Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Revert changes to the PartSetHeader and BlockID #510

Merged
merged 3 commits into from
Aug 25, 2021

Conversation

evan-forbes
Copy link
Member

@evan-forbes evan-forbes commented Aug 23, 2021

@evan-forbes evan-forbes self-assigned this Aug 23, 2021
BlockID: blockID,
POLRound: polRound,
Timestamp: tmtime.Now(),
DAHeader: daH,
Copy link
Member

@liamsi liamsi Aug 23, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is orthogonal to this PR but:
Why does the DAHeader need to be part of the proposal again? Is that the case in the specs? cc @adlerjohn

It's certainly not used for sampling during consensus. And if the proposer should commit to the erasured data, isn't that given by the header hash (which includes the data root) instead?

Copy link
Member Author

@evan-forbes evan-forbes Aug 23, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I was wondering the same thing. I defo don't think we need it. Reverting that might be slighty more complicated, as there have been quite a few changes since we merged #248 . We might want to revert that later or "manually" (not using git revert).

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The specs should be changed as needed to conform to the new Tendermint-as-black-box approach, so it's not really relevant in this time to chaos what the specs say. Since it's no longer necessary for validators to sample data, and since validators will in the happy path download the entire block data via parts, there's not much need anymore to include the DAH in the proposal. Note that the unhappy fallback case involves validator and other full nodes attempting to reconstruct the block by sampling individual shares until recovery is possible.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK, but there is still some nuance on what the validators should sing during consensus, right?

Are you tracking the changes on the spec side for the tendermint-as-black-box approach? Or should we just build the thing first and update the specs later only?

Copy link
Member Author

@evan-forbes evan-forbes Aug 24, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since the the DAH won't be in the Proposal or in the block, then I'm assuming that each validator/full node will have to recreate the DAH using the block data?

If so, should we keep the DAH in BlockMeta, or should we save it separately?

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are you tracking the changes on the spec side for the tendermint-as-black-box approach? Or should we just build the thing first and update the specs later only?

Build the thing first, then I'll do a pass after my vacation and make sure the specs and impl are aligned.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since the the DAH won't be in the Proposal or in the block, then I'm assuming that each validator/full node will have to recreate the DAH using the block data?

Wait, isn't the DAH in the block, and gossiped as parts?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Wait, isn't the DAH in the block, and gossiped as parts?

Yes! but we're thinking of getting rid of that too I think. I can't find the comment in ADR 007, but it was removed from the list of minimal changes to tendermint because it was not strictly necessay. This reminds that we did not discuss that properly, and totally should, so I'll open an issue.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks Evan. IMO, there is no need to gossip the DAH as part of the block parts. The partsetheader already authenticates the Block and the receiving nodes can reconstruct the the DAH. But I might be missing sth here.

Copy link
Member

@liamsi liamsi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks Evan! This looks good to me and will make upgrading from upstream simpler.
I'm wondering if we should also remove the DAHeader from the proposal now (https://github.com/celestiaorg/celestia-core/pull/510/files#r694342674).

Copy link
Member

@liamsi liamsi left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

👏🏼 thanks @evan-forbes

@evan-forbes evan-forbes merged commit 3e542b4 into master Aug 25, 2021
@evan-forbes evan-forbes deleted the evan/revert-PSH-changes branch August 25, 2021 11:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Revert changes to the BlockID and PartSetHeader
3 participants