state proofs: add block hash to LightBlockHeader #5663
Conversation
Replace the `Seed` value in the state proof light header with `BlockHash`. It serves a similar purpose in terms of the second-order consideration of defending against potential future hash-collision attacks (albeit a little bit weaker), but has the benefit of allowing validating the entire blockchain, by committing to the block hash itself in the state proof.
cce
changed the title
cce
changed the title
|
TestStateProofMessageCommitmentVerification is failing, which is an E2E test that looks like it spins up using vFuture |
The test set a configurable consensus protocol, "test-fast-stateproofs", and propagated the configuration to the child algod processes spawned by the test, but did not update the supported consensus params in the test process itself. As a result, when the test code tried to compute the light block header, it didn't have the block protocol (test-fast-stateproofs) in its config.Consensus.
|
Looks like a slight oversight in |
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## master #5663 +/- ##
==========================================
- Coverage 55.81% 55.79% -0.03%
==========================================
Files 476 476
Lines 67138 67146 +8
==========================================
- Hits 37476 37464 -12
- Misses 27141 27152 +11
- Partials 2521 2530 +9 ☔ View full report in Codecov by Sentry. |
|
Corresponding go-algorand-sdk change: algorand/go-algorand-sdk#589 |
do the other sdks need updates as well? |
Probably so, yes. Nobody is likely to be using the light block headers (which gained an extra field), but it might be that some users will be looking at the protocol config values (which also gained an extra field). What have we done in the past when adding new protocol configuration parameters to |
Replace the
Seedvalue in the state proof light header withBlockHash. It serves a similar purpose in terms of the second-order consideration of defending against potential future hash-collision attacks (albeit a little bit weaker), but has the benefit of allowing validating the entire blockchain, by committing to the block hash itself in the state proof.