The size is not negotiated at the start of the connection, it could
be different in each GDS call but I think that it is very
difficult to find an optimal value without parsing the statement
at the client side.

IMHO, Firebird server would have to use a dynamic buffer size
once it has parsed the statement and not wait a buffer size to
come from the client side.

--- In IB-Java@y..., "patcom123" <lister-filter@p...> wrote:
> Just made an observation (which is probably wrong) regarding the
> buffer size, it seems that the size is variable (at least across
> different compilations) and i was wondering if it would be possible
> to guess/approximate a more optimal value for it before tx/rxing data
> to the server?
>
> I guess that as it hasn't been done before, the size is *negotiated*
> at the start of the connection and remains fixed?
>
> Just a thought...
>
> JOHN