feat: implement 0x37 - CALLDATACOPY Opcode

[khaeljy]

Pull Request type

Please check the type of change your PR introduces:

• Bugfix
• Feature
• Code style update (formatting, renaming)
• Refactoring (no functional changes, no API changes)
• Build-related changes
• Documentation content changes
• Other (please describe):

What is the current behavior?

Resolves: #123

What is the new behavior?

Does this introduce a breaking change?

• Yes
• No

[Eikix]

Hey! Welcome, great first pr 🦾
Can we check if we can use slice instead of looping and copying?

[Eikix]

ah it's unlucky we can't assert equality for memory and felt dict, we should probably later create test utils for asserting equality @enitrat

[enitrat]

lgtm, will merge after thinking a bit about the type conversion thing :)

[github-actions]

Snapshot Comparison Report:

No changes in gas consumption.

[ClementWalter]

it looks non optimal to do two times memory.store_n because this function is quite expensive.

I think that it's cheaper to first pad data_to_copy, then to store_n only once, @enitrat

[enitrat]

@ClementWalter data_to_copy is a span and can't be padded, the solution would be to clone it into an array and then pad this array, but then I'm not sure it would actually be more effective. I will check and report back

[enitrat]

        if (copied.len() < size) {
            let n_zeroes = size - copied.len();
            self.memory.store_n(copied.pad_end_zeroes(n_zeroes).span(), dest_offset);
        } else {
            self.memory.store_n(copied, dest_offset);
        }

looks to be more optimized - I will re-verify once the requested changes have been applied

[enitrat]

looks to be more optimized - I will re-verify once the requested changes have been applied

Given that this implies cloning an entire memory segment just to pad zeroes, I think there might be a point at which it becomes more expensive to do a clone than it is to call twice store_n. Will research this later in #278

[khaeljy]

Tests will be fixed after merge PR #331

[enitrat]

Looks good now ! Waiting for bugfix PR merge to finish. Sorry it was a bit long but it was the first PR you opened, now it's going to be easier 💪

[enitrat]

💯

Enitrat approved the PR