I also consider the areas Brian mentions to be important. I might be
able to contribute on #3 (having built similar repository technology in
the past). It would be great if MDR were rearchitected to make it
possible to plug in off-the-shelf O-R mapping technology such as JDO or
Hibernate. When I implemented JDBC persistence for MDR, the generated
schemas came out quite ugly, and there's no way to configure them or add
O-R mapping since (as Brian says) the SPI is defined at the index level
rather than the class level.
License neutrality would also be a big improvement (GPL-compatibility
would be nice but it's probably way too much to ask).
In the past, I have gotten the impression that the reason that MDR
hasn't grown into its full potential as a multi-user repository is that
Sun's focus has been primarily on using it to enhance Netbeans. That's
only fair, since Sun has been the one putting in most of the resources.
However, if this is going to continue to be the case, perhaps others who
would like to see an "MDR-NG" would be interested in a fork or a brand
new project? Would Sun be supportive of such a development?
Aside: as part of The Eigenbase Project, we have ported an existing
metadata query language to work against MDR/JMI. Details here:
http://farrago.sourceforge.net/lurql.htmlWe're still working on the packaging to make LURQL easy to download and
play with (LURQL is currently included as a component in the full
Eigenbase distribution), but the doc may give some idea of what it's
capable of. This is something we'd like to find a way to contribute
eventually (licensing is currently GPL).
JVS
On Sun, 2005-12-04 at 19:42 -0800, John Sichi wrote:
>
> --- Brian Topping <
topping@...> wrote:
>
> > CC:
users@...
> > From: Brian Topping <
topping@...>
> > Date: Sun, 4 Dec 2005 11:22:10 -0800
> > To:
dev@...
> > Subject: Re: [mdr-dev] MDR, MOF2, JMI2, ...
> >
> >
> > On Dec 4, 2005, at 10:09 AM, Petr Hrebejk wrote:
> >
> > > I have a few questions for all of you who use the MDR. Is there a
> > > possibility that the community around MDR will cooperate on the
> > > development? How much work are MDR users prepared to invest into
> > MDR?
> >
> > My interest is proportional to the direction that the next revision
> > goes. In order that this work could serve as long as the existing
> > MDR has, I would like to see it rewritten from scratch with the
> > latest in a few areas:
> >
> > 1) Code generation (if it is still deemed necessary) is done with a
> > toolkit like
http://asm.objectweb.org/ instead of straight bytecode
> > manipulation.
> >
> > 2) Component management is handled by an open source IOC container
> > instead of lookup routines from the Netbeans runtime.
> >
> > 3) Repository access is done through something like JDO instead of
> > internal DBM-style management. This one is a bit more complex, but
> > JDO spec includes a flat-file backing store as well as SQL, and it
> > should get MDR much closer to being runtime configurable for robust
> > storage on SQL.
> >
> > I also find that it's very important that the license it is
> > distributed under allows binaries to be posted to locations such as
> > the Maven repository at
http://ibiblio.org/maven. Having to log in,
> >
> > jump through hoops to download the latest, then extract the important
> >
> > jars from the results is an unnecessary barrier to entry and source
> > of additional costs for users.
> >
> > Other than those minor details, I'm ready to get started right away!
> > :-)
> >
> > -b
>
>
>
>
> __________________________________________
> Yahoo! DSL ? Something to write home about.
> Just $16.99/mo. or less.
> dsl.yahoo.com
