W3C home > Mailing lists > Public > public-html@w3.org > January 2010

Re: ISSUE-27: rel-ownership Change Proposal discussion

From: Roy T. Fielding <fielding@gbiv.com>
Date: Thu, 21 Jan 2010 18:14:52 -0800
Cc: Mark Nottingham <mnot@mnot.net>, HTMLWG WG <public-html@w3.org>
Message-Id: <1C674B84-8A80-432D-92A3-1BF7CF70BA99@gbiv.com>
To: Tantek Çelik <tantek@cs.stanford.edu>
On Jan 20, 2010, at 8:17 PM, Tantek Çelik wrote:

> On Wed, Jan 20, 2010 at 7:19 PM, Mark Nottingham <mnot@mnot.net> wrote:
>> On 21/01/2010, at 2:10 PM, Tantek Celik wrote:
>>> While I admire a lot of the good work that has gone into "Web Linking", I must point out that in the decade+ since HTML4 (or 4.01), wiki-based methods of researching, exploring, brainstorming, proposing, testing, implementing, and iterating on rel values has actually served the web design and development communities reasonably well, with a good variety of rel values being published and interoperably implemented, whereas in that same decade+ time period, IETF/W3C-based processes have been largely absent or stagnant in terms of rel value development (with a few notable exceptions like Atom).
>>> Thus I do think in general we have a decade+ of experience that for this particular space of standards/extensions, community wiki-based methods have been superior (in direct contradiction of the assertion that "may not be a workable solution in the long run" - 10+ years is not a bad "long run" frankly), and I would hate to lose that moving forward.
>> Not to quibble, but you quote ten years,
> Yes, 10+ years that rel has been an extensible by *anyone* attribute
> in HTML4.x per the spec - that's merely a statement of historical
> fact.

Umm, it is a historical fact that rel has been extensible by anyone for
as long as it has existed, including when I redefined it in the HTML 2.0
DTD back in 1994.

>> whereas WHOIS says microformats.org was registered in 2005 -- the same year that Atom's link relation registry, which Web Linking is based upon, was created. When did you actually start using a wiki? Was there one somewhere else?
> A bit more of the history of adoption of various rel values and their
> development on the web (in contrast to email lists).
> There was very little widespread use of the rel attribute among web
> designers for anything other than rel="stylesheet" (and sometimes
> rel="alternate" for feeds) for many years after HTML4.x became REC.

Web designers are a subset of HTML users, let alone Web users.

> 2002: I first proposed (in a blog post on the web) that blogs should
> use rel="bookmark" (from HTML4) to more semantically markup existing
> blog post permalinks.
> http://tantek.com/log/2002/11.html#L20021128t1352 - soon thereafter it
> became widespread practice.

IIRC, Dave Raggett proposed the same thing in 1993 for HTML+.
I think it is great that you actually convinced current browser
developers to implement it.

> 2003: March, at SXSW I proposed marking up existing blogroll links to
> friends (existing widespread practice on the web) with rel="friend" to
> express that semantic. By December, Eric Meyer, Matt Mullenweg, and I
> proposed XFN 1.0 on a *website* (not an email list), and the XMDP
> profile format for formally defining them, which quickly became the
> dominant use of profile attribute. Again, the result was rapid
> widespread adoption/practice on blogs and personal sites. The
> announcement http://tantek.com/log/2003/12.html#L20031215 discussion
> and iteration continued collaboratively on blogs across the web, not
> email lists.  Perhaps not wiki-based development per se, but certainly
> *web-based* development.
> 2004: I invented and proposed rel="license" for more semantically
> expressing links to Creative Commons (and other) licenses.  Initially
> in a presentation at ETech (posted on my site)
> http://tantek.com/presentations/2004etech/realworldsemanticspres.html
> , and on the ETech *wiki*
> http://wiki.oreillynet.com/etech/index.cgi?RealWorldSemantics and then
> more formally documented on the (unfortunately now defunct) Technorati
> Developer's wiki:
> http://web.archive.org/web/20050321015449/http://developers.technorati.com/wiki/RelLicense
> Again, *web-based* (and mostly/eventually *wiki-based*)
> discussion/iteration, not email-list-based.

That explains why copyright was duplicated.

> 2005: January, rel-tag proposed/developed on the Technorati
> Developer's wiki
> http://web.archive.org/web/20050225023534/http://developers.technorati.com/wiki/RelTag
> . In June, all rel values and other microformats developed on the
> Technorati Developer's wiki were contributed to microformats.org,
> where they have been developed since, primarily on a wiki.

I think that's great.  It is not, however, a link relation registry
for the entire Web.  There is nothing wrong with using a wiki to
develop content that is submitted to IANA for formal registration,
where other people (not in your community) can find it.

>> From then on, numerous other rel-values have been
> researched/brainstormed/proposed/tested/experimented/iterated on the
> microformats.org wiki.
> A few more historical details/links: http://microformats.org/wiki/history
> Most comprehensive documentation of rel values on the web:
> http://microformats.org/wiki/existing-rel-values

And yet it is missing most of the pre-2000 relations.  Pity.

>>> I'm not saying that wiki-based methods have been perfect by any means, nor am I saying that they couldn't be improved. However, they've been the best we have seen *in practice*, and that's worth a lot IMHO.
>>> However, as I said, a lot of the good work has gone into "Web Linking", and would like to see the improvements/changes proposed there pursued  within the communities that have been carrying rel forward in the past 10+ years (primarily microformats.org), so that they can be evaluated/critiqued/iterated by those who have been working on them to date, and incorporated into the specs there as well.
>> According to my mailbox, you've been aware of the draft for more than a year (at the least; it's been around since 2006). Do you have additional feedback?
> I've given up on keeping up with innumerable different mailing lists
> for innumerable different standards. Email doesn't scale. Wikis do.
> http://microformats.org/wiki/wiki-better-than-email
> My additional feedback is the same as above: please propose your
> improvements/changes to rel values in the microformats community where
> those rel values have been largely developed, discussed, and
> documented, especially with respect to their use in *HTML* (in
> contrast to Atom or other formats) - which is of course, what this
> whole issue is about.

I like microformats, for obvious reasons, but I think you should
understand that the world does not consider the microformats wiki
to be a repository of standards for the Web as a whole.

Link relations are format-independent.  The folks who develop PDF
and SVG have just as much ownership of the value set as HTML or
microformats.  It is not reasonable to expect the rest of the world
to come to your doorstep for the definitions of a common standard,
particularly since you don't seem interested in listing the
relations used outside the microformat community.  Therefore,
we have lots of community-specific contributions, like those on
the microformat wiki, and one Internet registry to act as the
referral point from all names to the specific definitions in
those community spaces.

Received on Friday, 22 January 2010 02:15:22 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 10:15:57 UTC