incremental hash

[damip]

Todo

• define the trail hash
• take into account the trail hash in Execution outputs
• use the trail hash for PRNG seeding in Execution
• write the trail hash to the final state (compatible with bootstrap, included in XOF hash)
• read the trail hash from the final state
• initialize the trail hash in the final state to Hash::zero() on genesis

Checklist

• document all added functions
• try in sandbox /simulation/labnet
• unit tests on the added/changed features
  • make tests compile
  • make tests pass
• add logs allowing easy debugging in case the changes caused problems
• if the API has changed, update the API specification

[Leo-Besancon]

Another thing to handle is the reset of the final state.
You should delete your prefix at two places:
FinalState::reset()
FinalState::new(), if the reset_final_state argument is set to true

You can call the .delete_prefix() function on the db to do that.

[damip]

Another thing to handle is the reset of the final state. You should delete your prefix at two places: FinalState::reset() FinalState::new(), if the reset_final_state argument is set to true

You can call the .delete_prefix() function on the db to do that.

Improvement: remove the new(reset_final_state) and reset() concepts completely through #3728 => final_state is initialized from a db, and when re-attempting a bootstrap, the final_state object is just re-created.

[damip]

@Leo-Besancon the bootstrap test fails. Any idea ?

[Leo-Besancon]

@Leo-Besancon the bootstrap test fails. Any idea ?

I'd check if the helper function that inits the new hash key is called by the server. Otherwise, it's never created and the is_db_valid of the client fails.

Edit: it seems to be the case: 'Client's DB is not valid after bootstrap'!

[Leo-Besancon]

Looks like everything is taken care of.
Sidenote: We must not forget to init the execution trail hash in the db if we want to convert from a previous version (e.g. if we want to keep the testnet24 db for testnet25, or when converting the buildnet ledger)

[damip]

Looks like everything is taken care of. Sidenote: We must not forget to init the execution trail hash in the db if we want to convert from a previous version (e.g. if we want to keep the testnet24 db for testnet25, or when converting the buildnet ledger)

Yes but if we unify things as we expect to do, restarting from an existing state will preserve the previous hash which is good

[damip]

thanks @Leo-Besancon for the review and corrections !