[html4all] Omitting alt Attribute for Critical Content Wiki page

[Joshue O Connor]

Hi all,

This is a really interesting discussion. I also agree with Phil that an
algorithmic proposal for large amounts of content is an interesting one.
It could certainly be helped by parsing available metadata provided by
the author. Use of generic attributes like labelledby and describedby,
seem to me, like a way of future proofing the solution. This has been on
my mind as I don't want to argue for a solution that may not be ideal
and only solves a short term problem.

Gez's point about where responsibility lies is also noteworthy as this
is something that is easy to misunderstand. For example, if the user
agent when it encounters missing alt text and defaults to heuristics,
had a better algorithm, and authors used more context sensitive and
relevant file names then this /could/ result in better screen reader
output and user comprehension. This is of course a big problem for CMS's
or large web applications that dynamically generate file names etc, but
it's not IMO off the wall to suggest that the way this is done could be
improved.

So if the repair method could be also improved, along with user agent
heuristics then the markup language. in principle, does not have to go
down a wrong road which could itself cause greater problems in the
future. I am not saying this /is/ the case but wish to point out that we
need to be aware of describing solutions to a short term problem that
could have a more negative impact in the long run.

So on the foot of that. It's very interesting to me to try and
understand the limitations of using a namespaced solution only to this
issue.

Josh