> Since images are fetched separately the best way to deal with this is
> for the image data structure to carry the information. Intelligent clients
> can then offer exact placement, and others can just put them where they
> wish.
>
> GIFs have provision for cutomizable annotation which is both easy to
> add and decode. Other graphic formats have similar facilities.
>
> It is better to carry such layout information in the data structures
> of the graphic objects rather than overload HTML with duplicate information.
I agree that it is usually good to avoid duplication of information.
However, in this case unless the HTML specification has its scope expanded
to cover the specifics of these customizable annotations, this approach
will not receive any of the advantages of standardization. In particular,
intelligent clients will probably not address this issue with any
consistency.
In Brians's scenario, an HTML document that includes an IMG with an
alignment attribute (top, bottom, middle), an "intelligent client" would
not have much choice but to follow the alignment instruction rather than
searching thru a GIF file for a baseline annotation. Only if the IMG
construct included no alignment at all would the client feel confident in
dealing with baseline alignment.
Also, the baseline info is independent of the graphic format; the client
can process this instruction even if it is relying on other software to
render the image. If it is done in the IMG construct we have only to
specify one way to do it and clients can implement it once to cover all
supported graphic formats. If it is embedded in each different picture
format, the clients must actually process image data rather than simply
handing the image off to some generic picture-rendering code.
As to the issue of putting formatting info in HTML docs, this feature would
tie the formatting info to the image which is, by its nature, formatting
information already.
I don't want to beat a dead horse on this issue, but here is one more idea
--- perhaps "baseline" could be an acceptable value of the IMG alignment
attribute. This would direct the client to look for baseline info within
the graphic. Perhaps the HTML spec could specify how such info is to be
embedded in the most popular HTML-related graphic format (is this GIF?).
I firmly believe baseline alignment is not just a special-purpose feature.
Almost every document processor ever implemented is built around the
assumption that the combination of text and graphic elements on a page can
be used to express just about any visual information we can imagine. HTML,
SGML, and word processors place a structure on the text elements. Draw and
paint programs create and maintain structure within each graphical
element. Unfortunately, our current technology allows very few
relationships between graphic and text elements. The graphic can be placed
horizontally with some control (we can put spaces in front of the graphic
to move it to the right). Why don't we have a similar level of control for
moving a graphic vertically relative to a text line?
Paul Topping
Design Science, Inc.
[email protected]