Re: [Moderator Action] Re: CfC: Close ISSUE-206: meta-generator by Amicable Resolution

Henri Sivonen <>, 2012-08-02 08:10 +0000:

> But alerting users is what triggers the behavior in (some) markup
> generator developers that we're trying to avoid triggering!

OK, I hear you.

I am not sure that simply alerting the users with one single warning is
what would trigger that behavior in generator developers. I think that
behavior may be triggered by the *volume* of error messages that the
validator emits for output from their generator tools. But anyway, I have a
refinement to the proposal that I think can help get us past this.

> I expect that the kind of the markup generator developers who wanted to
> silence HTML4 validators with alt="" would want to make this warning go
> away, too,

Again, I wonder if instead what those tool developers are really wanting to
silence is the case where dozens or hundreds of error messages are emitted,
but yeah OK, understood.

> so I expect this proposal would not yield a worthwhile benefit
> compared to going back to the HTML4 situation but would add complexity
> on the way. That is to say, I think this compromise is worse than
> either Ted's proposal or making validators whine about the absence of
> the alt attribute as with HTML4.

OK, so to address that, here's a refinement on the proposal:

  Make the default behavior be that of a document contains any instances of
  img elements with the "incomplete" (or whatever) attribute, we have the
  post-validation validator UI highlight the text of the "Show error
  messages for img elements with 'incomplete' attributes." option --
  instead of emitting an actual warning in the error-messages section).

What I mean is, similar to the way we have the post-validation UI highlight
the name of the "bad attribute" in error messages for invalid input elements.

Because that gives a more passive/subtle indicator -- not an explicit error
or warning message -- I don't think it would trigger the generator-tool
developer behavior you've described. Yet it would still alert validator
users to a potential problem, and give the users a way to find out more
about the cause of the problem.


Michael[tm] Smith

Received on Thursday, 2 August 2012 08:49:30 UTC