Thanks for the tip. There is a difference between bulk and multiple statements, however. Bulk insert operates with supplied container(s) as parameter array(s) passed to the driver. From there, it entirely depends what driver does with it – some turn it into multiple statements, some (e.g. Oracle) do one statement and pass the array(s) to the db.

Since we currently treat SQL code as a “black box”, your suggestion is only applicable at the user level where user can choose whether to pass SQL string as multiple statements separated by ‘;’, or only one statement in the way you suggest.

Alex

insert foo select 1, ‘a’
union select 2, ‘b’
union select 3, ‘c’

It works quite well where I’ve tried it – especially it seems on SQLServer and Sybase where triggers run per-statement, not per-row.

James

std::deque and std::list are now suppoerted for bulk operations alongside std::vector (when it comes to input binding, std::vector offers better performance for POD types, since there is no internal caching and copying).

And then, internal extraction can now do bulk as well (for the drivers willing to cooperate).

We need testers, especially on non-windows platforms! Please check out the trunk and give it a try.

You can check documentation for unixODBC or iODBC and FreeTDS. Link to Poco::Data manual is in the post above (bear in mind that manual reflects features available in SVN).

Alex