't agree with this.
I do agree the consumers are impacted. I don't think we have information on
the number of airflow consumers by their telemetry setup.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
