- From: Al Gilman <asgilman@iamdigex.net>
- Date: Sat, 05 Jan 2002 14:17:37 -0500
- To: "Lofton Henderson" <lofton@rockynet.com>
- Cc: "Lynne Rosenthal" <lynne.rosenthal@nist.gov>, www-qa@w3.org
At 12:13 PM 2002-01-05 , Lofton Henderson wrote: >I wrote: > >"[...]one browser (IE 5.5) takes me to the top of the page, the other (NN >4.73) takes me to the bottom (thereby entirely missing the #docs-anchored >section)." > >To be precise, it is not "top of the page", but rather "correctly to the >anchored section" (which occurs near the start of the page, and which does >end up correctly positioned at the top of the screen.) > >-Lofton. > Great! Now we're truckin'. IE 5.5 does the right thing, but there are Netscape 4 versions out there that do rather unfortunate things. Hmm. Likewise #fragment references are routinely broken in the browser handling of redirects <http://www.w3.org/TR/cuap#uri>. So, in the light of the widespread use of browsers where the implementation of named-anchor link destinations is broken, we should think about a fail-soft form of citation. On the other hand, working on any W3C group is pretty dependent on being able to follow references into specifications and to threads in the email archives. For this I don't think there is a reasonable alternative to demanding that people use a browser with competent #fragment handling. Within the WAI I am possibly on the "coddle legacy" side of the median in the distribution of opinion. But being able to follow a reference to a point within a document is such an old, and IMHO essential, browser capability that I would probably push back some on this one. Not to say that there is any quick answer, but there may be different answers for Internet and Intranet, shall we say. Correspondence with the public and correspondence among members of a persistent and intentional group. Even for an Interest Group, I wonder whether we actually have anyone who suffers a real hardship in upgrading to some browser that does this much right. But the whole issue of how to handle discrepancies that are out there in the deployed stuff in the field is good material for QA to chew on. I will be interested to learn what evolves within the group on this subject. I confess I did not realize that any User Agents were broken worse than retreating to opening the page from the beginning when they can't cope with a #fragment. It sounds like a possible author-side workaround is to make sure that a citation includes, in its rendered text content, the exact text of the section title or a unique string starting the cited section when pointing into a document. Sigh. Under the immediate circumstances, it is possibly better to follow the usual practice and give the URIs for the documents themselves in a message inviting comment. Each of these documents will have references up and out to the context in the activity and document bundle, I trust. There is a sort of Call for Review template that is traditional, and basing informal invitations on that template is probably a good way to avoid trouble. Is QA doing anything to collect and publish what it knows of these infelicities in deployed implementations? This one didn't make the CUAP. Al
Received on Saturday, 5 January 2002 14:18:20 UTC