You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It is also informally known as: ocaml-kc (for Kyoto Cabinet - an inspiration for the implementation) and kv-hash (i.e., a key-value store based on a persistent hashtable approach) and possibly index-v2.
If we replace the index in this repo with the kv-hash alternative, then a lot of the current issues disappear. In particular, the "blocking merge" problem is not observed with kv-hash during irmin tree.exe bench replay of Tezos traces.
I'm opening this issue mainly so that there is a record that a lot of the other issues may be redundant. As an example, here are some issues that might be redundant if we move to index-v2:
A possible replacement for Index is here: https://github.com/tomjridge/kv-hash
It is also informally known as: ocaml-kc (for Kyoto Cabinet - an inspiration for the implementation) and kv-hash (i.e., a key-value store based on a persistent hashtable approach) and possibly index-v2.
If we replace the index in this repo with the kv-hash alternative, then a lot of the current issues disappear. In particular, the "blocking merge" problem is not observed with kv-hash during irmin tree.exe bench replay of Tezos traces.
I'm opening this issue mainly so that there is a record that a lot of the other issues may be redundant. As an example, here are some issues that might be redundant if we move to index-v2:
io_array'sget_entry_from_ioshouldn't be called #368find_instanceblocks for up to 2 min #323Index.try_mergedoesn't check for an ongoing merge #311The text was updated successfully, but these errors were encountered: