[tei-council] content model of <surface>

[Lou Burnard]

The reqsons it is this way is because the wg expressed the (admittedly 
rather vague) desire that old skool <surface> (for digital facsimile) 
and new skool (for genetic flummery) should have only limited co-existence.

My vote would be to do the P4 thing -- i.e. just alternate the whole lot 
and sort out any nonsense that surfaces later with schematron rules.


On 09/05/12 10:29, Sebastian Rahtz wrote:
> On 9 May 2012, at 10:15, Lou Burnard wrote:
>>
>> Looks still non deterministic to me. If you have<surface><lb/></surface>  which bit of the model is being matched?
>
> the point is that once you enter the second half of the model, things from model.global
> only occur after we have had at least one example of zone/line/surface/surfaceGrp
>
> its the same technique used in div and postscript.
>
> is the intention that a<surface>  must first get out of the way any graphics and glosslike elements, and
> then get on with its main business?
>
> if that main business can _start_ with anything from model.global, then we are screwed, cos
> <certainty>  comes in both the preliminary matter (glosslike) and the main matter (global), and
> non-determinism fires.
>
> the answer, of course, is to promiscuously interleave graphic and glosslike in the preliminary matter,
> so I'm going to implement that.
>
> scream loudly if
>    a) you have the faintest idea what I am talking about
>    b) you care one jot about whether its important that all the glossLike children of<surface>
> occur before  all the graphicLike children.
>
> Trust me, I'm in the zone on this one.
>
> Sebastian
>