Re: Help Wanted: Test Cases for "Fussy" Mode

* Terje Bless wrote:
>Also welcome would be examples of where «Fussy» mode actually does its job;
>namely catches errors that would normally have been overlooked.

It is by definition not possible to provide such examples. Fussy mode is
designed to complain about things that are known to break in common
browsers. If it breaks in common browsers it would normally not be
overlooked. Fussy mode - as currently implemented - is thus essentially

  [x] more confusing and unhelpful error messages

as it complains about many things that are known *not* to break in
common browsers. It is absurd to complain about each and every table out
there because <tbody> is missing, a HTML element many web authors do not
even know of. Similar for attribute minimization, the validator does no
longer like

  <option selected>...</option>

it tells you

  the name and VI delimiter can be omitted from an attribute
  specification only if SHORTTAG YES is specified

Not only that Validator users typically have no idea what "VI delimiter"
or "SHORTTAG YES" is and they validator does not attempt to help them to
fix the "problem", the "solution"

  <option selected="selected">...</option>

is known to *cause* problems with web browsers.

If fussy mode gets fixed to complain only about things that are real
problems, the vast majority of validator users would not notice fussy
mode as they looked at it in one or many of these common browsers and
those who do will not really benefit from it as loading their page in
Internet Explorer would most likely make them aware of the same issues.

There are so many features that would be of real benefit for Validator
users like textarea validation, batch validation, better error messages,
validation of attribute values, validator web service, etc.pp. I would
really appreciate if development gets focussed on these features. If you
really want to do fussy mode, do what WDG validator does, their "fussy
mode" does not even need a checkbox or considerations about beeing on or
off by default.

Received on Saturday, 30 August 2003 18:51:40 UTC