RE: User Agent Types

In a recent message, it was intimated that IE would be in the category of
technologies that rendered text as well as video. But, at last I think I
understand that IE does not, in fact, render video. An embedded object (e.g.
RealPlayer or Microsoft's player or Quicktime) plays the video.  However,
there is a very real problem that occurs here.  We want the user to be able
to use keyboard commands to control frame rate, to pause, etc.  We also want
the host program (such as IE) to have keyboard navigation commands.  If IE
doesn't know that the player is there, how does it know to pass keyboard
commands to the player?  Does there have to be some sort of command
hierarchy, ala HyperCard, where unused keyboard events are passed down (or
bubble up)?  Hmmm....

As I understood the general discussion at the F2F, any agent that renders
any type of material is responsible for making that material accessible.
Thus, a small screen display (such as my Palm III), when connected to the
web, might render text in a different fashion than a large screen display,
but since both are rendering text, both must make it accessible.  An audio
UA, whether designed for use by a blind user, or over a telephone, would
still have to make the text that is rendered accessible, as per our

Any agent that renders video would have to make that accessible.  If the
video were being rendered in an audio browser, it might be providing visual
descriptive text.  That would be rendering the video.  The only way to avoid
the responsibility for accessible content would be to ignore it entirely.

This suggests the general rule of thumb: If a UA makes a type of content
accessible to one group, it must, as far as possible, make it accessible to
all groups, either natively, or via third party AT.

Does this rule make sense to anyone other than me?

Denis Anson

Received on Wednesday, 30 December 1998 22:07:11 UTC