![]() Updated by running `python3 -m script.gen_requirements_all`. New or updated dependencies have been added to `requirements_all.txt`. Updated and included derived files by running: `python3 -m script.hassfest`. The has all fields filled out correctly. Documentation added/updated for [If the code communicates with devices, web services, or third-party tools: If user exposed functionality or configuration variables are added/changed: Tests have been added to verify that the new code works. The code has been formatted using Black (`black -fast homeassistant tests`) ![]() There is no commented out code in this PR. **Your PR cannot be merged unless tests pass** The code change is tested and works locally. ![]() Code quality improvements to existing code or addition of tests Breaking change (fix/feature causing existing functionality to break) Deprecation (breaking change to happen in the future) New feature (which adds functionality to an existing integration) Bugfix (non-breaking change which fixes an issue) Two tests got fixed in #82788 for the 3.8.3 upgrade, which are at least 2 spots that would need to fix. This has not been released yet.īefore upgrading, we should explore if we want cancellation and if we don't, we should fix the `StreamResponse` eating all the memory. In a new parameter was added that allows enabling the old cancellation behavior. It seems like it contains a bug where a `StreamResponse` will not raise when written to, causing our MJPEG streams to eat up all the memory (as it writes a new image every 0.5s to the stream without any back pressure). We did not anticipate such a behavioral change in a patch release and so didn't catch this in time. 3.8.3 included "fix" to align the behavior of aiohttp with the documentation, which is that request handlers no longer get cancelled when the client disconnects.
0 Comments
Leave a Reply. |