[tei-council] gaiji in PCDATA, or <g> in text

[Laurent Romary]

Le 9 oct. 05 à 15:57, Syd Bauman a écrit :

>> with the caveat that the macro does not work in all situations
>> (because of DTD weird things), so sometimes "text | tei.gLike" is
>> needed by hand
>>
>
> OK, I've taken a quick glance at all of these (list appended), and
> none of them are obvious candidates for excluding <g>, so I've left
> them all as permitting <g>.
>
> A few questions arise:
>
> <gramGrp>: why does this element have text at all? The content should
>            probably be something like
>            tei.Incl*, tei.gramInfo, ( tei.gramInfo | tei.Incl )*
>            no?

This is because you may use this element to tag semi-structured  
grammatical information where you may have text ('labels' or other  
comments) interspersed with actual elements like <pos>, <gen>, <num>  
etc.

>
> <re>: If character data is allowed, then <g> should be allowed. But
>       why does this element permit PCDATA at all? I thought it was
>       roughly analagous to <entry>, which does not.

Même cause, même effet. <re> is used to describe entry-like objects  
within a real entry (e.g. compounds). Such information is sometimes  
less structured then a real entry and may thus require that data be  
there in complement to the usual stuff (<gramGrp>, <form>, <sense>,  
etc.). Note also that we can have <dicScrap> (yark) and entryFree) as  
alternates to <entry> whereas there is only one <re>.

>
> <pVar>: the "Note:" information (<remarks>) does not match the
>         content model; which is correct?

It is even more weird that oVar is more consistent. My advice: change  
<remarks> to mirror oVar.
Best,
Laurent