On 10-8-2012 18:15, Roman Rokytskyy wrote:

> It is indeed tight optimization for single-byte charsets. However, if I
> remember corretly there is no penalty, since we falback to default Java
> mechanism. Also, I see no reason not to optimize for the charsets used
> in 95% or so installations of Firebird.
>
> As to whether it was tested against Charset - I do not remember, when
> the code was introduced, but I believe that was already 1.4.x times.

But at that time Java 1.3 was still supported so the mechanism used
would need to have worked on both versions (my assumption). In any case,
if I will first benchmark it before making a final change.

Do you know how it was tested at that time?

--
Mark Rotteveel