[DEPENDENCY_REFACTOR] - Fix the ValidatedStateUpdated dependency in the quorum proposal and vote tasks #3322
Conversation
* fix * remove dead test * fix lint * fix test * Fix vote tests --------- Co-authored-by: Keyao Shen <[email protected]>
jparr721
force-pushed
the
keyao/fix-state-dependency
branch
from
613c95f
to
5d8f197
Compare
…t handling from quorum proposal
| @@ -175,6 +177,30 @@ impl<TYPES: NodeType> HandleDepOutput for ProposalDependencyHandle<TYPES> { | |||
|
|
|||
| #[allow(clippy::no_effect_underscore_binding)] | |||
| async fn handle_dep_result(self, res: Self::Output) { | |||
| let high_qc_view_number = self.consensus.read().await.high_qc().view_number; | |||
There was a problem hiding this comment.
Followup on an earlier discussion, in all testing, this value, and the provided high qc value were never different, so it seemed like a red herring. I can add the change in though that effectively takes the high qc from the event if it's available, and falls back to the shared state otherwise.
| consensus | ||
| .write() | ||
| .await | ||
| .update_saved_leaves(proposed_leaf.clone()); |
There was a problem hiding this comment.
Would it make sense to call update_saved_leaves from within update_validated_state_map? The only place where both of them are not called one after another is DaTaskState. It seems to me that they could (and maybe even should) be called together even there and then it means that they are coupled so tightly that we should make sure that they are always called together. What do you think?
There was a problem hiding this comment.
Yeah we definitely want to do this. I had originally done it, but the change was not great. I plan to do it as a follow up after evaluating if ViewInner::Da is still relevant, as it made the implementation quite tricky. I’ll tag you in the follow up.
Closes #3296. WIP!
This PR:
This PR does not:
Key places to review: