Sümer Cip

I am also suspecting that there is some issue related with profiler+event loop happening however I could not yet confirm. In parallel, I am thinking to implement a very simple...

Thanks @euri10 , @sm-Fifteen . Still not %100 but I might have identified the problem: the tagging code is taking too much time as it is a basic linked list...

Hi everyone, I think I have identified the issue: the issue is when you call `yappi.get_func_stats()` from `dispatch` you are always enumerating *ALL* collected traces and then select ctx_id's into...

> Anyway, if it's a small change I'd still be happy to merge it. And I'm also happy to just discuss ideas on how to make things faster in general!...

> IMO this is a strong argument for stopping the profiling with an error if we detect that it's not going well. Agreed. My concern was to detect situations like...

> About related work: I'm starting a new job on Monday (joining Datadog's profiler team), and there is a decent chance I'll work on related stuff a lot going forward....

Hi. This is the issue we worked with few FastAPI devs on the subject: https://github.com/tiangolo/fastapi/issues/701 and the outcome of that issue was a simple middleware that you can use in...

We have recently released a new version of yappi (`1.3.6`) that has wheels compiled for most of the platforms. I would suggest using that version. I am closing the issue...

Thanks. I will look into this in detail. BTW, I realized we do not have any callgrind format unit test check. Do we have any way of ensuring the callgrind...

Thanks for the explanation. We have also discussed this before on https://github.com/tiangolo/fastapi/issues/701 and `clear_stats()` filtering might be worth implementing for using in long running servers/middleware as you proposed.