2000 01 31 ER telecon

Summary of action items and resolutions



"Crying wolf"

WL think it is easy to understand.

MC but we don't want to use issues that turn people off.

BM we basically define before saying that

WL "generate excessive warnings."

WC don't think it will internationalize well.

resolved: "crying wolf" is ok for now. will probably be cleaned up when technical editor goes through.

Author indications that a warning has been checked

CR a useful thing for us to have in a-prompt. mark as already been checked.

WC meta data stored in the document?

MC local file. implementation thing.

WC this for all checks?

CR not sure possible or desirable. if they add stuff to an HTML file people won't like it.

BM repair is expected ot modify.

WC store as RDF file, only add one line to the repaired document.

WL is this a technique for this doc.

CR is we checked a file, it had an image, not complex, don't need a longdesc. next time, check this file and would not put up the warning. if they have already said "not complex."

MC it can make a page more difficult to work with. like external meta data approach.

HB it would be useful to convey to analysis tools and site validators. "don't expect a d-link with this one, author has consciously defined."

MC one section rather than across the elements.

WL 'don't bug me again"

HB don't know how I would implement in the header?

WC not in header in RDF file that is separate

HB should be in the same file.

MC future issue. need XHTML and namespaces. We may want to define a set of attributes in a namespace that authors can use for this purpose.

@@WC take idea of using RDF to prevent excessive author warnings to CG.

Checking for a complex image

CR if not bullet or horizontal rule, do for everything else. size restriction?

WL asking for every picture for a londesc?

MC in bobby we do right now. talk about cry wolf, that's one thing that people hate.

HB longdesc = "" be a placeholder for author to say, "i don't need one." implied URI as value.

WC but that's only for markup aware people.

WL so to authoring tool issue.

MC for bobby this is a question of "does this need a description?" rather than how it is associated with it. if a longdesc exists for it, we don't pay attention. "longdesc" is shortcut for bobby, but don't ask the question in regards to "londesc".

WC how about ask at end, see one longdesc, how about convert to d-link?

CR going backwards.

WC actually standing in place.

@@WL check ATAG techniques doc for checking for longdesc/complex images techniques.

Combine similar techniques?

CR IMG element in one case, INPUT in another. in one case, may be able to have alt=" " other can't.

MC agree with that.

WL generally can't hurt no great gain made in combining.

resolved: leave 1.1.A and 1.1.C as is - i.e. don't combine.

Check APPLET for "alt" always needed?

WC "alt" required on APPLET as with IMG in HTML4?

MC will trigger HTML 4 validator.

WC applet is deprecated so would surprise me if required.

HB applet is there. required width and height, alt implied - optional.

WC text w/in applet attribute, enough?

WL no, should be required.

WC deprecated in favor of OBJECT.

HB last of objects should be text.

CR text w/in APPLET similar to londesc? so alt-text required.

HB we'd like to have it required. we aren't in position to change DTD.

CR not the HTML spec, but should prompt for alt-text regardless of finding text in APPLET.

resolved. yes, we should prompt for alt even if text in APPLET

Check APPLET and OBJECT for any HTML element

CR link to text only page.

HB many java objects generate material, static description be imperfect.

CR might want more than text description.

@@WC ask len to clarify his question about applet and object for any HTML document.

FRAME "title" check not automatic

CR assume if have frame title that it describes.

MC ask for every frame, also need a "longdesc." again, cry wolf.

WC only way is to verify with author.

CR we're gonna have to ask the author.

WC how to differentiate to the author between title and longdesc without using "title and longdesc."

CR title short.

WC longdesc be description of the frame

CR need for toc?

WC no

CR number of frames used? two frames, probably not. 6 frames. 1- logo, 2 ...etc.

HB all frames pointing to same longdesc.

WC can't think of a good example.

MC since they are both external files, if the longdesc is going to benefit have a browser that can get the frame any way.

WC three frame. top is navbar, left is navbar, main is content.

CR trigger on as least often as possible. doesn't seem to be needed.

HB name, title, longdesc all triggers.

MC not "name" is programmatic.

HB name is fallback?

MC name is what lynx uses.

resolved: len's right, title is not an automatic check. prompt for longdesc on frame if there are 3 or more. suggest longdesc is same for all 3 frames since relation among them.

CR longdesc describes relationship between frames.

HB same frame referenced from other pages.

WC longdesc really should have been on frameset rather than frame...

$Date: 2000/11/08 08:17:43 $ Wendy Chisholm