[tei-council] Re: The Naming of Classes

[Lou Burnard]

Christian Wittern wrote:

> 
> For model classes that are populated from modules (like the
> *dictionary* stuff), this can not anymore be guessed from the name.
> Since in practice the knowledge were something comes from is needed to
> write customizations, it would be useful to see this reflected
> somewhat in the naming.

I agree with this objective: I tried to do this to some extent by using 
"entry" as the prototype name, since one always talks about dictionary 
entries. But I agree that more attention is needed here. The dictionary 
chapter is the one which uses classes most heavily at present, and I 
need to study the way it does so a bit more carefully.

> 
> *Part and *Like are not really clearly distinguishable. And *Like
> really sounds a bit silly.  In what way is chunk pLike? If siblings
> are the criterium, pSiblings would maybe make more sense?
> 


I am sorry if the two are not distinguishable, since the distinction is 
really rather critical! "chunk" is "pLike" in the sense that its members 
are "like" ps -- they occur at the same hierarchic level as p  and they 
are semantically associated. I agree that this is a bit of a stretch for 
  "chunk" though, since the members of that class really have nothing in 
common except for their hierarchic position. I hope you agree that the 
distinction works better with e.g. bibLike and bibPart.

I didn't choose "sibling" because it seemed to me that many xLike class 
members would have siblings which were *not* xLike. For example hiLike 
things are not really dateLike, but they are both siblings (and both 
"pPart"s!). But I've no quarrel with changing "like" to "sib". (It would 
get rid of my all time favourite which no-one has yet commented 
on/noticed i.e. "model.uLike" )

As Sebastian points out, deciding  whether something is "xLike" or 
"yPart" for a given x which is a member of y is likely often to be a 
rather subjective decision. The TEI is full of rather subjective 
decisions about how things should be named, and I'm very happy to review 
particular cases, or -- if the cases turn out to be rather numerous -- 
to rethink the whole exercise. But what alternative proposals do we have?

> All the best,
> 
> Christian
> 
>