chore: upgrade to libp2p 0.38 #463
Conversation
|
Let me know in case you need any help on the rust-libp2p side. Also, there is v0.39.1 now. |
|
@mxinden thanks and thanks for letting me know of the 0.39.1! I think this one is looking pretty good other than the bet I made some months ago on multiaddrs staying peerid-less will be a chore, but I might just take up on your help offering when looking at our dht test related things, when I get to look at them next. I do remember there being a change in behaviour I wasn't able to track down to a PR, but maybe the whole idea in the test was originally faulty. I'll ping you later! Looking at the changelog for the latest release I think I'll take care of all of the issues found during this upgrade before moving on to multiaddr switching. |
I am not sure I follow. Can you expand on this? |
|
@mxinden thanks for your review on #472. I've been quite busy so also apologies for the delay:
So when we last worked on rust-ipfs in 0.34 days and before that, I understood that the wanted direction for rust-libp2p was to avoid supporting CLI use cases (which rust-ipfs mostly is) and keep the separation between multiaddresses and peerids. I understand this changed in libp2p/rust-libp2p#1931 -- before that it was explicitly unsupported to dial multiaddresses with peerids. Instead of arguing about this on the issues I chose to just work around the limitations, which now need to be cleaned up. |
472: chore: upgrade to libp2p 0.39 r=koivunej a=CHr15F0x This PR is a follow up of #463. It includes the following: - upgrading libp2p to [v.0.39.1](https://github.com/libp2p/rust-libp2p/releases/tag/v0.39.1) - upgrading some other deps (e.g.: prost, prost-base, quick-protobuf, base64), **excluding cid and multihash** - fixes to tests that started failing following upgrade to libp2p v.0.38 in #463 Incoming shortly: - [x] respective changelog update - [x] ~~I haven't run the changes against current workflow, so some additional fixes could be necessary;~~ edit: though I did check against all tests locally numerous times during dev, albeit only in a linux/x64 environment; btw looks like the workflows are fine with the changes Co-authored-by: Joonas Koivunen <joonas.koivunen@gmail.com> Co-authored-by: Krzysztof Lis <klis33@gmail.com>
Thanks @koivunej.
Please do on future issues. Always appreciate additional perspectives. |
Upgrades dependencies once more, no feature fixes.
The added networkbehaviour methods would allow for less branches in the src/lib.rs regarding listening addresses. Let's see if this'll build.
TODO: