Hans Larsen

> why not adding the did file as well Because we have _some_ control over our own canisters, we don't have control over users canisters. Pushing the problem to infrastructure...

> Textual representation is not optimized for writing. There's no better way right now. So let's add shorthands for record, tuples, vectors and optional and make it a good one.

> did file is an inherent part of Candid That's misleading. Candid itself does not require schemas, as you also encode the type in the serialization. If Candid wasn't built...

@rossberg > would essentially turn Candid into an untyped language I don't think so. If anything, I'm advocating for a properly typed integer, where `{0} That said, we could add...

> blob is vec nat8, do you expect it to be upgrade to vec nat64 at some point? `blob` being `vec nat8` is an implementation detail; from a typing standpoint...

> In this case, the type should be vec nat. Theory != reality. Not every developers are omniscient on the API and best practices. > suffix Nah this was Rust...

These are not in the public spec yet (they're internal APIs likely to change), so I'll mark this as blocked.

Thanks for the feedback. The builder itself already exists; https://github.com/dfinity/agent-rs/blob/next/ic-agent/src/agent/builder.rs#L4 (and is used in icx and dfx). We could have the `url` field in the Config type be `Option` and...

> why Agent::new and AgentConfig are public at all. Some historical reason, maybe a change in dfx or something. There was initially no builder, but now that there is a...

It seems we are using the Waiter to throttle. Are you using the proper Waiter instance? https://github.com/dfinity/agent-rs/blob/next/ic-agent/src/agent/mod.rs#L603-L605 ``` waiter .wait() .map_err(|_| AgentError::TimeoutWaitingForResponse())?; ```