feat: validate_timestamp implementation

[TAdev0]

• issue Implement validate_timestamp function. #21
• follows contribution guide
• code change includes tests

This PR implements validate_timestamp functions, which checks that the proposed timestamp for the new block is strictly greater than the median of the previous 11 timestamps.

My implementation is simplified compared to ZeroSync

ZeroSync uses cairo zero hints to retrieve the last 11 timestamps by reading 11 contiguous memory cells in the VM, and then sorts the new array of 11 elements in order to find the median. I think this step can be skipped, it doesn't seem necessary to sort anything in our implementation as prev_timestamps in ChainState is necessary already sorted right?

also, what happens if prev_timestamps contains only a few values or even none (for example when syncing the very first blocks)? I'm assuming there are at least 6 values (11 - 5 = 6 to get the mean) , implementation will break if its not the case. But i guess for ZeroSync its the same? Trying to read 11 consecutives values in vm memory if there are less than 11 will fail with reading unknow value error

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
raito	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Aug 5, 2024 9:18am

[TAdev0]

yeah, ZeroSync uses this internal function:

// Compute the median value of an array of 11 timestamps
//
func compute_timestamps_median{range_check_ptr}(timestamp_array: felt*) -> felt {
    alloc_locals;
    // Compute the median using a hint
    local median;
    %{
        timestamps = []
        for i in range(ids.TIMESTAMP_COUNT):
            timestamps.append( memory[ids.timestamp_array + i] )
        timestamps.sort()
        ids.median = timestamps[ids.TIMESTAMP_MEDIAN_INDEX]
    %}
    verify_timestamps_median(timestamp_array, median);
    return median;
}

which implies that timestamp_array pointer points to an array of 11 elements

[maciejka]

Chain state will be verified by the zk proof so it makes sense to assume it is valid, i.e. timestamps are sorted and of valid length.
Just small stylistic issue.

[maciejka]

@TAdev0 please join project tg group: https://t.me/RaitoStarknet