[tei-council] Re : build failure

[Sebastian Rahtz]

On 14 Dec 2013, at 11:10, Lou Burnard <lou.burnard at retired.ox.ac.uk> wrote:

> Sho nuff. Cue long boring discussion about whether validation shd always include schematron constraints or not. 

Hmm. I just looked at this.

If I switch the Schematron validation to only look at examples marked as valid, that will get the result you want today,
but the warnings will remain switched off for ever, and so not kick in when the real moment arrives to kill the beasts.

So, suppose we have element <foo> with inherited attribute @bar. we have an example using <foo bar=“…/>
in the elementSpec for <foo>. We have decided to deprecate @bar, so we mark the example as invalid, 
and add @validUntil to the <attDef> for “bar” (in its attribute class). Come 2014, the warning on @bar kicks in,
so the build fails, and someone goes in and removes that <attDef>. However, they are not warned to emend the example for <foo> which
is now permanently marked as not needing validation, so it goes forward with its wrong example. This is bad bad bad.
In the case where the whole element is to be zapped (<relationGrp>), things will be OK, but not otherwise. Remember also
examples in the chapters, which would not be caught by removal of the elementSpec

Therefore I suggest that your change to mark those examples as invalid may have been unwise, and should be reverted.
I’d be happier to live with the build marked as cloudy-with-a-chance-of-warnings for a while until we have a more
foolproof method.
--
Sebastian Rahtz
Director (Research) of Academic IT
University of Oxford IT Services
13 Banbury Road, Oxford OX2 6NN. Phone +44 1865 283431