Identifiers 2.0 Public RFC results

[Markus Demleitner]

Hi Tom,

On Fri, Sep 25, 2015 at 11:55:48AM -0400, Tom McGlynn (NASA/GSFC Code 660.1) wrote:
> It's not a big deal -- the redundancy is harmless -- but I think it
> illustrates why it would be nice to have the definition of unreserved in the
> document.   It doesn't take much space and it makes it much more
> self-contained -- you can check its consistency without having to make
> external references.  I'd prefer to be able to use a single document to be
> able to tell if my ID's are valid or not.

The trouble is, that it woulnd't stop at unreserved; the document
re-uses lots of other symbols from RFC 3986, like fragment,
path-abempty, gen-delims, sub-delims; if I include those, I have to
pull in productions these depend on, and their explanations, and
fairly soon I'm hosting a large part of the RFC in my text.

I wanted to avoid that partly for lazyn^H^H^H^H^Htext economy. But
also, I might get them slightly wrong, in which case Identifiers
would contradict RFC 3986 in subtle ways cropping up years down the
road when they might be baked into applications. Or there might be
bugs in RFC 3986 that will get fixed there but would stay on in
Identifiers.

However, my most important reason is that Identifiers tries to
reflect how I think implementations should work: Start with a library
that's RFC 3986-compliant and then add what little extra stuff we
want in the VO in a thin wrapper.  For that, copying material from
RFC 3986 would actually be detrimental, as implementors would always
have to check what's RFC stuff that's already in their library and
what's special to IVOIDs.

Of course, I completely subscribe to the opinion that reading
grammars distributed over two documents is a strain.  But I believe
the alternative is worse, and I believed I could save most people
from that strain by simply giving an overview in 2.1 that should be
"enough for non-demanding applications" (if you excuse me quoting
myself).

Perhaps it should be made clearer that sections 2.1, 3, and 4 are
what sound, solid people should be reading, whereas 2.[2-6] are
primarily for standards lawyers, validator writers and other
similarly unpopular characters?

Cheers,

             Markus