Re: Emacs as word processor

[John Yates]

On Sat, Nov 23, 2013 at 10:24 AM, Eli Zaretskii <address@hidden> wrote:

> > On Fri, 22 Nov 2013 4:47:05 PM, John Yates <address@hidden> wrote:
> >

> > I want to be able to say "This is a chapter title" or "This is a step in a
> > recipe" or - most commonly - "This is a top level paragraph with no
> > particular distinctive property".

>

> On Sat, 23 Nov 2013 10:13:38 AM, John Yates <address@hidden> wrote:
>

> Are you saying that I can customize an emacs face to specify
> inter-paragraph space? a bullet glyph or numbering style? first line and
> subsequent line indentation?  That is definitely not the case with my
> emacs, current as of Nov 8th.

That's unfair: you said nothing about those features in the text to
which I responded.  You just talked about the difference between
imperative and declarative approaches to specifying attributes.  Now
you've changed the subject, so I no longer understand what are we
discussing.

This is a thread regarding adding a WYSIWYG capability to emacs.   I guess I am guilty of silently assuming a vague shared sense of what a generic contemporary WYSIWYG editor does in these sorts of situations:

Chapter title: probably force a new page, possibly number, larger possibly bolded font, possibly centered, extra trailing vertical space.

Step in a recipe: increase indentation slightly, call out a list using bullets or numbering

Top level paragraph with no particular distinctive property: revert to defaults

If you are saying that these features don't exist in Emacs, I agree:
they don't.  But I don't see the significance of that fact, since
everybody agrees that Emacs is not a WYSIWYG word processor at this
time.

I understood this thread to be - at least in part - about developing a vision of how one _might_ move emacs from its current state to one offering some kind of WYSIWYG capability.

If you are saying that these features could never be part of a face
spec, then I don't think I agree; please explain why you think so.

I did not say that.  Quite the contrary.  In saying I would give the idea thought I would have hoped that it was clear I felt the suggestion could not be dismissed out of hand.

I now have given it a bit more thought.  Bottom line, I agree that one probably _could_ implement paragraph styles using faces but I believe that to do so would be a poor design.

A better design would provide s separate, easily navigable representation of a document's (largely hierarchical) structural elements.  Such structural element objects provides obvious sites to hang per element "paragraph-oriented" formatting directives as well as a natural framework for property inheritance.  Properties (many new) would be attached to these elements  (e.g. indentation, inter-element vertical spacing, list properties, face).  Then, just as a collection of text properties bundled into a face and attached to a span of text does a good job of implementing what some WYSIWYG editors call "character styles", so too a collection of properties bundled into a face-like object and attached to a structural element could do a good job of implementing "paragraph styles".  With such bundles of properties attached to structural elements instead of spans of text calling them simply "faces" would surely be confusing.  We might call them "paragraph faces" or "structural faces" or "structural styles" or whatever...

Finally a new text property would be needed to bind a span of text to the most deeply nested structural element to which it belongs.  (Though I see no technical difficulty including that property in a face I can see no virtue in doing so.)

/john