Re: Resources and state

Yes! I think this is very important to keep in mind:

James Cheney wrote:
> *However*, these are important research issues for provenance, and there 
> is little, if any standardization on how to do this, in contrast to the 
> various provenance models (OPM, PML, Provenir, ...) we have been 
> discussing which seem perfectly adequate for static, non-aggregate 
> resources.
> I recall that at the end of the WG, issues such as containers, 
> versioning, recipe links and so on were raised and included among the 
> concepts, but that there seemed not to be a strong consensus that they 
> were essential or that technology for these has converged enough to 
> justify standardization.  I don't think the goal of the WG is to try to 
> solve (what I view as) research issues.
> So I would be inclined to to agree that we should avoid mission creep 
> concerning resource state, and maybe go further: while recognizing the 
> importance of dynamic state, versioning, containers, etc. for 
> provenance, we should scope the model as narrowly as possible, and ask 
> for each concept whether it is really something that needs to be 
> standardized in order to provide minimal , and whether it is 
> well-understood enough not to be a research problem.  For some of these, 
> we might consider extensibility mechanisms (eg OPM-syle "profiles") to 
> accommodate experimentation without presuming to standardize something 
> prematurely.

So my view is that we don't try to solve the hard research problems, but focus 
on areas where consensus already exists.  But, I also don't think we should 
arbitrarily limit what we propose to the well understood cases, if we don't need 
to, because defining such limits may itself be a hard research problem.


Received on Thursday, 2 June 2011 14:46:39 UTC