Ismail Khoffi
Ismail Khoffi
Can we document the decision here int this repo as well as a light encoding spec for app devs and clients? The spec served as a previous guideline of implementing...
I've opened a separate [issue](https://github.com/celestiaorg/celestia-core/issues/759) for the above so we can discuss the length delimiter here only. @tzdybal brought up some good points above. If we use bytes not number...
If we use bytes, I would still vote for varint encoding!
OK, after thinking about this further, I do think varint encoding is not so bad after all: it will always work (up to 10 bytes for an unsigned integer ......
ref: https://github.com/lazyledger/lazyledger-core/pull/351
I agree with you @tomtau. Also, @marbar3778 is right that the proto files (and hence the generated code) should match the tendermint repo. Actually discussing signed vs unsinged integers has...
I'd recommend not prioritizing this. It might be good to make this a post-launch concern even.
As all this sounds good but it's mostly about optimizations, I've attached this to Testnet. We could even do some of this after mainnet though. > AFAIK we don't have...
rsmt2d is a bit different with this regard, as we are currently refactoring it to streamline the APIs and make it production-ready. So if you have good ideas there, you...
> But measuring differences would require us to implement that first, so it hard to understand the exact numbers of gains without doing actual work here. Yeah, you are right...