Re: [VE][247] New Error Message Suggestion

On Sun, 31 Aug 2003, Philip TAYLOR [PC336/H-XP] wrote:

> > Validator's explanations should be written for people who
> > have an elementary understanding of what validation is. And for them, the
> > _correct_ explanation would be: "This program implies an SGML declaration
> > containing SHORTTAG NONE, contrary to the HTML 4 specifications, and
> > therefore reports NET-Enabling Start-tag as an error".
>
> I'm sorry, I /don't/ agree; "NET-enabling Start-tag" is simply gobbledygook,

Sure it is. I was just saying what the correct (honest) explanation would
be, _if_ a validator behaves the way it does. There is no way to explain
its behavior in simpler language, since it _is_ performing sort-of
validation and claiming to be a validator.

If it is renamed to be an HTML and XHTML checker, then things change quite
a bit. No need to be cautious about issuing messages intended to be useful
for practical (X)HTML authoring - which are bound to be nonsensical in the
context of generic SGML validation.

Then again, a can of worms is opened. As soon as you honestly admit that
you are doing "checking" in a broad sense of the word, you are committed
to admitting that it is inherently subjective. You might _still_
distinguish between objectively verifiable claims about violations of a
specification and subjective warnings, and even classifying the latter.

> "The sequence <FOO />* can be interpreted in at least two different ways,
>  depending on the DOCTYPE of the document; for HMTL 4.01 Strict*, the '/'
>  terminates the tag <FOO (with an implied '>').  However, since many
>  browsers don't interpret it this way, even in the presence of an HMTL 4.01 Strict*
>  DOCTYPE, it is best to avoid it completely in pure HTML documents and
>  reserve its use solely for those written in XHTML".

Well, it really does not depend on the DOCTYPE but on the SGML declaration
- and for HTML 4.01, the SGML declaration is fixed. And if the user has
specified an HTML 4.01 DOCTYPE, it's very illogical (well, just wrong) to
play by different rules.

In practical terms, the "fussy" mode lumps things together in an
unconstructive way. The SHORTREF issue was seldom a practical issue
(except in some contexts where the real problem was a failure to use
quotes around an attribute value, see the old saga at
http://www.cs.tut.fi/~jkorpela/qattr.html )
and it became one _only_ because
- people were told to use XHTML and they understood part of it, _and_
- people started using validators without knowing what they are doing.
I'm afraid that no formulation of an error message or a warning
will help in either of these problems. Note that it is a problem
_only_ in the sense that people get confused with validator
messages. Browsers interpret <img /> as <img> and don't care about no end
tags. So in practical terms, the message would be useful only to the very
small group of authors who actually tried to use SHORTREF features.
(And those people can surely be expected to know SGMLese.)

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/

Received on Sunday, 31 August 2003 13:44:40 UTC