RE: spaces and alt-text. proposal for the ERT.

I thought the consensus was that alt=" " (and even alt="  " [or more 
spaces]) is, strictly speaking, not valid html (because of the whole 
leading/trailing-spaces-may-be-ignored thing) but that this did not trip up 
any known browsers.

The consensus was that alt="  " (or any number of spaces) should probably 
be rendered the same way as alt=" ", but browser behavior on this point was 
variable.

I also thought the consensus was that alt=" " is NOT the same thing as 
alt="" no matter what the interpretation.

My version of Lynx (pretty new, but maybe not the very latest) treats 
alt="" VERY differently than alt=" ".

According to working draft of the WAI UA guidelines, alt="" SHOULD hide the 
image (and any associate link).  This IS what my version of Lynx does.

I don't think the difference between alt="" and alt=" " is author 
preference.  It is more an obscure technical issue.

We can "trump" the whole issue by deciding that alt="" should NEVER be 
allowed by the WCAG.  This is an overly zealous rule.  It has the virtue of 
being very simple.  We can leave the validity of alt=" " to the validation 
services (they don't currently report an error).

I don't think, in practice, you should flag for alt=" " since then you must 
also test for all permutations of     and spaces.  Certainly this 
kind of test can be put into an algorithm, but if an author thinks that 
their alt text is best represented by (some form of) spaces, who are we to 
question that?  We don't actually have a test for GOOD alt text (without 
human intervention anyway) except for the obvious things like alt="foo.gif 
(1234 bytes)".

My point is, if the author goes to the trouble of inserting alt text (even 
alt=" ") then they must have thought about it for a second or two.  This is 
what we are trying to promote!  There are no authoring products that 
generate alt=" " for you, although some do generate alt="" and stuff like 
alt="foo.gif (1234 bytes)".  For spacer gifs, alt=" " is logical, whereas 
alt="" is ALWAYS questionable.  It is not possible to tell if alt="" is by 
design or accident.  This is NOT true for alt=" ".

--Bruce


On Tuesday, December 07, 1999 6:14 PM, Wendy A Chisholm [SMTP:wendy@w3.org] 
wrote:
> At 01:02 PM 12/7/99 , Charles McCathieNevile wrote:
>> 3b. THere is no defined handling for alt=" ", so it might no have the 
desired
>> effect
>
> using "" versus " " has been argued back and forth.  It is my
> interpretation of sifting through the archives that if the "desired 
effect"
> is a space between two characters, using an image for this is
> inappropriate.  We suggest style sheets.
>
> Also, some people have favored "" while others like " ".  Most of the
> arguments have been based on the current state of the art handling of
> spaces.  However, it appears that "" and " " are currently handled the 
same
> [http://www.w3.org/WAI/GL/tests/spaces.html].  I was the only one to
> respond to the tests, so it only includes the 3 browsers I was able to 
test.
>
> Many people pointed to the HTML4 spec where it says that "leading or
> trailing spaces" may be ignored.  However, is a single space trailing or
> leading?
>
>     is inappropriate since it is intended to be used as a 
typesetting
> hint as Nir described on 17 November:
> http://lists.w3.org/Archives/Public/w3c-wai-gl/1999OctDec/0070.html
> Alan Flavell also makes a good point about nbsp not being considered 
"white
> space" 
http://lists.w3.org/Archives/Public/w3c-wai-gl/1999OctDec/0069.html
>
> Therefore, I don't think it matters much if we suggest that people use "" 

> or " ".  Personally, it seems to be a matter of author preference. 
 Neither
> one is going to cause horrible things to happen (although "" has
> historically - i.e. it supposedly used to be ignored by Lynx but doesn't
> seem to be anymore).
>
> If we look at the long term, I would suggest using a space since spaces 
are
> supposed to be preserved in XML.  However, I also agree that it is a 
matter
> of separating content from presentation.  A space seems to say, "there is 

> formatting going on here" while null says, "it's decorative."
>
> Again, I think it boils down to author preference.
>
> >3d. If the image is decorative, use alt="" but ask if the author would 
like
> >to associate a longdesc
> I think it is up to the author, thus why I left it as either a blank or a 

> short description.  It is my understanding that some users appreciate
> knowing what is on the page. If the author wants to support short
> descriptions for them, it is their choice.
>
> As we look forwards to tools that will allow users to navigate the
> structure of a page, if a user chooses to navigate past images then we
> won't have to worry as much about users wading through alt-text that they 

> don't care about.

Received on Tuesday, 7 December 1999 18:47:05 UTC