[tei-council] TEI Conformance

[Syd Bauman]

> you'll be amazed (:-}) to hear that I think all these statements
> are what we should be moving away from,

Yes, actually, I am amazed that it's _all_; I had only expected
_most_. :-)


> > * the TEI is not in the business of certifying things
> I would say that is why we have the Consortium

Those who wrote the Charter did not agree, apparently. (Although
certifying things *could* be part of what we do, it is not *why* we
have a Consortium.)


> > * conformance is like validity: either you are or you aren't
> the current draft of conformance does not really
> support this statement

To what draft are you referring? Nothing in
http://svn.sourceforge.net/viewvc/tei/trunk/P5/Source/Guidelines/en/CF-Conformance.xml?view=markup
seems to support a variety of degrees, levels, or types of
conformance. Did I miss something?


> > * conformance must permit scholars to create their own,
> >   conformant, arcane encodings for things we haven't thought of
> >   
> s/conformant//    and I would agree...

And this is the absolute and most important crux of the issue that I
am repeatedly harping on. It is *absolutely essential*, no less, that
TEI continue to support as "conformant" encodings that are outside
the off-the-shelf TEI box. If we don't, we will find lots of projects
that are faced with the choice between tag abuse and loss of funding.
Most will choose tag abuse, or (as sometimes happens now) just say
that they are TEI conformant when they're not.


> > * conformance should make it easy for scholars to do so
> Are we in the business of doing it right or making
> it easy?

Notice my wording: we "must" permit scholars to do so, we "should"
make it easy for them. We are in the business of doing it right
first, but not to the exclusion of all else. (Otherwise, we would
have dropped support for P5 DTDs long ago :-)


> > * it is very important that we continue to make efforts to lower 
> >   the barriers to TEI usage
> sort of. But I want to _raise_ the barrier of conformance....

Raise it? It is already so high that many, if not most, projects
don't actually meet it. Instead they do something that results in
"TEI like" encoding, e.g. starting with TEILITE.DTD and modifying it
directly.


> > And yes, that second bullet point entirely negates the idea of
> > different "levels" or "types" of conformance. What is being discussed
> > at http://www.tei-c.org.uk/wiki/index.php/Conformance is *not*
> > conformance, but rather "degrees of interchangeability" or some such.
> I might agree that the phrase "degrees of interchangeability"
> is more useful. 

Good! You, James, and at least one or two other people on this list
have all agreed to my proposal that we consider "degrees of
interchangeability" the place where we discuss things like the fact
that it is easier to interchange documents that conform to a strict
subset of tei_all, etc., reserving the word "conformance" for only an
adherence to a very encompassing set of general rules much as we had
in P4.

However, every time the discussion comes up the word "conformance" is
used with the former meaning, or at least is ambiguous.

Sigh.


> But in that case I don't think conformance means very much at all,
> and we might as well drop the word if it means no more than a
> self-assessed moral claim to being good.

I think this means you really don't understand TEI conformance at the
deepest levels. It is not about making interchange trivial, or even
easy. I would say it is about making what is needed for interchange
explicit.

But I'm not sure at all where this "self-assessed moral claim to
being good" business comes from. Let's take P4 conformance as the
example, since we haven't defined P5's yet. The rules for conformance
are very explicit, very clear, and (for the most part) syntactically
not hard to test at all. On the other hand, they are hard enough to
follow that many projects choose to just hack the TEI Lite DTD, and
semantically they are impossible to test. In P4 is conformance "no
more than a self-assessed moral claim to being good"? 

Personally, I'd like the rules for conformance to P5 to share all the
features of P4's, except the "hard to follow" part.


I really gotta go ...