Add interop framework #26
Conversation
| - **Data vs Value Transfer** | ||
| - A data-transfer protocol allows information in general to be transmitted cross-chain. | ||
| - A value-transfer protocol allows cross-chain transmission of fungible assets. | ||
| - Cross-chain value transfer is a form of exchange. |
There was a problem hiding this comment.
Thinking about it, some might say that value transfer is actually a type of data transfer. What about defining data transfer as a "non-value" transfer or "generalized data transfer" and then mentioning that a value-transfer protocol is a "specialized" type of data transfer?
| - **Push vs Pull** | ||
| - A push protocol allows for the entire cross-chain transmission lifecycle, up to and including delivery to a receiver, to be arranged on the source chain without a user transaction on the destination chain. The protocol may rely on **third party relayers** to complete the lifecycle. | ||
| - A pull protocol requires a user (or application) to complete the lifecycle with a transaction on the destination chain. The protocol can be said to require **self-relaying**. |
There was a problem hiding this comment.
Genuinely curious about how to define push vs. pull in a way that accurately characterizes every existing protocol (I have proposed this one). As a general rule, I would say that a permissionless push-based protocol is actually a pull-based protocol. Third-party relayers could also be considered part of the users who seek to consume a message. In the other side, if the system itself—responsible for digesting and validating—is the only one that can execute the cross-chain operation, then we are entering push territory.
Let me know your thoughts
Documenting some of the design choices that have come up, and giving them names so we can discuss and decide.