Support batching in 1.6 driver about tarantool-python HOT 4 CLOSED

would like to have batching support in the client.

from tarantool-python.

A dirty workaround might help for now. Here is a batch insert variant.

from itertools import islice

def load_tarantool(t, items, space_name):
    space_sid = t.schema.get_space(space_name).sid
    items = iter(items)
    chunk_size = 1000
    chunk = tuple(islice(items, chunk_size))
    n = 0
    while chunk:
        payload = b''.join(bytes(RequestInsert(t, space_sid, item)) for item in chunk)
        t._socket.sendall(payload)
        try:
            assert all(Response(t, t._read_response()) for _ in chunk)
        except SchemaReloadException as e:
            t.update_schema(e.schema_version)
            continue
        n += len(chunk)
        chunk = tuple(islice(items, chunk_size))
    print(f'inserted {n} items') 

from tarantool-python.

Yes, something to reduce round-trips from client to database would certainly be a great feature.

from tarantool-python.

There is no any native support for batching in basic IPROTO requests. Just sending a sequence of inserts isn't nice since some of them may fail in the middle and we won't have any possibility to handle it properly.

On the other hand, crud module has *_many methods which is essentially a batch insert/replace/upsert for sharded systems. These handles will be supported as part of the #205 issue. So I'm closing this one in favor of #205

from tarantool-python.