Re: gnome-db team

[Vivien Malerba <malerba linuxave net>]

On mar, 28 mar 2000, Rodrigo Moya wrote:
> Hi all!
> 
> We've successfully, after some months, completed the gnome-db team for
> the moment. Right now, we're in a position I wanted some time ago, that
> is, there is one person for each provider, so now I can go on holidays
> for a couple of years, until you reach the 1.0 version....
> 
> No, I'm not going anywhere without gnome-db, so now I want to
> concentrate on the rest of gnome-db = lots of bonobo, client libs,
> gda-builder....
> 
> So, here's the presentation so that you know each other. That's the
> team:
> 
> Stephan Heinze <stephan.heinze@xcom.de> - oracle
> Vivien Malerba <malerba@linuxave.net> - postgres
> Nick Gorham <nick@lurcher.org> - ODBC
> Chris Wiegand <cwiegand@urgentmail.com> - MySQL
> Cédric & another guy (sorry, don't remember your names/addresses) -
> GnomeDbReport

Nice to see so many people :)
 

> So, I'd like to start defining the config component stuff. The bonobo
> part is clear, isn't it? We'll code a sample bonobo control which will
> just include a container widget created by each of the providers' config
> component. My idea about what the component will do is that it should do
> whatever you want (and can). That is, the ODBC widget will be just a DSN
> configurator, as the windows version does, but the oracle provider can,
> for example, allow the creation of users, databases, ... management of
> tablespaces, quotas, etc, and any other provider whatever is possible
> with the underlying database. What do you think?

I think that there should actually be several bonobo components provided by
each provider, so that an application can choose which one to display. The
different components I see would be:
* one to allow configuration as defined above, but limited to configure
datasources
* one to allow the creation/destruction of databases
* one to allow the manipulation of users and their rights on tabes and databases
* one to allow the creation of tables and other objects more or less specific to
each provider (sequences,...)
* and other ones for tablespaces, quotas, etc.

What needs to be done is that a client must be able to know which components
are avaliable for a provider (maybe with the supports() function)

> 
> Also, I'd like all of the providers' people to say your opinion about
> the interface in the servers, the missing schemas (this is very
> important), .... because as we're starting to have several stable
> providers, I don't want to add more and more providers if the design
> does not fit any kind of data source.
> 

I will make a small modification to the TYPES schema to include the number of
arguments necessary to create a table (for the majority it will be 0, but for
example for data type varchar it will be 1, at least under Postgres). We should
also add a schema to know, for a given data type, what are the type of the
arguments necessary. [ Rodrigo, we talked about this at GUADEC ]
Have a llok at the doc in CVS to see the state of the SCHEMA stuff as I did in
Postgres (they can be modified if you need for your provider)

I'm at the moment busy defining the basis for generic XML queries, I'll submit
my thoughts to you probably next week.

Cheers
Vivien