Enum WriteOp 

pub enum WriteOp<'a> {
    Single {
        query: &'a str,
        params: Vec<DbValue>,
        mode: FetchMode,
    },
    BatchDDL {
        queries: &'a [&'a str],
    },
    BatchParams {
        query: &'a str,
        params_list: Vec<Vec<DbValue>>,
        isolate_rows: bool,
    },
}

Unified write operation types.

Variants

Single

Single query with optional return.

Fields

query: &'a str

params: Vec<DbValue>

mode: FetchMode

BatchDDL

Batch of DDL statements executed in a single transaction.

Fields

queries: &'a [&'a str]

BatchParams

Same query executed once per parameter set, in a single transaction.

Use for batched INSERT … ON CONFLICT, UPDATEs, or any non-Postgres backend. For a plain high-volume insert into one table, PgHandler::copy_in is ~30–50× faster — see its docs for a full copy_in-vs-BatchParams decision guide.

Fields

query: &'a str

params_list: Vec<Vec<DbValue>>

isolate_rows: bool

Per-row error isolation.

• true — a bad row is contained and the rest of the batch still commits. Both PgHandler and the multi-backend BaseHandler wrap each row in a SAVEPOINT (standard SQL: Postgres, MySQL/InnoDB, SQLite), so a failed row rolls back alone instead of aborting the transaction.
• false — all-or-nothing: no per-row savepoints, so the first error rolls back the whole batch. ~2× faster than the isolated path. Use for trusted bulk inserts where partial success isn’t needed. For the fastest plain bulk load, prefer PgHandler::copy_in.