"Helen Borrie" <helebor@...> wrote:

>
> >I think this idea is not good enough. To obtain a database info you
> >must be connected to a database first. Personally, I'd like to be
> >able to get engine capabilities before connecting or creating
> >database.
>
> Agreed - Jim, do you have some wires crossed here? Your suggestion was
> aimed at a mechanism to report the attributes of the database engine,

i.e.,

> presumably a service call of some sort; but then you proceed to suggest
> that it could be a *database* info structure. Why would you write the
> engine capabilities into databases?

First, we can attach the server without opening any database. This is done
via the Services API.

Second, did anyone look at isc_info_svc_capabilities before starting this
discussion? It doesn't work exactly as we want now, but it implements the
same idea.

I'd prefer to get the server/engine capabilities via the Services API and
the local capabilities - via isc_database_info() call. This sounds logical
to me. I know that isc_database_info() was designed to return the
server/engine info as well, but there was no server-level API that time. Now
we have a choice.

But it has serious problems with Jim's idea of making the Services API a
separate provider which is completely independent from the engine. So I
think the future of the Services API should be discussed first.


Dmitry