dpc
I'm investigating a similiar problem with ``` #[derive(Clone, Debug, Serialize, Deserialize, Encodable, Decodable)] pub struct MintConfigConsensus { /// The set of public keys for blind-signing all peers and note ///...
Trying to bench it, but can't: https://github.com/fedimint/fedimint/pull/7517
Status: we suspect this is some weird wasm miscompilation / optimization problem, and refactoring it is a big gnarly, so this is on hold.
@maan2003 Is this still an issue?
Ah, yes, I keep mixing these up.
@elsirion Informed by my mistakes in #7620 it seems to me that validating this crypto at decoding level is the right choice by default, as tracking if something was validated...
IMO, this will just rot in a backlog, and is a lot of engineering effort and complexity for increasingly lower benefits.
I like that `Params` are zero-copy, and I haven't noticed usability issues with the lifetimes. In my toy astra/mathcing project I barely every write down the lifetimes: https://github.com/dpc/htmx-sorta/blob/5e9d2fe026ad9e2baae20368e59c7237c0671161/src/routes.rs#L64 The only...
> It's when you use it in a complicated web framework abstraction the lifetime pollution becomes apparent. This is how matchit is used in axum: https://github.com/tokio-rs/axum/blob/368c3ee08fc3896358d3bd2bfc8cc67f2925c6ef/axum/src/routing/url_params.rs#L24 That doesn't look bad....
BTW. Probably good idea to make this controlled via env var (optionally), so in scritpts and for backward compat it can be set once for all invocations.