[html5] That alt tag!

[Jukka K. Korpela]

2013-06-09 15:48, Bob McClelland wrote:
> Is 'alt' in or out, for validation?

You should always have a useful alt attribute for each img element, to 
make your pages more accessible. This is far more important than the 
position taken in any HTML specification or validator.

>   The validator always used to complain if I didn't have one, then 
> recently it stopped doing (And I read reports that "I didn't need to 
> use an alt tag any more)  and today it's complaining again!

The  WHATWG HTML Living Standard is a document that may and will be 
modified daily, or almost daily, without prior notification. The related 
validators, in addition to validating against some mutable document like 
that, are all described as experimental and may also change at any moment.

> All this chopping and changing drives me potty and gives rise to two 
> questions- 1) What is the current position regarding alt tags,

Note that it's an attribute (alt=...), not a tag (<alt>).

Any answer to your question may be outdated at any moment, even between 
the moments that someone writes an answer and someone else reads it, so 
a specific answer is not particularly interesting. You can check today's 
position in WHATWG Living HTML at
http://www.whatwg.org/specs/web-apps/current-work/multipage/embedded-content-1.html#alt
(I suppose the structure will be retained even though the content 
changes, so this URL should work at least for a long time.)

The changes are really related to a rather detailed issue, namely HTML 
generators (such as software that generates an HTML image gallery from 
user-submitted images), which may have no information about the intended 
meaning and role of an image. Some people think that generators should 
be allowed to omit alt attributes, in order to prevent them from 
emitting some nonsensical or plain wrong alt attributes. Others think 
that a good HTML document always contains alt attributes for images, and 
if it is sometimes impossible to produce good HTML, then that's not a 
reason for changing the rules.

Moreover, authors of validators may have their own ideas and preferences 
and they may choose to implement checks that differ from the specs, 
thinking that it's the specs that need to be fixed. They might also be 
just waiting to see how some debate over a feature will end, then fix 
the validator accordinglt.

> and 2) what's the best way to be informed whenever a change (to any 
> part of the whole html5 spec) is implemented?

It depends on what you mean by "the whole html5 spec". In a broad sense, 
HTML5 is defined in over a hundred documents, almost all of which are 
work in progress, not finalized specifications. And WHATWG does not call 
its HTML "HTML5" any more.

At the start of
http://www.whatwg.org/specs/web-apps/current-work/multipage/index.html
you will find some links related to version differences and change 
notifications, under "Version history". But since the notifications are 
frequent and rather technical, I think most people won't find them 
useful unless they actively participate in the development of WHATWG 
HTML Living Standard.

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/