chore: split db abstraction into new crate #8594
Conversation
onbjerg
added
C-debt
onbjerg
force-pushed
the
onbjerg/reth-db-api
branch
from
594e640
to
acd15ac
Compare
|
Removed dbi caching in 37fca74 since mdbx already does this and it would complicate doing this split if we wanted to keep it around (and complicate adding support for custom tables), see https://erthink.github.io/libmdbx/group__c__dbi.html#ga9bef4a9fdf27655e9343bbbf8b6fc5a1
|
onbjerg
requested review from
rkrasiuk,
rakita,
joshieDo,
shekhirin,
mattsse,
Rjected and
gakonst
as code owners
| @@ -1,5 +1,4 @@ | |||
| //! Storage sharded key | |||
|
|
|||
There was a problem hiding this comment.
its a conversation starter
| impl $crate::table::Table for $name { | ||
| const TABLE: Tables = Tables::$name; | ||
| impl reth_db_api::table::Table for $name { | ||
| const NAME: &'static str = table_names::$name; |
There was a problem hiding this comment.
Does this also address @prestwich's limitation of not being able to define custom tables in ExEx?
There was a problem hiding this comment.
not entirely, most of the code still assumes a fixed set of tables by using the generated Tables enum
There was a problem hiding this comment.
This LOC was one of the primary issues and is now resolved, but i haven't identified others yet
onbjerg
force-pushed
the
onbjerg/reth-db-api
branch
from
0f0fe5c
to
9762234
Compare
There was a problem hiding this comment.
one suggestion when doing these kinds of refactors, I try to avoid most dep changes on the initial PR with re-exports and then update them once the pr is merged because doing it like this can result of a lot of conflicts and makes the initial pr harder to review
Splits the db abstraction into a new crate
reth-db-api.Todo:
TABLEassociated type inTable(otherwise this PR is not possible)reth-db-api(e.g.impl_uints!(u8, u16, ...))Ref #8512