Toby said:
The fact that WYSIWYG editors generate dodgy output is not down to bad
implementations, it's because they are based on a fundamentally flawed
principle: that HTML is a visual description.
To a large extent, the combination of (X)HTML & CSS *is* a visual description.
It is not *only* that, but that is a large part of it.
Note that any true WYSIWYG editor has to read the CSS for the document being
worked on, and decent ones not only allow styles to be attached to the visible
features, but also the CSS can be edited too. Even Dreamweaver 4 did that, and
that came out years ago! So I take it for granted that a WYSIWYG editor will
help develop both (X)HTML & CSS in parallel.
I accept that additional "views" would be valuable. An "outline view". A
"linearisation view". A "site navigation view". But, given the fact that most
people who access my pages will be viewing them on a screen, it makes sense to
make that a major view.
No doubt it would be possible to create a good graphical HTML editor,
but it could not be described as WYSIWYG. Think along the lines of
Mozilla's DOM inspector, which shows the document tree and a preview
of the file. Now consider making that into an editing interface.
I still want, indeed demand, indeed *get*, an editor that shows me a
resemblence to what the page I'm working on will appear as to most of my
target audience. I want, and get, the ability to type into cells, type
directly into a heading, drag & drop, examine the relationship between
headings & text & images, etc. I find it a very convenient working style, and
one that gets better each year for a whole set of web pages. I see no reason
to compromise - I'm not asking for a fundamentally new type of tool, but
rather significant improvements to an existing type of tool.