feat(anvil): support millisecond mining with correct block.timestamp #8010
Conversation
|
Need to add tests. DO NOT MERGE |
|
@mattsse ptal |
There was a problem hiding this comment.
hmm, this does not really address because this only changes the tracked timestamp value and has no effect on interval mining.
I think the right fix for this issue, which is much simpler would be here:
foundry/crates/anvil/src/eth/miner.rs
Lines 157 to 158 in 487892d
and we could leave the setblocktimestamp interval alone, we can't even express sub 1s intervals anyway.
| @@ -1845,12 +1845,17 @@ impl EthApi { | |||
| Ok(true) | |||
| } | |||
|
|
|||
| /// Sets an interval for the block timestamp | |||
| /// Sets an interval for the block timestamp in milliseconds | |||
There was a problem hiding this comment.
this is incorrect and should be seconds
| @@ -581,7 +581,7 @@ pub enum EthRequest { | |||
| feature = "serde", | |||
| serde(rename = "anvil_setBlockTimestampInterval", with = "sequence") | |||
| )] | |||
| EvmSetBlockTimeStampInterval(u64), | |||
| EvmSetBlockTimeStampInterval(f64), | |||
There was a problem hiding this comment.
this is okay due to how json numbers work and the existing test should cover this as well.
| // Set interval in TimeManager as milliseconds. | ||
| backend.time.set_block_timestamp_interval((interval).as_millis().try_into().unwrap()); |
There was a problem hiding this comment.
why is this additional call now required?
| @@ -68,11 +81,11 @@ impl TimeManager { | |||
| self.add_offset(seconds as i128) | |||
| } | |||
|
|
|||
| /// Sets the exact timestamp to use in the next block | |||
| /// Sets the exact timestamp (`next_exact_timestamp`) to use in the next block | |||
| /// Fails if it's before (or at the same time) the last timestamp | |||
| @@ -100,30 +113,58 @@ impl TimeManager { | |||
| } | |||
| } | |||
|
|
|||
| /// Updates `wall_clock_timestamp` by `interval_ms` | |||
| pub fn update_wall_clock_timestamp_by_interval(&self, interval_ms: u64) { | |||
| let current_wall_timestamp = self.wall_clock_timestamp.read().unwrap(); | |||
| fn compute_next_timestamp(&self) -> (u64, Option<i128>) { | ||
| let current = duration_since_unix_epoch().as_secs() as i128; | ||
| fn compute_next_timestamp(&self, update_wall: bool) -> (u64, Option<i128>) { | ||
| let current = duration_since_unix_epoch().as_secs() as i128; // TODO(yash): Getting current time here as seconds. |
| // Sleep for 1s | ||
| tokio::time::sleep(Duration::from_secs(1)).await; |
|
Superseded by #9160. Turns out its a lot simpler than this approach. |
Motivation
Fixes #7931
Closes #9142
Solution
Adds
wall_clock_timestampinTimeManagerwith millisecond precision.This
wall_clock_timestampis then used as a reference in interval mining to update thenext_timestamp.Moreover, we now interpret
TimeManager.intervalas milliseconds instead of seconds. This brings breaking changes to the following anvil API method:evm_set_block_timestamp_interval.Note: We still interpret
--block-timefrom CLI as seconds. Hence, for millisecond mining user sets the--block-timeas0.05or0.75etc.