"RegTAP" broken out and updated

[Markus Demleitner]

Dear Registry community,

As promised in São Paulo, I've now broken out the stuff pertaining to
the "relational registry" (which I've nicknamed RegTAP in some
places) from the draft RegistryInterfaces2 document.  The latter
remains at 

https://volute.googlecode.com/svn/trunk/projects/registry/RegistryInterface

The "RI2" draft is now lacking anything on a searchable registry
except a note that RegTAP is now somewhere else.  I'm still convinced
we should drive it on to bring it through the standards process in
lockstep with RegTAP.  It would suck to have RegTAP *and* RI1 out
there as "current standards" at the same time.

Anyway, this mail is mainly to announce 

http://volute.googlecode.com/svn/trunk/projects/registry/regtap/RegTAP-fmt.html

which is the broken-out spec for RegTAP.   

As I said in SP, I feel it's urgent to fix the Registry situation,
which would mean that *something* ought to be in RFC by this time
next year latest.  Hence, I'd very much love any kind of contribution,
feedback, and first and foremost implementations.  

If you want, you can have a look at how we're doing things over here
at http://svn.ari.uni-heidelberg.de/svn/gavo/hdinputs/rr/ (which is
not always *really* pretty; the interesting part is
res/vorgrammar.py, which may shock you, but almost all the complexity
is about supporting OAI and STC-X).

As before, you're also welcome to edit the stuff at

https://volute.googlecode.com/svn/trunk/projects/registry/regtap


[[[[[[Diatribe on STC following, you're free to ignore it]]]]]]

You'll notice that I've not yet entirely taken out the stuff on STC
coverage; it's just grayed out.  I could pull it into another spec,
but I feel that'd be a shame; this is fairly easy stuff, potentially
highly useful, and if it's broken out, it'll probably never be
passed, and we probably won't have usable STC searches in the
registry for another 10 years.

Admittedly, it's hard to collect the information in these tables from
current STC-X, but that can be remedied by recommending that people
only use a handful of STC constructs (everything else would be
ignored).  In that way, people could finally search against STC
constraints in the Registry.  Wouldn't *that* be cool, for a change?
At least my take on it is that we've waited long enough for more
sophisticated stuff to come along, and really: requiring that
resource record authors futz around with MOCs or similar such magic
will probably not improve uptake.

Also, of course, it's a separate data model that can be located
separately, which means that people are free to ignore it anyway.

If you feel strongly that the STC stuff will delay the progress of
this (hopefully) standard, I'll quickly take it out.  You can even do
it yourself, the SVN URL is above.

But please have a look at it first.

Thanks,

         Markus