Menu
▾
▴
#620 which part of TEI Guidelines takes precedence
James and Lou said on 2013-11-11 during the Council meeting that the Guidelines prose takes precedence over the specifications. I can't find this stated in the Guidelines but believe it should be added. When doing so, be sure not to call the Guidelines prose "the prose" because it could lead someone to think we are referring to text appearing within a spec as opposed to the content model specified therein.
Please take the following as an expression of concern, made (obviously) without knowing the context of the decision above. And hopefully unnecessary.
At least two milestones are relevant as the basis for what I want to say: (i) the switch from Pizza Chef to ODD, as a tighter and in many ways more accessible mechanism for defining TEI vocabulary and sharing it (the requirement of supplying the ODD also matters) and (ii) the recent delegation of even more power to the ODD language.
Back in 2001, when I fell in love with Lou at first sight, he preached about the Chef and about how the Guidelines supplement the SGML (freshly turned into SGML-like XML) with instructions for interpretation and hints for interoperability. Then came ODD, which made it possible to state many requirements more thoroughly, explicitly, and to more closely connect the verbal description with its machine-readable expression, along the Knuthian principles. (There was a shift of balance there, and this is the core of this note).
Very recently, ODD gained even more power, abstracting half a step further away from RNG, and arguably maturing in the process.
Now, I am somewhat concerned with the message that Kevin relayed above, at least the way I understand it, because what it makes me think of is a step backwards towards the times when the TEI had to rely on prose, because it had no means of uniform and powerful expression. But now it does! And the message that can be (hopefully mis-)read as saying "when in doubt, ignore the ODD and rely on the prose" seems to me exactly the wrong message to send, because it suggests that the balance is getting pushed back, away from the freshly upgraded ODD, and towards the prose (and in which language? isn't the ODD a way to formally express what the individual translators may render imprecisely in their target languages?)
The role of the Council is to maintain and develop the Guidelines, among others by maintaining the underlying ODD. So, I would say, if there is a discrepancy, it means that the Council messed up (a perfectly human thing) and should get the problem fixed -- and thus "when in doubt, communicate": maybe someone is late with the prose description, maybe they forgot to commit, or maybe the two are in agreement, except that someone expressed that agreement very poorly.
Bah, as I said above, I hope that this note is completely unnecessary. Keep up the good frog!
;-)
I agree with this and share Piotr's concern; to me, the spec should take precedence over the chapter prose (although if they're in disagreement, it's obviously an error and needs to be corrected).
Assigning to myself to take forward.
If there are inconsistencies between what is stated in different parts of the Guidelines concerning the same phenomenon, those inconsistencies MUST be resolved. The resolution should NOT take the form of saying "this part of the Guidelines (always) takes precedence over this other part". If I ever said anything like that, I was provoked into doing so by Martin's assertion that the element specs were all you need to read. However there is a difference between being inconsistent and being under (or over) specified, which is what I think Piotr is rightly concerned about.
In my opinion there are some aspects of the TEI tags intended meaning that cannot be expressed in a complete way by the specs, so I would not assert that specs always takes precedence over Guidelines. But we cannot say the contrary either, since in http://www.tei-c.org/release/doc/tei-p5-doc/en/html/USE.html#CF conformance to schema is required for conformance. The only solution I imagine is adding a statement saying that the notion of semantic conformance cannot be completely enforced in a formal way so the prose is to be considered part of the conformance notion; if there is an evident inconsistency between Guidelines prose and specs that is to be considered and error to be reported to Council.