Re: Interest in HTML Conformance?

David Bianco (bianco@misty.larc.nasa.gov)
Tue, 12 Apr 1994 13:32:29 GMT


Liam Relihan writes:
> On Tue, 12 Apr 1994, David Bianco wrote:

>
> > browser which does not conform to the higher levels of the spec would
> > be lost when viewing these pages. Of course, it'd be nice if both
> > document providers and document consumers would migrate up the food
> > chain into sophisticated tools which support high levels of the spec,
> > but then we've wasted the time we put into defining the lower levels.
>
> I don't think so...the lower level specs will be needed for those upwardly
> mobile clients.
>

I guess my point is that I predict there will be very few of the
clients who will need to be upwardly mobile. When the spec is
released, we hope that client developers will conform to the spec.
Once the existing clients conform, why would anyone write a browser
which is not at least on par with the ones already in use? When Lynx,
Mosaic, et al are at level 3, who will want to use a level 1 client?
Not many...

> > I propose instead that each distinct draft of the HTML (HTML+?) specs
> > be assigned a version number. This would allow a tangible, definitive
> > specification for HTML to be established, and also allow it to be
> > enhanced without the confusion of trying to figure out exactly what
> > constitutes "standard" HTML and what doesn't. Clients can still claim
> > conformance to a specific version number, which is similar to claiming
> > conformance to a level of HTML, but since only one level is current at
> > any given time, there should be no confusion over what your browser
> > supports.
>
> This is very similar to the level system...obviously the higher the level
> of your client the better, so I suppose only "one level is current".
>

Yep. It's a subtle difference, but I believe it's an important one.
I really believe that the added complexity of "levels" of support
which are all current will turn many developers and users off. It
stands to make the WWW a lot more complicated than it already is.
However, with a version numbering system, only one version is current
at any one time so there is minimal confusion.

> I reckon we more or less agree with each other, except that you are
> proposing that the versions be defined as development takes
> place...however, I still think that we shall need to decide what is "base
> HTML".
>

A base definition is definately called for. "HTML 1.0", for want of a
better name, would ideally encompass most if not all of the current
practices. I think current practice would make an excellent base for
the standard.

David