test: re-enable the simulation test
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: Eric Torreborre <[email protected]>
Signed-off-by: jeluard <[email protected]>
Signed-off-by: jeluard <[email protected]>
Signed-off-by: Roland Kuhn <[email protected]>
Signed-off-by: Roland Kuhn <[email protected]>
Signed-off-by: Roland Kuhn <[email protected]>
Signed-off-by: Roland Kuhn <[email protected]>
also complain when the exact rollback point is not found, instead of silently moving to a different point's state also complain when trying to roll back to a point before the tip Signed-off-by: Roland Kuhn <[email protected]>
Signed-off-by: jeluard <[email protected]>
Signed-off-by: jeluard <[email protected]>
feat: make sure mithril package files are stable
Signed-off-by: jeluard <[email protected]>
Signed-off-by: jeluard <[email protected]>
Signed-off-by: Roland Kuhn <[email protected]>
This is an attempt at simplifying our snapshot-production process to be leaner and involve less tools. The current solutions requires a running node, Ogmios attached to it, and a script monitoring the whole. Plus, the data must be extracted while the node is syncing, which is not only unpractical to do repeatedly, but also quite brittle as it can sometimes fails due to concurrency issues (especially on mainnet, when building some of that data takes time). Plus, adding intermediaries like Ogmios in the loop, even if written in Haskell using the Cardano libraries, adds room for mistakes and discrepancies. There are already various transformations that are required on top of the raw ledger data, so it is better to reduce the surface and operate directly on the base data. Using the ledger state queries over the wire is unnecessary. The current commit only replaces the 'fetch.mjs' script, although the end goal is to produce the test vectors we currently produce using the 'generate.mjs' script (which combines data fetched from multiple epochs). As for the input snapshots, they can now be obtained from a working database using the db-analyzer (from ouroboros-consensus). Although this can take several hours on mainnet, it is easier to instrument than manually removing blocks from the immutable db and making queries while the node is syncing. Signed-off-by: KtorZ <[email protected]>