Images/ALT

I think we need to do two things regarding the Images section in the
Page Author Guidelines:

- separate image/ALT issues from image-map/object
- move forward regarding ALT text value.

The section is too large as it and it doesn't really talk about the
issues with ALT value. Plus the sensitive map issues are specific.

Regarding the second point, I'd like to get a clear picture of the
respective roles of ALT/TITLE/A-TITLE and LONGDESC.

I know it's a hot topic and I'm not trying to play 100% consensus
here (after all I'm not the chair of this group :-)

I suggest the following:

- IMG ALT is a short textual description of the image (context free)
- IMG TITLE is a short textual description of the image _function_
- A TITLE (with or without IMG inside) is a description of its HREF target
- IM LONGDESC is a long textual description of the image

Then there are a lot of "applications" of these base rules (and I
think the guidelines should be as detailed as possible).

- IMG used as decoration-only gets a TITLE = ""
- IMG used as decoration-only gets a ALT = "deco" or
  ALT=<real description>, authors is free to choose.
- IMG used as LI gets a TITLE = "bullet", or maybe "*" or "item", we
  need to pick up one. I vote for "*".
- "Drawing of, Photo of, Cartoon of.." etc should be omitted for small
  images, especially when there is a group of such images. 
- Image of text should get ALT = text (but better be done using CSS)
- Do not recommend for specific punctuation or use of [], let browser
  add value
- Do not recommend minimum image size (fix browsers)
- Mixed graphic/text image gets both description and text (text if
  usually function that goes in TITLE)
- For set of images used in place of image maps: use Len Kasday
  guidelines of sequenced ALT "overall desc..." "desc..." "last"


My main point is that the function of the image should go in the image
TITLE, not its ALT.

I think it's better to make the ALT itself context-free, as it allows
for things like embedded desc in the image format or semi-auto filling
ALT tools (where finding the context might be hard).

The Browser guidelines will need to say in details how to use TITLE
for tooltip, ALT and TITLE together, etc, and how to default things
when some attributes are missing (for instance, in the context of A,
with no TITLE nor ALT, a browser could do pre-fetch the pointed
document TITLE and use that). I suspect the Browser configuration
aspects are going to play a big role.

Received on Friday, 23 January 1998 09:24:55 UTC