Checking program bytecode when submitting it for storage

[ameba23]

Currently it is possible to store a program (using the programs pallet's set_program) even if it's bytecode is invalid, or it has a configuration interface which is not compatible with the program. When registering there is also no check that the provided program pointer points to valid bytecode.

This is maybe rather an issue for the programs repo - as that should provide us with a way of checking validity of bytecode etc. But relevant here is where that would get called from in entropy-core.

[JesseAbram]

so ya I think this should look like something like

fn validate_program_input(bytecode, messge, aux_data,  config_data) -> bool {
    // simulate return bool
}

// same idea here 
fn validate_hashing_algo

and it can live in two places. One would be core where you send the data over to a TSS and have it send back the result. However I think that is a worse option then having this compile down to wasm and run it client side (wasm for in the browser could be rust for program devs, however, I do believe there is already a way for program devs to test this in the programs repo, by compiling it to wasm and running their tests against this)

TLDR: I think in the programs repo, compiled to wasm, pushed to npm, pulled in by the sdk is the move

[ameba23]

doing it on the client side seems good - as long as there isn't a reason why someone would want to intentionally publish some garbage programs. i can't think why they would want to, but just putting the idea out there.

[JesseAbram]

doing it on the client side seems good - as long as there isn't a reason why someone would want to intentionally publish some garbage programs. i can't think why they would want to, but just putting the idea out there.

I mean possible but the idea is to charge for state rent so that isn't an attack vector

[mixmix]

This seems related to the "Dry Run" conversation @ameba23 . Maybe we could group all those related issue into a "Milestone"

Aside: when we were discussing "dry run" we were meaning "check content is valid/ functional before committing things to chain". When I said "dry run" to someone on my team they thought about --dry-run in some tools where the system mocks the actions and shows output, but doesn't actually mutate anything. => perhaps we need a different name ... e.g. "pre-flight-checks" ?

[ameba23]

Linking to the associated issue in the programs repo: entropyxyz/programs#95