On 11 May 2006 at 9:05, Dmitry Yemanov wrote:

>Y-valve would be required to always parse SPB as otherwise it cannot know
>what it should to - forward the request to the engine or handle it itself.

Yes, but only if service must be done locally ("service" param
doesn't include network path).
May be we should develop some machinery for "partial handling" of
SPB. In this case Y-valve parser may be the "last resort" handler.

>> Because utility are ODS-bound. And so do engines.
>
>Particularly true. fb_lock_print is not ODS bound, neither is gfix
>-shut/sweep/etc.

gfix -shut call the engine, AFAIU.

>Which engine will open the database being backed up depends on ODS. But GBAK
>will complete successfully anyway, producing a most recent backup version.

Really? You can run FB2 gbak against IB5 server?

>If you think that we should be able to produce different backup versions
>within a single server installation, then what should happen with standalone
>GBAK utility? Do you suggest to ship multiple GBAKs too?

No. As I wrote a long time ago, gbak must become a stupid utility
that just receive binary stream from engine and store it in file(s).

--
SY, Dimitry Sibiryakov.