On Sat, Apr 08, 2000 at 10:44:56AM +0200, Asger K. Alstrup Nielsen wrote:
> [Properties of individual characters or words]
> > After some reflection I think that contry is a bad example, most of the
> > regular logical attributes should be word related and thus some sort of (simple)
> > inset, and not font related.
> >
> > Will this simplify the underlying architecture?
>
> I can only refer to the documentation that I wrote about the new kernel
> from last year. All these issues are discussed in that document, and
> the conclusion is the same as Lgb has come to:
>
> We need a attributes with character resolution.
> Call it CharAttrib, TextAttrib, CharParams, TextParams, or whatever.
> It is all the same.
Note that that I agree with that, sure that we need attribute with
character resolution. What I dont agree is that Contry should be, as an example,
one of those attribute.
Probably the same applies to language, we don't a expect language to be related
to a single character. Any example against?
> Beware: With the current LyX text representation data structure, we
> might get a substantial increase in memory usage, unless a flyweigth
> pattern is employed.
There are certain compromises to be made.
> Regarding resolution: There is no point in limiting this support to
> word boundaries, because then you can not reuse the mechanism for
> the font attributes. Consider things like "_F_ile -> _S_ave".
Maybe I wasn't clear (that isn't difficult ;) I am not against the char resolution
attributes, I simply think that we should have middle range attributes that are
perhaps best descrived as insets. And logical structures as Contry or Name should
be one of those cases.
Clear now?
> Greets,
>
> Asger
Greets,
--
José
