[WIP] Family Specific Transmitters #16704
Conversation
…P contract transmitter
…tly supporting SVM and EVM
0xsuryansh
requested review from
winder,
asoliman92,
matYang,
makramkd,
dimkouv,
0xAustinWang and
0xnogo
AER Report: CI Core ran successfully ✅AER Report: Operator UI CI ran successfully ✅ |
0xsuryansh
changed the title
… into CCIP-5276_family_specific_transmitters # Conflicts: # core/capabilities/ccip/ocrimpls/contract_transmitter.go # core/capabilities/ccip/oraclecreator/plugin.go
0xsuryansh
changed the title
|
| // ContractTransmitterFactory creates commit/execute transmitters for a specific chain family. | ||
| type ContractTransmitterFactory interface { |
There was a problem hiding this comment.
Feels like it would make more sense to define this in ocrimpls
There was a problem hiding this comment.
My thought is that this is a factory interface for different transmitters , the implementation of which exist in the impl folder ?
| type ExtraDataDecoded struct { | ||
| ExtraArgsDecoded map[string]any | ||
| DestExecDataDecoded []map[string]any | ||
| } |
There was a problem hiding this comment.
Can we have doc comments on public methods, members, and types that are introduced?
| offrampAddress string, | ||
| defaultMethod, priceOnlyMethod string, | ||
| ) ocr3types.ContractTransmitter[[]byte] { | ||
| return &ccipTransmitter{ |
There was a problem hiding this comment.
Why are we using the same exact object for all these methods? The point of the separate transmitter per family is to literally have a different implementation. I don't think this PR really improves much upon the baseline.
There was a problem hiding this comment.
The design intentionally uses the same underlying transmitter type, but the key is that its behavior is driven by the injected toCalldataFn. Each factory provides its own implementation of this function, which tailors the calldata encoding and ultimately the transaction submission logic to the specific chain (EVM or Solana).
While the outer object is the same for common functionality the chain-specific logic lives in the callback.
There was a problem hiding this comment.
I don't like this approach at all and I was the one that (unfortunately) introduced it.
0xsuryansh
changed the title
Created a
ContractTransmitterFactoryinterface with two functionsNewCommitTransmitterNewExecTransmitterThis interface has two implementation
EVMContractTransmitterFactoryandSVMContractTransmitterFactorywith chain family specificToCalldataFunc