It's worth speculating how this would work in an alternate universe where authors cared more about usability and user agents made more intelligent use of HTML. Most of the usecases mentioned involve differentiating sets of links, and deciding how they should be treated on behalf of users. HTML includes many ways of classifying anchor links into groups (rel/rev, type, hreflang, class), such that users really ought to be able to make such decisions for themselves. In the mirrorverse, user agents would come with customizable behaviours for classifications such as:

rel="external"

rel="help"

type="application/pdf"

--
Benjamin Hawkes-Lewis

Reply via email to