> Benjamin Franz writes:
>
> > So I repeat: It is time to officially disband the HTML WG.
>
> No, it is not.
>
> I can't put it in any better than another Spyglass employee did, at our
> recent partner's conference, so I'll try to quote:
>
> "The Web community needs to be informed of the cost of maintaining
> millions of pages of content which match a language with no formal
> description."
In the interest of keeping this thread somewhat contained, I am replying
to Eric Sink to address a point raised by three other people as well -
the need for a work group for the development of a good markup language:
I agree.
There is without question a need for *a* work group to develop a good
platform independant markup language I, and I really believe there should be
a work group for one. It just can't be HTML.
HTML has some fatal flaws in terms of becoming (because it most certainly
is not now) such a platform independant markup language. These flaws are in
the relationship between the language and the people using it.
The people who conceptualized HTML thought in terms of highly technical
users communicating high grade information. They designed a language
that *assumed* that most everyone matched that audience. Thus they didn't
have to make explicit in the language itself the link to SGML - everyone
using and developing the language could be assumed to have that knowledge.
They could be expected to understand you can't say <i>this<b>
text</i> changes</b> appearance in SGML. They could be expected
to read DTDs (or to learn to do so). So far no real problem. A well
defined set of target users and language.
Well, since January 1993, the assumption about *who* was going to be
using the language has been radically violated: It is a mass medium used
primarily by non-technical users who wouldn't know what SGML was if it
bit them. The language itself provides no hint to the non-technical user
that you *have* to have a <title> in a document, or that you can't say <b><p
align=center>Centered Bold Text</p></b> or any of a million other things
you can do wrong. Most of them have no idea what a 'markup language' is
(other than HTML is supposed to be one) and couldn't read the DTD anyway.
David C. mentioned that he is a proponent of the HALSoft validator. So am
I. But I also looked at their counter of people using it. Slightly over
130,000 [1] documents have been submitted to them for validation. The real
number of unique pages submitted is certainly at least a factor of two lower
due to repeat customers. Even if someone fixes their page on the first
try, they will send the same page at least one more time to check their
fixes. Even counting mirror sites, I doubt you could push the net count
to over 200K submissions.
All those who believe that the number of validated HTML pages is within two
orders of magnitude of the number of web pages currently in existence,
please raise your hands now. A salesman will be by your table to provide
a propsectus on some very fine beach front propery in Florida. No checks
accepted, small bills only please.
In case someone is going to suggest that the low number is because of the
huge number of people downloading the validator for themselves, think
again: only 441 builds of the validator tool kit have been made [2].
I'm one of them.
> Sure, we need a user's guide to HTML. There are some.
Dozens, maybe hundreds, actually [3]. Most of them are wrong. Both
NCSA's [4] and Netscape's [5] (which is where *most* people get their first
'official' guidance to HTML) guide's are definitely at variance with the
DTDs. And let's not forget the world of commercial book publishing.
I do not think I have seen even *ONE* such book get it right.
> But we MUST have a formal description of HTML. That's the bottom line
> agenda of this group. It's essential.
Sure. And actually we *already* have a formal description of HTML.
Several in fact [6]...If the IETF would simply accept them as official we
would have *completed* a major part of the mandate of this group.
> If you have a better way of reaching that goal, then by all means, jump
> in and solve the problem. Until that surfaces, it will continue to be
> my policy to support what we have, and to help make it better. Right
> now, that is this IETF Working Group.
I am fully behind both those goals. I simply do not believe it can be
achieved within Content-type: text/html. HTML was hijacked two and half
years ago - when the <IMG> tag was introduced. We have *NO* prospect of
regaining control of it. Anyone who believes we can has not been paying
attention to what is going on around them and how HTML is being used in
practice. It has been getting worse and worse at an accelerating pace
and shows no sign at all of reversing that trend: 'She's gonna blow,
Captain!'
When I propose disestablishing the HTML-WG I am *not* proposing stopping
development of a powerful SGML based hypertext markup language. I am
proposing that a *new* group be formed, with a new Content-type as its
mandate and that it *start* by fixing the mistakes that let this
horrible situation develop in the first place. It needs to BEGIN with
SGML compliance and tools.
Learn from all the mistakes made with HTML and clean out the
garbage that HTML has accreted. *Start* with stylesheets and completely
eliminate physical markup from the language. No more
<UL/OL/LIST/MENU/DIR> type splits, no more tags and attributes that are
needlessly abbrievated until they carry no useful mnemonic value (FIG,
REL, REV, --, ...). No documents without the <!DOCTYPE> explicitly stated.
DTD based validation software for all three (MS-Windows, Mac, Unix) major
platforms. Natural language versions of the DTDs. Modular DTDs. True
international language support. Explicit <ALT> handling of unknown
markup in a way that allows *single* documents to be maintained from the
most advanced feature set to the simplest.
HTML is an experiment gone Frankenstein. It is time to learn
from the mistakes and start a *new* experiment.
-- Benjamin Franz[1] HALSoft Validation Server Stats; <URL:http://www.halsoft.com/html-val-svc/service_stats.html> [2] HTML Check Tool Kit Build Statistics; <URL:http://www.halsoft.com/html-tk/stats.html>
[3] Yahoo - Computers and Internet:Internet:World Wide Web:Beginner's Guides; <URL:http://www.yahoo.com/Computers_and_Internet/Internet/World_Wide_Web/Beginner_s_Guides/>
[4] A Beginner's Guide to HTML; "The primary exception to the pairing rule is the <P> tag. There is no such thing as </P>. " <URL:http://www.ncsa.uiuc.edu/General/Internet/WWW/HTMLPrimer.html#A1.8.2>
[5] Extensions to HTML; <URL:http://home.mcom.com/assist/net_sites/html_extensions.html>
[6] HAL Software Systems HTML Service and Products; <URL:http://www.halsoft.com/html/>