light-clients: update revision number for cf-guest and cf-solana to 1

[dhruvja]

This is to increase compatibility with other libraries which fail to
properly decode proto3 message with missing revision number field
(missing fields should be decoded as zero in proto3).

[mina86]

I don’t follow, why do we need this?

[dhruvja]

well we used revision number as 1 everywhere since ibc-go used to omit it if it is 0. In our case it doesnt really matter what we use a revision number but i think, using 1 everywhere seems like a good option to maintain backward compatibility.

[mina86]

Right, but my point is why are we changing the revision number? Why can’t it be zero?

[dhruvja]

if we make it to 0, we wouldnt be able to establish a connection with cosmos since it omits the field if the value is 0.

[mina86]

Yes, that’s how protobuf works. In proto3 zero fields are not encoded but then when decoding zero is implied. So Cosmos should decode it as zero just fine.

[dhruvja]

but apparently it doesnt, i am not sure why. When we were having revision height as 0, we were getting missing field issue on ibc-go.

[mina86]

OK, though now the rollout is potentially tricky, no? This needs to be updated in the solana-ibc and on Picasso at the same time, no?

[dhruvja]

well no, since picasso uses cf-guest from composable-ibc.

[mina86]

But that uses this cf-guest as dependency, no?

[dhruvja]

the cf-guest light client on picasso doesnt implement ClientStateExecution trait so we wont have an issue over there.

[mina86]

LGTM