J
Jukka K. Korpela
jake said:I was reviewing somebody's site on one of the other
groups and noticed that they had the longdesc as a literal.
This just causes a '404' error if anyone tries to follow the link that
gets generated... at least in HPR.
Quite naturally. As an almost purely theoretical point, if one wants to
embed the description into the attribute, it would be formally correct to
do so using a data: URL, e.g.,
longdesc="data:,Hello%20world"
(I wonder whether HPR supports that.)
For an incorrect attribute like longdesc="Hello world", a browser _might_
do some guesswork and decide first that it's incorrect (a URL must not
contain a naked space) and then, as error recovery, present the text to
the user as if it were the content of a document pointed to by the
attribute. But this would be far-fetched and could result in worse than
meaningless data presented to the user.
The longdesc attribute is virtually useless, or a little worse: people
who care about accessibility can spend their time writing such attributes
instead of doing something productive, like including an explicit link to
an explanation when needed.
I wonder where that idea is comes from ?
Originally from just sloppy reading of the spec, I guess. The idea has
survived because most people that put longdesc attributes onto their
pages _never_ see them in action, or see them e.g. in a popup window on
Mozilla when they right-click on the image and select "Properties".