Re: [Gzz] New incomplete PEG-- please comment

[Benja Fallenstein]

Tuomas Lukka wrote:

> On Thu, Oct 31, 2002 at 12:00:12AM +0100, Benja Fallenstein wrote:
>  
>
> > [...]
> > In Gzz, let's make the primitive *formatted*, xanalogical
> > text. Like it should be possible to link to and transclude
> > all text, it should be possible to emphasize and color
> > all text, for example.
> >    
>
> Applause. Excellent reasoning, and proposition.

Thanks :-)

> But lots of problems: where do you *stop* the allowed formatting?
> Should there be
>
>         - bold
>         - bold, underlined, 15pt
>         - subscript/superscript (equations)
>         - integral signs with bounds (a la TeX)
>         - tables
>         - images
>
> Where is the line drawn?
>
> This should probably be an issue.

As I said below, I'd like to consider this to be external policy 
matters: The mechanism proposed by this PEG is to attach styles to 
individual characters/units in the enfilade, and to allow users of the 
enfilade to query these, where each style is simply a URI (= cell id). 
How to interpret these is completely open. This seems like a good thing, 
since it allows lots of experimentation on the higher levels.

This *is* drawing a line-- a piece of text is simply a string of 
characters with xanalogical ids and a set of styles attached to them, 
and things you cannot represent through this mechanism aren't catered 
for. In the above, "bold" and "bold, underlined, 15pt" will be easily 
represented, "subscript/superscript" should be ok too (like in HTML), 
"integral signs with bounds" could be a little harder, "tables" probably 
can't be represented well, and "images" are completely out of scope (but 
we have them already by using ImageSpans! :-) ).

However, I have a simple, orthogonal proposal that could help with a lot 
of these issues in a similarly abstract way as this PEG handles styles 
attached to spans: a new UnitSpan which is defined simply by a URI and 
interpreted by outside code. That way, we could insert anything 
"special" we like into an enfilade, interpreting it through the 
connections of the cell with the same id, or something like that (just 
as with styles). This could be good for sane document-specific 
"characters" (such as images showing buttons on the screen) or as a 
general inclusion mechanism (to include tables or equations, both of 
which I think should be defined in structure, not inside the enfilade).

If there's interest in this, I'll write a PEG that discusses the 
different options in detail.

> > However, there are two major problems with this exciting idea.
> > The first I discovered was: What if you have the same characters
> > in two different contexts? 
> >    
>
> Actually, this argument is a misunderstanding of the idea:
> I first thought it would be like this and was against it because
> of this. But: the idea was actually that the formatting links
> would be a part of the *document*, and applied only to that
> document.

Yes, as said below, I was able to deduce this. But as also said below, 
this takes away the property I like most about this scheme ;-)

> > This can happen even in the same
> > document: the same text can appear in a header and somewhere
> > in the body, possibly in a reference to the header
> > or something. 
> >    
>
> But this is again true even in that model.

Exactly...

- Benja