Re: 1st draft of reference card

 I should have explained

1 this draft of a reference card is an attempt to clarify my thinking as to
what might be appropriate and I posted it to the list as an outcome from
today's EO IG conference call and as a follow up to Friday 24th discussion
in Peterborough

2 my references to the formatting apply solely to the text as it appears in
the email - I was concerned that I wouldn't have removed all of the section
breaks, fonts, styles etc. that I had used in my own working version - and
that the lack of formatting styles would make it difficult to navigate.

>GF:  I'd leave out the five-word limit, since someone might take this as a
>browser >limitation.   Instead say something like, "...write a concise
>description of a few words >or a short sentence...," which offers more
>flexibility.

In response to GF's comment I agree, but I put in the 5 word limit to cope
with situations where the author has specified the height and width of an
image placeholder in such a way as to overlap the space for the alt text
and obscure some of the text. Any strong feelings on whether or not this is
an obselete or unnecessary consideration?

>GF:  In the GL group, there's a l-o-n-g debate raging about LONGDESC vs
>the d-link.  >Before recommending the d-link, I'd review the guidelines
>and arguments for both >(available at w3.org/wai/gl).  Personally, I think
>both should be supported, since >pre-5.0 browsers won't handle LONGDESC.

 I opted for d-link here as a current and backwards compatibility
compromise - I don't think there is room to include both in a reference
card / flyer. I consulted the PA GL which were used in Peterborough on
Saturday 25th July, and they recommended that "When most browsers in use
support longdesc use longdesc. Until then use d-links [Priority 1]. ... If
you use both "longdesc" and d-link or if you use "rel" to link the image
and d-link, a tool under development will be able to convert d-links to
"longdesc" and / or OBJECTS as well as remove d-links automatically is
desired". So, on the 80 20 principle, I opted for d-link and "rel" as it
seems most relevant now - although it can be updated to longdesc as
appropriate when it is widely supported. The PA GL reference to the
automated tool for converting d-link and "rel" seemed to offer the promise
of a currently working technique which would not represent an expensive
maintenance overhead or require extensive retro-fitting.

>GF:  Actually, it's best to include the e-mail address *in* the link
>itself, like this:

Agreed, Geoff. I'll update this version.



Best wishes - Stella

Received on Thursday, 30 July 1998 17:30:52 UTC