Vsevolod Tolstopyatov

Also, these feature flags should have a clear priority over debug mode -- it should be possible to e.g. enable stacktrace recovery without enabling debug mode itself for more fine-grained...

UDP: Fleet disables stacktrace recovery because of `setName` overhead, but would like to enable it back

Yes, that's it

Nice write-up, thanks! I have an idea that should cover most of the reports regarding the inflexibility of structured concurrency. Have to wrap my mind around it to write a...

See also #3679 and #3681

@dkhalanskyjb WDYT? Do you have any objections or comments? This PR is draft because it has a bunch of tests failing, mostly guide-related once where coroutine id is part of...

>I don't understand why the debug mode is enabled in production I tend to think completely the opposite. We should eliminate as many barriers as possible for folks to keep...

Thanks for the detailed explanation. For me, it looks like a very subtle problem without delay design -- depending on the application's machinery, single-threaded dispatchers (multi-threaded dispatchers as well, but...

For things like timings, it's indeed preferable to introduce your own dispatcher, I believe we cannot introduce a general enough mechanism that fits them all. >16 is arbitrary, it is...

Unfortunately, it is not `sample` problem but a much deeper one. We have a flow invariant that explicitly prohibits emissions from context different from the context target flow is being...