Browser ompatibility with implementation bugs (was: Re: default.css)

On Mon, 18 May 1998, Ian Hickson wrote:

> Of course, this will require reading the DOCTYPE - and inferring one should
> it be missing - and parsing according to the correct DTD. (I am day dreaming
> about the day where we see DTDs shipped with a browser... which are actually
> used by the parser!)

I don't think this is a good idea. In the matter of bugward (ha!)
compatibility I think the best thing to do is to simply check the DTD. If
it's non-existant, enable bug compatibility mode. If it's existant, disable
it. Since bug compatibility will mostly be added checks (no space without a
</P>, allow improperly nested lists and so on), you simply disable these
checks if the user supplied a Document Type Declaration, or even just an
HTML 4.0 declaration. This way older documents will work with the bugs,
while newer ones will work according to spec. And after a browser generation
you can phase out the old stuff completely. Yes, there *will* be people who
wrote bad HTML 4.0 and included a DTD and have their pages broken (or more
likely, appear simply a little different than they expected). That's what we
*want*. It's a lot better than people writing *proper* HTML 4.0 and having
their pages illegible or inaccessible.

Personally I find the whole idea ludicrous. Every single HTML author ends up
coding according to what NS or MS asks them to, and then both companies
refuse to change their recommendations because they're afraid they'll
contradict their own recommendations. When will this end? You can't produce
a program that supposedly handles a well-defined concept (HTML, HTTP, Java,
whatever) and force the author/programmer to change his style to suit your
broken implementation. You've got it backwards. But I'm sure that I'm not
really the one who's got the problem here. I'm wondering how the people on
the W3C's HTML WG feel, the people who slaved for so long to give us a
specification that incorporated all the broken features (HTML 3.2) and then
another specification that is backward-compatible with those features (HTML
4.0) and then sat back and watched the industry (of which some of them were
part of) ignore their efforts.

What was the point of 3.2 and 4.0? Why do Netscape and Microsoft pay their
subscriptions to the W3C and the salaries of the employees who work in its
working groups if they value bugs more than the specifications they *pay*
for? Mr. Wilson brought forth the example of MS Office incompatibilities
causing user gripe. This is not a good analogy. The web is not the same as
your hard disk. I'm confident that if Microsoft came up with a version of IE
that truly, faithfully, completely followed a *specific* standard (and not
"parts of HTML 4.0", meaning not even proper HTML 2.0), you'd be seeing
little "best viewed with IE" buttons all over the Web in a matter of days.

What every author of Web pages wants these days is the ability to go to
the W3C's web site, print a copy of the latest recommendation (or the latest
recommendation he knows is implemented) and be able to code to that, without
testing it against any browser or looking up any other recommendation. And
that is the most valuable feature browser manufacturers can offer their
customers, period.

-- Stephanos Piperoglou -- sp249@cam.ac.uk -------------------
All tribal myths are true, for a given value of `true'.
                         - Terry Pratchett, The Last Continent
------------------------- http://www.thor.cam.ac.uk/~sp249/ --

Received on Tuesday, 19 May 1998 15:14:10 UTC