- From: Jonas Sicking <jonas@sicking.cc>
- Date: Tue, 18 Aug 2009 10:44:38 -0700
- To: Henri Sivonen <hsivonen@iki.fi>
- Cc: Jan Richards <jan.richards@utoronto.ca>, HTMLWG WG <public-html@w3.org>, W3C WAI-XTECH <wai-xtech@w3.org>
On Tue, Aug 18, 2009 at 2:53 AM, Henri Sivonen<hsivonen@iki.fi> wrote: > On Aug 18, 2009, at 11:20, Jonas Sicking wrote: > >> My general impression of @alt is that relatively speaking it has been >> pretty successful, as far as "bolt-on" accessibility attributes go. >> Looking images that needs description (i.e. is non-presentational) vs >> tables that needs description (due to having a complex layout), my >> impression is that @alt has been much more successful than >> @summary[2]. This makes me wonder why. I can think of two reasons: >> >> 1. @alt is required, and thus a missing @alt is flagged by validators > > It doesn't explain the difference relative to summary, though, because > there's no validator-related reason to put in the predominantly bogus > summaries that have been found in the wild. I'm not sure how the bogus @summary values are related to this? The point is that there are orders of magnitude more @alt than @summary. Though technically some of that can be explained by @summary not being needed on every table. However I strongly suspect that even if you looked at just the tables that needed it, you'd see a relatively small percentage with @summary at all, vs. how common @alt is. >> 2. @alt used to be displayed by IE as a tooltip > > If the alt tooltip in IE made alt better, the campaign not to show alt as > tooltip in other browsers has been horribly misguided. :-/ Indeed. I simply don't know. > In any case, there are many other more plausible reasons including: > > 3. Alt is the foremost accessibility evangelism item. Introductions to > accessibility cover alt right away but don't necessarily cover summary. > > 4. For sighted authors, alt is conceptually easier to grasp and easier to > test in Lynx or in a GUI browser with image loading turned off. > > 5. SEO propaganda. Indeed. >> It looks to me like there are two options here: >> >> 1. Fill out a best guess value. The DC document uses the example >> "Photo X of Y" as an example. >> 2. Not fill out any alt attribute at all. >> >> The advantage of using 1 is that it allows us to say that @alt should >> be a required attribute. The downside is that it creates a bunch of >> @alt attributes which contain fairly useless information. And with no >> way for AT users to tell that the information is useless. I don't know >> how big of a problem this useless information is for AT users. I seem >> to recall some AT user saying that it's easy enough to skip an >> attribute if it appears to contain useless info. > > I've decided to no longer advocate what solution should be adopted, but I > think the signals for authoring tool developers should be coherent across > HTML 5 validation and ATAG 2. If there's WAI consensus that #1 is the right > thing to do, ATAG 2 needs to be edited to say so to keep things coherent. > >> The problem with 2 is that it goes against the requirements of the >> spec. I.e. the tool would create invalid HTML. I'm not convinced this >> is a problem. > > If it isn't from the tool vendor point of view, why do BlueGriffon and iWeb > do what they do? You'll never have things working perfectly. No matter what solution we argue for there will be some people/vendors doing things wrong. The presence of errors doesn't mean that model needs to be changed. >> This seems very similar to me to how the spec says "Tables must not be >> used as layout aids". However it's quite possible today to use >> dreamweaver to create a page that uses tables as layout. Thus >> dreamweaver today can already create invalid HTML. The responsibility >> to not do that falls both on the tool and the user. > > That's substantially different, because the spec doesn't give computable > criteria for deciding if a table is a layout table, so the requirement isn't > machine-checkable. To make alt similar, the spec would need to make the > presence of the alt syntax to depend on matters that aren't computable. The point remains that tool can't enforce all the requirements of the spec IMHO. / Jonas
Received on Tuesday, 18 August 2009 17:45:39 UTC