Gossamer team progress update 22.07.2025

[P1sar]

Initial Update

Cargo size

PRs

• Dedup dependencies between dependencies and dev-dependencies section PR has 2 approves but it’s not merged yet
• Remove unused dependencies found by using cargo udeps (PR merged) - We are going to be working on a script to parse the cargo udeps output and filter the false positives to add it as part of the CI
• Bump hickory-proto since the version 0.24.1 is flagged as vulnerable, this PR was closed in favor of another one from dependabot but it’s not merged yet so maybe we can help someway there
  Issues
• Fork dependency from untrusted source as part of paritytech. This is already addressed by Basti in this PR he found a better way to solve it without forking.

WebRTC

Have been mostly looking into existing PRs and Issues across smoldot and polkadot-sdk
It looks like most of WebRTC is NOT implemented on the Smoldot side. Looking in to possibly brining litep2p in to smoldot
Polkadot-sdk branches with some WebRTC are not compiling. Looking into that problem.
We define a list of issues on LiteP2P side that need to be handled, so we are going to start with them.

• WebRTC backpressure #45
• webrtc: Panics for non-stun packets #107
• Multistream-select implementation used by WebRTC doesn’t conform to spec #75
• Improve error handling in WebRTC-related Noise functions #43
• Add support for fallback protocols #42
• Implement new substream type for WebRTC #41
• webrtc: Improve robustness by avoiding panics #350
• webrtc: Optimize substream AsyncRead and AsyncWrite #351
• webrtc: Optimize decoding with appropriate buffer size #352

We are talking with Alexandru Vasile regarding all these questions.

Open questions: Would it make sense to bring Litep2p to Smoldot?

Speculative availability

• Experimenting with pushing candidate data to Prospective Parachains
• Need access to Parity logs and metrics platform or dumpdata to see in more detail where problems occur
• We are working on launching our own testnet + validator within existing testnets with proper CI/CD pipelines to test future changes. Plus, looking into Zombinets to reproduce problems and test changes.
• Started putting design together here

Approval rewards:

Prepared list of changes for RFC-119. After some peer review, we will propose these changes to the PR by the end of the week.
Looking into availability recovery & distribution as the off-chain part for Approvals Rewards. These subsystems are responsible for tracking uploads/downloads of chunks and
Started drafting changes for RFC here

Off-chain runtime

Not started

DoS protectio

Not startetd

[josepot]

It looks like most of WebRTC is NOT implemented on the Smoldot side.

What is it, exactly, that it is not implemented?

Looking in to possibly brining litep2p in to smoldot

Why?

[timwu20]

What is it, exactly, that it is not implemented?

I think Kirill was relaying an assumption I had made where I had initially thought Smoldot utilized rust libp2p implementation to establish webrtc-direct connections. After further investigation into Smoldot we determined we were incorrect in making that assumption.

Why?

I think for the same reasons why substrate migrated to litep2p. The most compelling reasons being the more efficient CPU usage and better data throughput at higher byte counts. After more research though, given how the network connections are abstracted away from the core smoldot libraries I don't think it makes much sense to migrate to litep2p given this could only be used when running light clients outside of the browser environment.