Replies: 1 comment 13 replies
-
Okay. So, it would be feasible to instrument the internal chain consumer to also synchronize from a given point when changing patterns. What isn't possible (at least in the current design à without a major overhaul) is to have synchronization going on at multiple levels on-chain. Said differently, in order to synchronize something from the past, Kupo will have to rollback the entire database from that point on. The synchronization isn't per pattern, there's only one synchronisation thread. |
Beta Was this translation helpful? Give feedback.
13 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Here's my narration of how I think kupo works. Please correct me where its apparent I don't know what I'm talking about.
On start, kupo looks at all blocks since the
--since
arg ( defaulting to origin) for matches on--match
args.Kupo reads blocks sequentially (caveat: rollbacks) in a single sequence.
On restarting, kupo will go the last checkpoint (?) and continue syncing.
Kupo allows adding and deleting patterns while running.
However, matches will only be indexed for a given pattern in a given block if the pattern is added before the block has been read by kupo.
If there is some delay in learning of a new "interesting" address, and kupo has already read this block there is no way to add this address and get a complete utxo set for it.
How bad would it be to have kupo be able to handle multiple block readings?
Beta Was this translation helpful? Give feedback.
All reactions