Re: CfC: Close ISSUE-206: meta-generator by Amicable Resolution

Henri Sivonen <hsivonen@iki.fi>, 2012-08-02 09:17 +0300:

> On Thu, Aug 2, 2012 at 1:03 AM, Sam Ruby <rubys@intertwingly.net> wrote:
> > Ted (Mike, Henri, others):
> >
> > Is there any possibility of merging this proposal with the "Mint a new
> > attribute for relaxing alt attribute conformance criteria in certain
> > situations" proposal:
> >
> > http://www.w3.org/html/wg/wiki/User:Eoconnor/ISSUE-206
> >
> > Yes, there are clear differences (they even are enumerated in Laura's
> > updated proposal), but it appears to me that these two proposals are
> > relatively close in spirit.  The essential difference (other than spelling)
> > appears to be the default?
> 
> The default validator behavior is the whole point of Ted's proposal,
> so merging "If a conformance checker encounters an <img> element with
> an incomplete attribute specified, it should by default report it.
> However, it may supply an option to suppress such an error." from
> Laura's proposal would defeat the whole point of Ted's proposal. As
> for the name of the attribute, any of "incomplete", "relaxed" or
> "noalt" would work for me. (Other names might be sensible, too.)
> 
> I don't really see the point of Laura's proposal. If the default was
> to report the presence of the "incomplete" attribute, the validator
> might as well reports the absence of the alt attribute instead. As far
> as I can tell, with the default behavior Laura is proposing, then the
> attribute would be redundant.

Here's a proposal for a compromise that might allow the two CPs to be merged:

  A. Add a "Show error messages for img elements with "incomplete" attributes.

  B. Make that option off (unchecked) by default.
  
  C. Make the default validator behavior be that if a document contains any
     instances of img elements that have the relaxed/incomplete attribute
     (or whatever the name ends up as), emit a single warning per document:

       Warning: This document contains at least one "incomplete" attribute,
       which indicates it may have images that lack text alternatives. To
       see error messages showing the locations of those images, use the
       "Show error messages for img elements with "incomplete" option.
  
   That would ensure that users are always alerted to the presence of the
   attribute, by being shown that warning message. But the warning message
   is show only once per document, instead of the dozens of times or
   hundreds of times or whatever the error messages will otherwise be shown.

  D. Make the default validator behavior by that errors for the "incomplete"
     attribute are not shown. But because of C above, the users will still
     be made aware that the document has potential problems -- they will
     still by default always see that one Warning message.

Please consider this proposal carefully. I recognize it's not perfect.
We're not going to find a perfect solution to this problem. But given how
close the two CPs are to one another otherwise, I think this proposal to
merge them with the default behaviors is something we have a chance of
getting agreement about, if everybody can give ground just a little.

  --Mike

-- 
Michael[tm] Smith http://people.w3.org/mike

Received on Thursday, 2 August 2012 07:04:20 UTC