We should have a pool of connections which we use for queries, which will open up more concurrency. We should also have a single, separate connection used for updates (SqlStorage::transaction should use this connection). Since queries will no longer use the same connection as updates, there is no longer a risk of concurrent queries seeing a partially updated state. Thus, this is a step towards #400 .
Sqlx has connection pooling built in, so it may be best to wait and do this after #271
We should have a pool of connections which we use for queries, which will open up more concurrency. We should also have a single, separate connection used for updates (
SqlStorage::transactionshould use this connection). Since queries will no longer use the same connection as updates, there is no longer a risk of concurrent queries seeing a partially updated state. Thus, this is a step towards #400 .Sqlx has connection pooling built in, so it may be best to wait and do this after #271