[ntp:questions] Re: Busted link on http://www.eecis.udel.edu/~mills/ntp/html/index.html

Dale Worley worley at dragon.ariadne.com
Sun Oct 19 19:55:49 UTC 2003


vjs at calcite.rhyolite.com (Vernon Schryver) writes:
> If it doesn't work in any one of those, it's bad HTML even if the W3C
> and the IETF say its perfectly good and proper.  I've had to give up
> on some perfectly standard HTTPS pages because ancient versions of IE
> and current versions of Safari cannot handle them.
> 
> Well, that does assume that your purpose is publishing to a wide audience.

As Vernon says, the ultimate purpose is to be useful, so what
restricts the audience is not good.  OTOH, things that are not
conformant to the standards are likely to not work with *future*
tools.

I've always used the ancient construction:

        <h2><a name="foo">This is the heading</a></h2>

IIRC, that was valid in HTML version 1, so it has good compatibility
with old tools.  Putting an ID attribute on things is a newfangled
innovation and I'm not sure I hold with it.

In regard to the value of an A-NAME attribute, or an H2-ID attribute,
digging through the HTML 4.1 specification makes it clear that it may
not start with a '#'.  Though no doubt many browsers will cope with
that mistake, since they have to be compatible with the mistakes of
the past.

As for popular reference works, there is an oddity.  While a
dictionary of 1000 pages is likely to be extremely accurate, a 1000
page "Ultimate Reference to HTML" is likely to be quite poor.  There
seem to be special publishers who write massive references on popular
topics, whose basic assumption is that the reader is incapable of
thinking through anything.  So the book is organized as "If you want
to do X, do A, B, and then C."  The exception is O'Reilly, whose books
are of consistently high quality.

I do wish to thank Dr. Mills for cleaning up this problem.

Dale



More information about the questions mailing list