Thomas Eizinger

Results 1888 comments of Thomas Eizinger

I have a similar motivation. I wanted to explore with moving the printing of `span`s to the end of the log line instead of at the beginning. That would make...

> Can you make those public? Maybe instead of that, they could be moved to a `tracing_subscriber_utils` crate? That way, they are reusable for others without extending the public API...

> That way it would be possible to introduce breaking changes to the utility crate without affecting the main one. Yep. Just needs someone to step up and create it....

You can still pre-slice the contents if you want. The API doesn't change (still accepts a `Transmit`). What I am suggesting is that `quinn-udp` can do an upper bound slicing...

If we don't want to move the slicing into `quinn-udp`, then I think we should at least add a check that `contents` is not bigger than 65535. In the code...

> > Ensuring that we only pass a slice of at most 65535 bytes if the kernel supports GSO. > > If I recall correctly, the maximum UDP datagram size...

Another datapoint for moving this logic into `quinn-udp` is that it is quite easy to make very subtle mistakes here, see https://github.com/firezone/firezone/pull/8932.

In that case, I think the API is quite misleading considering that similar operations on `Vec` or slices in Rust behave very differently. I guess we can't detect whether the...

And how much value does it have to match the C API? I am using Rust so that I don't have to deal with (so many) C APIs! If there...

> Most Rust data structures panic if you access something out of bounds. We can't do that so an error seems like the next closest thing. I have to retract...