At 03:59 PM 10/28/2004, Dmitry Yemanov wrote:

>"Ann W. Harrison"" <aharrison@...> wrote:
> > A more productive post-V2
> > approach might be to use the Vulcan internal SQL to execute
> > procedures directly, without the BLR intermediate layer.
>
>Yes, I was thinking about this too. And, generally, I like the idea. I will
>definitely support it as soon as someone explains how SPs are going to be
>serialized in the database.

Serialized?

> Or are we going to take the MS approach and make
>them interpretable rather than compilable?

I'd stick with compiled. The "compiled into BLR" is a misnomer -
it's really more a translation from a more-or-less human readable
language (SQL) into a language only a computer could love.

> Or do we require the source code
>to be available and recompile a SP on the first reference?

That's the model I'd expect to see.

>This will make lots of people (those hiding their sources
>in distributed databases) quite unhappy.

Right. And we'll need a solution for that - probably something
original like a special privilege that allows the procedure to
be read internally for compilation but not returned to the user.
Or encryption. What we've got at the moment is barely even
security by obscurity, given that ISQL will pretty print BLR...

Regards,


Ann