search

ChainAgnostic / namespaces

Chain-Agnostic Namespaces host informative specs and profiles of CAIPs (Chain-Agnostic Improvement Proposals) per blockchain ecosystem
https://namespaces.chainAgnostic.org/
53 stars 65 forks source link

CAIP-122 example doesn't conform to spec #66

Closed bumblefudge closed 1 year ago

bumblefudge commented 1 year ago

As @silverpill pointed out in another thread, the example in CAIP-122 lists an address in Solana-native format, NOT in CAIP-10 format as specified above in the spec. Shouldn't it line 116 read solana:5eykt4UsFv8P8NJdTREpY1vzqKqZKvdp:GwAF45zjfyGzUbd3i3hXxzGeuchzEZXwpRYHZM5912F1 instead of GwAF45zjfyGzUbd3i3hXxzGeuchzEZXwpRYHZM5912F1 ?

@haardikk21 @ukstv @zachferland someone who knows solana help me out here :D

haardikk21 commented 1 year ago

Good catch, definitely an oversight in the namespace spec.

The point of contention here is that SIWx was inspired by EIP-4361 - which does not require CAIP-10 in the address field, but CAIP-122 does.

Does this imply that the EIP-155 namespace spec for CAIP-122 will be divergent from EIP-4361? Or do we remove the CAIP-10 requirement from CAIP-122 entirely?

silverpill commented 1 year ago

From the user's perspective, plain address is better, because it looks familiar and usually matches account ID used by wallet. CAIP-10 identifiers are harder to read and understand. I think the most sensible approach is to use plain address when displaying a message, and use CAIP-10 identifiers internally

haardikk21 commented 1 year ago

From the user's perspective, plain address is better, because it looks familiar and usually matches account ID used by wallet. CAIP-10 identifiers are harder to read and understand. I think the most sensible approach is to use plain address when displaying a message, and use CAIP-10 identifiers internally

I think the question about how closely we want to tie EIP-155 CAIP-122 spec with EIP-4361 remains. Regardless of whether the address is displayed in the message or not, is a CAIP-10 URN mandated in the data model?

It matters because when verifying signatures, or validating SIWE objects, or creating CACAO (CAIP-74) objects - the data model values matter because they're not re-parsed from the message representation - unless, of course, all of those specs then include notes to basically slice the URN into only it's address part

bumblefudge commented 1 year ago

closed by #236