Struct WriteOptions 

pub struct WriteOptions {
    pub mode: WriteMode,
    pub batch_size: usize,
    pub key_columns: Vec<String>,
    pub bulk_mode: BulkMode,
    pub copy_format: CopyFormat,
    pub atomic: bool,
    pub isolate_failures: bool,
    pub verbose: bool,
}

Configuration for a batched write.

Fields

mode: WriteMode

Conflict semantics. Default WriteMode::Insert.

batch_size: usize

Rows per batch. 0 is treated as DEFAULT_WRITE_BATCH.

key_columns: Vec<String>

Conflict-key columns for WriteMode::Skip / WriteMode::Upsert. Must be a subset of columns. Empty is rejected for those modes.

bulk_mode: BulkMode

Route batches through the destination’s native bulk loader (BulkMode::Auto / BulkMode::On). Ignored for the conflict modes, whose bulk loaders carry no MERGE semantics.

copy_format: CopyFormat

Postgres COPY wire format for the bulk path. Other backends ignore it.

atomic: bool

Wrap the whole write in one outer transaction: all batches commit together or roll back together. Uses the same per-backend BEGIN/COMMIT/ROLLBACK as migrate’s atomic apply — MSSQL adds SET XACT_ABORT ON; Oracle gets an explicit COMMIT. A batch failure under atomic rolls the whole write back and the report records the failing batch.

isolate_failures: bool

On a batch failure, retry that batch one row at a time to pinpoint the rejected rows (recorded as RejectedRows) and let the good rows through. Off by default — the cheaper per-batch granularity rejects the whole failing batch. Mutually weakened by atomic: under one outer transaction a row probe cannot commit partial good rows, so isolate_failures only refines the diagnosis, not the commit boundary.

verbose: bool

Emit per-batch diagnostics on stderr (bulk path selection / fallback). Mirrors the CLI --verbose flag.

Trait Implementations

impl Debug for WriteOptions

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for WriteOptions

fn default() -> Self

Returns the “default value” for a type.