> The recent spate of mailings about software which converts various
> document formats into HTML makes one wonder whether trying desperately to
> cram every last feature into the HTML specification will really provide
> an adequate solution.
It won't! And this has been clear to a lot of people for quite a
while. But allowing every file format under the sun, isn't the
solution either.
You can do this today, btw. HTTP already use the MIME mechanism you
are referring to, to label the file it transmits. All Web browsers I
have used, have a mapping between mime types and external programs to
handle files of that type. This mapping is much better than the
extension based mapping you suggest.
But even though you *can* use a lot of different file formats, the
core of the Web, the actual Web itself, consists of interlinked HTML
files. The only thing that keeps the Web together (apart from chewing
gum, and hope) is a common file format, that a lot of clients, on a
lot of different platforms, can handle. in some way or other.
So what do you do? Well, the scalable solution is where you have a way
to describe the data format/data structure of the file you are
transmitting, and a way to tell the browser what to do with it.
A pretty interesting suggestion for using SGML do describe the data
format and style sheets to tell the browsers what to do is [1]. This
is less esoteric than you might think. Many of the parts necessary for
an implementation of this type exists already. It's more a question of
gluing them together.
- Steinar
[1] http://www.ncsa.uiuc.edu/SDG/IT94/Proceedings/Autools/sperberg-mcqueen/sperberg.html