This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
Let the spec say that implementations MAY use heuristics to suppress announcing/presenting/making available the longdesc URL to the user. (But add that this does not apply when the content is text *and* the UA/AT makes use of the MAY option to present text to to the user.) MOTIVATION: * To avoid that users are encouraged to check obviously erroneous or pointless URLs. * To avoid that users’ trust in the longdesc attribute is affected negatively WHEN IT APPLIES * I shall not requeset that the spec describes the this in exact detail, but it seems to me that the circumstances that it would be useful to use heuristics in order to evaluate the longdesc, are roughly equivalent to the validation warnings described in bug 21501.
In general I think this is a bad idea, and in practice I think that unless it is extremely carefully phrased, it provides a way to justify not supporting longdesc: "The vast majority of longdescs are useless, so a valid heuristic is to say that if there is a longdesc it is useless and suppress it". Easy to implement, likely to be recommended by some strong and respected advocates of HTML, but I think highly counter-productive to accessibility in general and the purposes of this specification in particular. It might make sense to modify this proposal to provide an exception case for 21566, but IMHO only when it has been *determined* that the specific longdesc is useless. (There's still weasel-room with that proposal, but this spec is not going to solve all the problems of human behaviour, it will just tell people who want to get this right how to do so.)
(In reply to comment #1) > In general I think this is a bad idea, and in practice I think that unless > it is extremely carefully phrased, it provides a way to justify not > supporting longdesc: "The vast majority of longdescs are useless, so a valid > heuristic is to say that if there is a longdesc it is useless and suppress > it". Easy to implement, likely to be recommended by some strong and > respected advocates of HTML, but I think highly counter-productive to > accessibility in general and the purposes of this specification in > particular. > > It might make sense to modify this proposal to provide an exception case for > 21566, but IMHO only when it has been *determined* that the specific > longdesc is useless. (There's still weasel-room with that proposal, but this > spec is not going to solve all the problems of human behaviour, it will just > tell people who want to get this right how to do so.) I understand your thinking. But as for those 96% procent of 'useless', then not all of those are not necessarily useless. Useless and wrong are not the same, if UAs use the MAY option to present textual content. Also, I think that the critical voices will in general be more favorable if it is shown that longdesc is in fact something that users can rely on.
The specification will make no requirements on invalid longdescs