Use RETD for return values larger than a word.

[sezna]

Return the result from the main function, or the public ABI functions in the case of contracts, using the return opcode.

If the value is not held in one register, i.e. if it is a reference to some stack or heap memory, do we need to copy that memory somewhere? cc @adlerjohn

[adlerjohn]

CALL specifies a list of ranges that can be written to for return values. That's how values are supposed to be returned from a cross-contract call. The register in RETURN is supposed to be used for an error code from an external context mostly.

[sezna]

In script+input data mode, are the specified ranges given as part of the initial stack frame?

[adlerjohn]

Ah, so if it's a predicate context, the return value is true or false (for a well-behaved predicate). Anything not true is considered false for the purposes of predicate validity.

In a script context, the value is simply returned and can be used by applications as an actual return value or an error code. If you want to return more than a single register worth of data, the plan is to have a new opcode (e.g. return data RETD) that can only be used in a script context and returns a single memory range. The returned value can be committed to in the receipts root.

[sezna]

Ok. So, until RETD is introduced, I will just return a register? And if it is a pointer or something that doesn't fit, then that's just unfortunate, until we can return data.

[adlerjohn]

Yep, but only from a script context. Internal (i.e. contract) context return data using the writable ranges given with CALL.

[sezna]

RET is implemented, but we need to use RETD for data larger than a word.