TAG Rejects HLink
TAG, the W3C's Technical Architecture Group, announced recently that XLink "should be used for hypertext references in user-interface oriented applications", and that "it is the unanimous opinion of the TAG that XLink should be used for hypertext references in XHTML 2.0." The announcement seemed to directly repudiate HLink, a new linking solution for XHTML.
Steven Pemberton, chair of the HTML WG, suggested that HLink and XLink were two ways of doing the same thing: "The idea is that you can define linking on markup languages using XLink concepts, and you could even define XLink itself using HLink. [HLink] is not a divergence from XLink, but an enrichment..."
The TAG suggested, further, that, "given that MathML and SVG already use XLink for hypertext references, that would seem to be precedent for using XLink in XHTML." To which Pemberton responded: the "fact that SMIL doesn't use XLink would seem to be a precedent for not using XLink in XHTML."
In response to TAG's claim that creating HLink falls outside XHTML 2.0's charter -- one "design goal" goal of which is "to use generic XML technologies as much as possible" -- Pemberton says that "'as much as possible' was added exactly to cover XLink, because it prevents us from doing the things we want to be able to do."
The first public XHTML 2.0 Working Draft, released in early August, explicitly refused to use XLink to provide hypertext linking. That refusal was foreshadowed as early as the XLink Final Call when members of the HTML WG suggested that XLink had failed to meet some of its requirements, particularly those most related to HTML linking semantics.
The TAG's repudiation touched off a wide-ranging conversation, including both W3C-procedural and purely technical issues, in the XML developer community. The publication of TAG's finding followed very quickly on the heels of the HTML Working Group's public release of HLink.
Says Tim Bray, member of TAG, "...I agree that consensus developed quicker than one might expect...the HLink draft had been kicking around for quite some time albeit never in a publicly-visible place, so the people on the TAG have had more time to think about it than the general public."
Questions of W3C and TAG procedure were at the forefront of the ensuing community discussion. One XML developer said that "the problem is that there are some people who feel that the process was violated. That set a bad tone...The TAG message didn't come across...as an invitation to feedback and discussion, but [as] a pronouncement ex cathedra."
Shane McCarron, one of the editors of XHTML 2.0, said that the TAG's repudiation of HLink went "far beyond the scope of the TAG's brief." Ann Navarro, another XHTML 2.0 editor, said that the TAG's insistence on XLink "is inappropriate in the face of explicit direction by the governing body of the W3C -- the TAG...does not override AC decisions. This 'suggestion' certainly seems to attempt to do so."
The status of the TAG's repudiation of HLink is unclear. The announcement seemed to arise out of the context of the existing scope of XLink issue, which was raised by Tim Berners-Lee. But the "resolution summary" of the xlink-scope issue, though still labeled a "draft", says that its own status is a "personal view only", presumably of its author, Tim Berners-Lee.
Subsequent comments about TAG's announcement from TAG members added to the confusion surrounding the status of its repudiation of HLink. Says Norm Walsh, a member of TAG, "The TAG was asked to form an opinion about an issue. We did so. We talked long and hard and with considerable agony about the issue. And...we reached a position where we all agreed. I accepted the action to write a message telling the world what that opinion was."
Walsh further adds, "The opinion of the TAG is just that, an opinion. I don't feel (and I don't believe anyone on the TAG feels) that our opinion commands anyone to do anything. We don't have any authority to tell anyone what to do."
At least two members of TAG recognize that it is difficult, if not impossible for unanimous, publicly disseminated TAG decisions to be taken as "opinion" only. According to the minutes of a 9 September TAG teleconference, Paul Cotton, a TAG member, said that "any time we say something it's being taken as gospel and either we have believers or disbelievers. When the [TAG] speaks, people notice." Another TAG member, David Orchard, added, "This is good, but it means we have to think about the impact of our decisions."
Mike Champion suggested that part of the problem was that the TAG's repudiation of HLink was understood as a formal statement of the TAG's authority under its charter. Said Champion: "many readers...concluded that TAG decided to issue an Architectural Recommendation to the effect that there are too many linking technologies out here already, and we should all learn to stop worrying and love XLink."
But that interpretation of the TAG's repudiation is one which Tim Bray warns against: "We developed an opinion and published it. I'm not sure where things go from here, process-wise, but we thought this over, we read the documents, and we developed an opinion, and we published it...At the moment, the TAG is unanimously on the record as disagreeing, from the point of view of Web Architecture, with the HTML WG's intent to discard XLink. Either we are wildly unrepresentative of the rest of the world, or there's a real issue here."
Speaking on behalf of the HTML WG, Steven Pemberton said, in response to the TAG's repudiation of HLink, that "it is not clear you have the authority to say that." The HTML WG formulated HLink, Pemberton suggested, because the expected formation of a "rechartered Linking Work Group" never materialized.
Also in XML-Deviant
No standard is useful without software which implements it. Tim Bray wondered whether HLink is "any more likely to be implemented by browser vendors than XLink has been? I'm pessimistic about the chances of either of them until we get some competition and innovation back in the browser market."
Navarro, however, suggested that the HTML WG concerns which motivated HLink are connected with the needs of both the user community and vendors: "...we have representatives from the 3 largest market-share browser vendors on our working group, and we've needed to argue this point and produce HLink...since those vendors are present, assume what you will from those design goals."