Re: img@relaxed CP [was: CfC: Close ISSUE-206: meta-generator by Amicable Resolution]

On 8/5/12 5:13 PM, "Leif Halvard Silli" <>

>Leonard Rosenthol, Sun, 5 Aug 2012 09:21:44 -0700:
>> PDF  has Alt as an optional [snip] MAY
>> PDF/UA  however, [ forbids ] missing Alt, but not an empty one.
>1. do you have both tooltips - AKA @title - and @alt?

Tooltips only exist on certain objects, such as annotations and fields.
They don't exist on text or images.

We do, of course, have captions for images - which would be similar to
title (I believe).  And for text, there is something called ActualText
which can be used for things such as abbreviations of human-readable
explanations for symbols (eg. "the artist formerly known as Prince").

>2. does an empty @alt imply that the object it sits on
>   is just presentational/decoration?

In PDF, all content is presentational - at least in that we don't make a
distinction.  PDF proper doesn't care about the empty alt and would just
pass that on to AT (Assistive Tech).  PDF/UA, however, would consider that
invalid and such a file wouldn't pass (human) validation.

>3. could an object have both a - possibly empty -
>   alt plus a non-empty tooltip?

In PDF, yes.  In PDF/UA, no.

>4. is there any semantic difference between an item with no alt
>   vs one with an empty alt vs an alt with just white space?

There isn't anything in the standard(s) that would make such a difference,
but in practice as it relates to communication with AT, if there is no
Alt, then there is nothing sent to AT for it to read for that object.  If
there is something (be it empty or whitespace), then that will be sent -
how the AT device handles it, I don't know?!?!

>>HTML pays different attention - and applies different principles - to
>different elements. If one uses <OBJECT> as an image element, then the
>validator will flag nothing regardless of whether one fills the
><object> with fallback text or not. Thus, we could say that only WCAG
>rules the ground. But, for <img>, then it is neither the PDF model nor
>the PDF/UA model that counts.

Agreed that the rules should vary for different object types - as they do
in PDF/UA as well. 

>But a mixed model where there are rigid
>rules for what empty alt, no alt and non-empty alt means, then there is
>a second level of what is valid  plus another level of ARIA attributes
>that are not properly taken account whether when calculating the
>meaning or when checking for conformance.

But can everything truly be checked by machine?  Others have said that
proper WCAG compliance (and I would suspect ARIA as well) will require
human intervention.  So then some of those rules aren't ones you are apply
at the validator.   Also, are they all rules that belong in HTML proper vs
in some other related standard (be it WCAG, ARIA, etc.)??


Received on Monday, 6 August 2012 01:41:07 UTC