Open marten-seemann opened 1 year ago
Is it premature to stop testing mplex until this comment https://github.com/libp2p/specs/issues/553#issuecomment-1592857250 hold true and it's actually deprecated in the libp2p/specs?
The first step would be for every client to have yamux enabled by default (which requires stable support in their underlying libp2ps, and proper testing) Once that is done & we are confident in all yamuxes, we can update the specs to remove mandatory support
mplex also seems to acheive better performance than yamux: https://github.com/ChainSafe/js-libp2p-yamux/issues/42 in js-libp2p
This is more about documenting what we're already doing now. We're not fixing bugs or performance issues in mplex, instead, we're already skipping tests (see the transport integration test effort in go-libp2p).
If we resolve the specs issue first and then disable the tests or vice versa will most likely not make any difference.
mplex is being phased out. go-libp2p and rust-libp2p have removed it from the list of default muxers a long time ago. Kubo is considering removing it as well: https://github.com/ipfs/kubo/issues/9958.