Giovanni Barillari

@apapsch I would rather prefer a default to `0664`, configurable to `0666` if needed.

@RobertoPrevato regarding tests, do you have any advice? I don't see any specific tests for the scribe module. Should I just add a test in `test_responses` for this?

Hey @RobertoPrevato I'm the one who actually feels sorry for this; I hadn't chance to continue working on this since my last comment. The current implementation kinda provides the low-level...

From preliminary tests it seems performance is not affected. At least on MacOS arm.

Yes, this is kinda a duplicate of #323, at least the root cause is the same. Given the README states: > Due to the nature of such handlers some libraries...

I'm afraid you can't really interact with contextvars in Granian access log. Those log lines are emitted after the whole async application object returned, thus the span – and trace...

> It would be nice to let the app be aware of such cases and raise an exception in the `send()` method: https://asgi.readthedocs.io/en/latest/specs/www.html#disconnected-client-send-exception. This would require to move from spec...

#225 has some initial work on this. If anyone want to take over and open a refreshed PR for this I'll be happy to review and merge it.

jfyi, starlette 0.38.1 was just released with `pathsend` support removed. Given you rely on FastAPI, you should wait for them to bump the Starlette dependency; with that Granian should work...

@2-5 yeah, I'm sorry about that. I'm actually considering to just drop the `--opt` option and related code (maybe after deprecating it for a few releases). The actual benefits have...