RFC: agent instructions #42
Conversation
|
This pull request has been mentioned on Polkadot Forum. There might be relevant details there: https://forum.polkadot.network/t/xcm-rfc-agent-instructions/3819/1 |
| ``` | ||
|
|
||
| ### Implementation details | ||
| These three instruction allow for different implementations of proxying transactions via an |
There was a problem hiding this comment.
| These three instruction allow for different implementations of proxying transactions via an | |
| These three instruction allow for different implementations of proxying extrinsics via an |
| ## Summary | ||
|
|
||
| The RFC introduces the addition of agent instructions into XCM, specifically introducing three | ||
| new instructions: SetAgent, RemoveAgent, and TransactAsAgent. These instructions allow a |
There was a problem hiding this comment.
Throughout
| new instructions: SetAgent, RemoveAgent, and TransactAsAgent. These instructions allow a | |
| new instructions: `SetAgent`, `RemoveAgent`, and `TransactAsAgent`. These instructions allow a |
There was a problem hiding this comment.
There seems to be a general misuse of the term "transaction". This introduces confusion when discussing actual transactions (i.e. signed extrinsics submitted directly to a runtime) vs. transacting through means such as cross-chain messaging.
I think it would also be worthwhile to note, although implementation-dependent, that a single agent could be set with multiple permissions, just like you can designate an account as both a Staking and Governance proxy. The implementation of instructions like RemoveAgent will need to consider this, meaning whatever stores agents locally (e.g. Proxy pallet) will need a data structure that is friendly to finding all agents with a given MultiLocation. An alternative would be to change the signature of RemoveAgent to remove just one particular agent and/or add a RemoveAllAgents instruction that a chain could choose not to implement should its agent storage structure not be conducive to removing many.
Summary
The RFC introduces the addition of agent instructions into XCM, specifically introducing three
new instructions: SetAgent, RemoveAgent, and TransactAsAgent. These instructions empower a
location to designate an agent to operate on its behalf, enabling the agent to dispatch calls on behalf of the origin. This is particularly beneficial for local accounts, which can act as
agents for remote locations. This reduces the number of Transact instructions the remote
location needs to send. An agent is also able to dispatch calls via XCM using the
TransactAsAgent instruction.