Thomas Eizinger

> I'd also be open to adding zstd compression for the cached baseline JSON file if you think that's worth it independently I am leaning towards that being separate. Compared...

> How can I help? I'm happy to review draft PRs or whatever else you need. I'll open a draft PR once I have some questions and/or an initial design,...

Having seen the internals now, I have a question: Why are we building the docs for the _current_ revision in a separate directory? Why not just reuse the default location?...

Right, that makes sense. I'll see if I can optimise that on my end by just adding another cache!

Damn, that is amazing! :) Looking at our current CI, I just had another idea. At the moment, we are creating and parsing two rustdocs, only to then say 0...

> Another use case for this is in [hyperium/tonic](https://github.com/hyperium/tonic) where there are multiple crates that are intended for testing purposes only and are not published anywhere. We solve that by...

This includes a tiny change in behaviour. A remote can observe that the authentication handshake failed and for them, the connection is thus never established. With `allow_block_list`, the connection is...

@conectado The latest panic seems to be here: https://github.com/algesten/str0m/blob/ff331bd958bb91ad6d189e5aa32e550883c8b036/src/io/stun.rs#L82. Are we reading a zero-length packet?

@conectado You need to drop acdbe8fba5300088e473610487d9c40eeb78a178 during rebase to avoid conflicts as that PR already merged now :) I had just cherry-picked the commit.

> > > > > @conectado The latest panic seems to be here: https://github.com/algesten/str0m/blob/ff331bd958bb91ad6d189e5aa32e550883c8b036/src/io/stun.rs#L82. > > > > > Are we reading a zero-length packet? > > > > >...