The addendum claims that input devices "shouldn't be visible in
the markup" -- a position that would require the user of a web-based
OCR application to select a scanner over a camera for each page of
text, while a user of a teleconferencing application would need to
select a camera over a scanner for each photo.
Actually, it's worse than that. If the HTML working group chair's
suggestion of using <input type="file" name=... accept=[mime types]>
was used for both device input and filestore file upload, then
the browser operator would have to choose between the filesystem
and all the appropriate devices each time. That could get tedious,
and impede applications (such as conferencing or spoken language
instruction) that would go much smoother if a default were specified.
Does anyone think that suggestion is good user interface design?
Is there any doubt that the W3C specifications are omitting the
proper solution for such applications? It's no wonder that the
browser vendors haven't followed the spec that way for device upload.
Cheers,
James