Re: [DotGNU]RDF based discovery system

[James Michael DuPont]

--- Norbert Bollow <address@hidden> wrote:
> James Michael DuPont <address@hidden> wrote:
> 
> > The point is that a Discovery service for free software needs to be
> > thought through carefully.
> 
> I'd like to start my response by proposing definitions for the terms
> "discovery problem" and "offerings".
> 
> With "offering" I mean an XML description of something that is
> obtainable and potentially desirable, for example a commercial
> service, or a chunk of GPL'd code that is available for download,
> or some other kind of information.
> 
> The "discovery problem" is to find, among many offerings, one or more
> that will meet your needs.
> 
> I'm primarily interested in a discovery system that allows to
> locate business services (not limited to webservices).  If the
> same system also turns out to be useful for finding free software,
> great.  However, if a proposed system has design limitations that
> prevent it from doing everything that would be desirable in a
> "discovery service for free software", that should not prevent
> DotGNU from moving forward, as fast as reasonably possible, with
> busilding a system that is useful for discovery of business
> services.  DotGNU needs such a system in order to fulfil its
> mission of being a successful competitor to the .NET threat.
> 
> On the other hand, I'm getting the impression that you're primarily
> interested in a discovery system that allows to locate chunks of
> code that may be suitable for re-use in other Free Software
> projects.  That's a somewhat different type of discovery problem,
> not necessarily more difficult than the discovery problem for
> business services, but different and less well understood.  It would
> certainly be nice to have a good solution to this problem in DotGNU,
> but that's not among DotGNU's key objectives.
> 
> > As I said, it is my challenge to you and the other dotgnu people to
> > deal with the question of openness and informationsharing.
> 
> My response to this challenge is that DotGNU makes, as far as that
> is possible without this interfering with the project's primary
> objectives, all information and decision-making processes publicly
> visible.
> 
> However, my impression is that you're continually pushing for DotGNU
> to include issues of some kinds of _automation_ of "openness and
> informationsharing" among its primary objectives.  That is not going
> to happen, because that's not part of the original DotGNU vision and
> including it among the primary objectives would reduce our chances
> of attaining the original objectives.
> 
> It's totally ok when these issues are part of your personal agenda,
> and part of exprimental projects listed on the "Proposals" wiki page,
> but I wonder whether you'd be willing to stop trying to push them
> onto the list of core objectives of DotGNU, or into development
> projects where the maintainer has already said "no"?

Norbert,

You are right that I am talking about finding services, implemented as
parts of source code, from that source code. 

Lets say that I have a md5 digest, and want to find all the functions
that operate on that. That would be a valid case. 

Or lets say, I am looking for a function to to a FFT on an array of
data. This would also be something to be discovered.

The "no" seems to be more a knee jerk reaction than anything planned or
well defined. I dont understand exactly what the "No" applies to, I can
see that your application of discovery can be seen in many ways. 

The issue of finding business services is a valid issue, and the issues
of finding parts of sourcecode are similar to, but not the same. I can
accept that finding source code is not seen as the primary goal of the
discovery service. I guess that programs and functions could be
published as fine tuned services via this system.

The question of fine tuneness is a valid one, on the top level, 
the type of metadata from the DGMX files would be the first level of
the discovery system.

I am currently offline and on the road right now. I am now implementing
my patches to pnet to make it work the way that I want it to work. The
question of the Meta Object Description maybe more fined tuned than
what is needed for the discovery service.

mike



=====
James Michael DuPont
http://introspector.sourceforge.net/

__________________________________________________
Do you Yahoo!?
Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop!
http://platinum.yahoo.com