Re: Images as alternatives to text instead of the reverse

On Wed, 21 Aug 1996, Steve Knoblock wrote:

> >But traditional printing is not platform indendant. In the example you have
> >spell checked. In other words, the computer has no idea that the graphic is
> >REPRESENTING text, so it cannot do intelligent textual things with it. In
> >books, this is obviously not a problem, but on the Web it is.
> >
> You're saying that the "display type" is representing text, and that needs
> to be indexable and searchable. In that case, I am with you. If a graphic
> were representing text, we should be able to do intelligent things like
> searching for the textual content. That is one of the reasons I want to use
> css, to eliminate the use of graphics for titles or text so they are
> indexable. I think I misunderstood what Stephanos was proposing, as the
> subject---using images instead of text---seemed contrary to the direction of
> css.
> >Alt text says: "the image is the 'real' version of the idea, the text is
> >secondary." Because of this, alt text is not usually indexed (AFAIK). The
> Okay. And the I can see the merits of being able to use tags in the alt text.
> >proposal under discussion says the opposite. "This is really text...treat it
> >like text...but graphic clients may want to use this stylzed version." Since
> >they are opposites, they should have different HTML representations.
> >
> Then I agree, we do need a way of saying that the text is the real deal and
> that the graphic is for presentation. At least for things like a graphical
> cover-art or logo with textual content where you may want that text to be
> indexed.
> Now, that I've seen Stephanos <SPAN STYLE="altimage: url(images/ablogo.gif)
> ; align: center"> usage, I support the idea for logos and the like.

Of course, this is exactly what the OBJECT tag does. Perhaps we could
generalize just a wee bit and change "altimage" to "replace". That way you
could replace any HTML element with an arbitrary object for rendering
purposes. The original content would, of course, still be searchable or
indexable in the usual manner. (I know this has been proposed before, but
I couldn't find it in the archives.) I would say this addition, since it
breaks no existing stylesheets and the underlying mechanism (inlining
objects) is already implemented in most browsers, could well fit into

In a later version of CSS (CSS2?) we could go further, with
"insert" and "append". All three of these elements could take
multiple url()s or "text string"s as arguments. The rendering engine could
step through the list, and render the first object it can handle. I f it
doesn't understand any of them, it could render the tag as if the
insert/append/replace property were not present.

Example: (I always did like HTML3)

In stylesheet:
NOTE.CAUTION { insert: url(caution.mov) url(caution.gif) "Caution: " }

<NOTE CLASS=CAUTION><P>This is a cautionary note... take heed!</NOTE>