Re: Validator fails to detect invalid document

Nick Kew <nick@webthing.com>:
> On Mon, 11 Nov 2002, Bjoern Hoehrmann wrote:
>
>>> Depending on feedback, this may or may not change before release.
>>> Prohibiting this particular SHORTTAG feature is strictly speaking up to
the
>>> HTML WG [...]
>
> But we know there are bugs in the spec, and that they regard XHTML
> as the bugfix.

SGML shorttag features are not bugs in the specs--the browsers are the ones
having bugs.

>   (b) Though valid de jure HTML, it is not de facto valid.

Since when is "browser compatible" a synonym for "valid"? Lynx gets it
right, AFAIK.

>> Exactly. Don't touch their stuff.
>
> Disagree.

I'd be mad at a police officer who stops me in my car for driving too slow,
just because /he thinks/ there should be a minimal velocity.

> IMO the approach of htmlhelp and valet (default to
> a parse mode that complains of shorttags) is right.

Than they're not validators but advisors. The one hosted by the spec making
institution, should be a real validator.
Anyhow, the W3C CSS Validator knows to distinguish "Errors" and "Warnings",
wasn't that an option for the W3C Markup Validator, too? In particular all
used shorttag features should result in a warning, including " />" in XHTML
documents delivered as text/html. Missing charset declaration is only worth
a warning, too, if no characters from outside the default charset are being
used.

> Validator is aiming for more simplicity, so "strict sgml"
> mode should be relegated to the "advanced" form.

IMHO, it should be the other way round, give an option to /disable/ strict
SGML mode.

Christoph Päper

Received on Monday, 11 November 2002 07:21:09 UTC