Re: [dtp-dev] Question Concerning Statically vs. Dynamically Defined DB

[rcernich@xxxxxxxxxx]

Hey Larry,

> For informational purposes, this is how version information is handled in
WTP:
>
> There is an extension point (org.eclipse.wst.rdb.core.
> databaseRecognition).  This extension point can be used to provide a
> class that returns a boolean indicating whether the version provided
> in the DatabaseDefinition is the same as the version on the server.
> The connection wizard uses this facility to determine if the
> versions are the same, if they are not, then a warning dialog is
> displayed and the user can determine whether or not they want to
> connect anyway.   The user is given the option to connect anyway, so
> that if a new version or service pack for the database is released
> that the databaseRecognizer does not recognize, the user can still
connect.
>
> ( Note that there are no databaseRecognizers implemented in WTP so
> the warning dialog is never displayed in WTP.)
>
>
> I would argue against your comment in the defect ("
> I would add that if the vendor/version is not defined, the
> ConnectionInfoImpl should fallback to use the generic JDBC
> dbdefinition and log this to both syserr (so the developer notices
> this in his debug env) and the Eclipse log.
> ")  as this will block the user from utilizing vendor specific
> features such as custom catalog loaders and databaseDefinitions. The
> user needs to be allowed to decide which databaseDefinition to use
> in this case.
>

I understand your concern.  However, I don't think it is all that bad.  If
the vendor/version is not defined or does not match anything registered
with the framework, the user cannot use the connection at all.  I think it
would be better to have the connection be usable, even in a limited
capacity, than not usable at all.  That said, I think it should be amended
to use the databaseRecognizer extension prior to moving along to the
generic support.

Sound reasonable?

Rob