Re: Toward Graceful Deployment of Tables

Brian Behlendorf (
Tue, 14 Mar 1995 16:43:07 -0800 (PST)

On Tue, 14 Mar 1995, Luke 路客 wrote:
> On Tue, 14 Mar 1995, Dan Connolly wrote:
> >I wrote:
> >>
> >> Let me take the liberty to quote the specification you conveniently forgot:
> >>
> >> (which I find it hard to believe because you are a co-author of it :)
> >>
> >> 2.2 Undefined Tag and Attribute Names
> >
> >Just for the record: that section is in there because I couldn't get
> >it removed (i.e. it has too much momentum), not because I think it
> >belongs there. I don't think it provides a viable evolution strategy.
> >I think it will be a support nightmare. I think it is a support
> >nightmare.
> Any major upgrade of any systems will bring support nightmare in terms of
> compatibility issue. You think provide different versions of documents via
> a format negotiation is less nighmarish? I don't think so.

It is, particularly when one (HTML 2.0) can be generated from another
(HTML 3.0), and that generation happens via scripts, a publishing/CVS
system, or within the server itself though I would prefer the first or
second method. Thus, you can just edit in HTML 3.0 and all is well - I
bet you could even write a HTML 3.0 -> MozillaHTML script converting the
style sheet info into per-tag attributes.