On 17/1/09 23:30, L. David Baron wrote:
On Saturday 2009-01-17 22:25 +0200, Henri Sivonen wrote:
The story of RDF is very different. Of the top four engines, only Gecko
has RDF functionality. It was implemented at a time when RDF was a young
W3C REC and stuff that were W3C RECs were implemented less critically
than nowadays.
Actually, the implementation was well underway *before* RDF was a
W3C REC, done by a team led by one of the designers of RDF. In
other words, it was in Gecko because there were RDF advocates at
Netscape (although advocating, I think, a somewhat different RDF
than the current RDF recommendations).
Yes, Netscape had this stuff when it was still called MCF. W3C's RDF
took ideas from several input activities, including MCF, Microsoft
XML-Data, PICS, and requirements from the Dublin Core community. But it
looks more like MCF than the others.
MCF was originally proposed by R.V.Guha at Apple; it followed him from
Apple to Netscape in 1997, and when the Mozilla sources were later
thrown over the wall, there was a lot of MCF in there.
MCF White Paper, 1996 http://www.guha.com/mcf/wp.html
spec, http://www.guha.com/mcf/mcf_spec.html
While this was at Apple, there was a product/viewer called HotSauce /
Project X, and some early grassroots adoption of MCF as a text format
for publishing website summaries.
http://web.archive.org/web/19961224042753/http://hotsauce.apple.com/
http://downlode.org/Etext/MCF/macworld_online.html
It was at this stage that dialog started with the Library scene and
Dublin Core folk, about how it related to their notion of catalogue
records, and to the evolving PICS labelling system, format and protocol
being built at W3C.
eg.
http://www.ssrc.hku.hk/tb-issues/TidBITS-355.html#lnk3
http://web.archive.org/web/19980215092626/http://www.ariadne.ac.uk/issue7/mcf/
The MCF/RSS relationship is a whole other story, eg. see
http://www.scripting.com/midas/mcf.html
http://www.scripting.com/frontier/siteMap.mcf
http://web.archive.org/web/19990222114619/http://www.xspace.net/hotsauce/sites.html
Then the thing moved to Netscape. Tim Bray helped Guha XMLize the spec,
which was submitted to W3C in 1997, where it joined the existing efforts
to extend PICS to include text labels and more structure -
http://www.w3.org/TR/NOTE-pics-ng-metadata
http://www.daml.org/committee/minutes/2000-12-07-RDF-design-rationale.ppt
http://searchenginewatch.com/2165291
So the June 97 spec was
http://www.w3.org/TR/NOTE-MCF-XML/
.. you can see from the figures that the technology was very RDF-shaped,
http://www.w3.org/TR/NOTE-MCF-XML/#sec2. Also a tutorial at
http://www.w3.org/TR/NOTE-MCF-XML/MCF-tutorial.html
Netscape press release accompanying June 13 1997 submission -
http://web.archive.org/web/20010308150737/http://cgi.netscape.com/newsref/pr/newsrelease432.html
Less than 4 months later, this came out as a W3C Working Draft called
"RDF": http://www.w3.org/TR/WD-rdf-syntax-971002/
... in a shape that didn't really change much subsequently. RDF wasn't
the same design exactly as MCF but the ancestry is clear enough.
And getting back to the original point, yeah Mozilla had MCF sitemaps
code in there.
Revisiting
http://www.prnewswire.com/cgi-bin/stories.pl?ACCT=104&STORY=/www/story/9-8-97/312711&EDATE=
http://www.irt.org/articles/js086/ and the like, it's clear that RDF was
very much a child of the 1st browser wars.
In retrospect the direction it took within Mozilla didn't do anyone much
good. The earliest MCF apps were about public data on the public Web,
feeds, sitemaps and so on. But eventually the ambition to be a complete
information hub led to MCF/RDF being used for pretty much everything
*inside* Mozilla. And I don't think that turned out very well.
http://www.mozilla.org/rdf/doc/api.html etc. The RDF vocabularies it
used were poorly or never documented (I have some guilt there) and when
Netscape went away, the incentive to connect to public data on the Web
seemed to drop (no more tie-ins with the 'what's related' annotation
server, 'dmoz' etc.). RDF drifted from being a Web data format to be
consumed *by* the browser, into an engineering tool to be used in the
construction *of* the browser, ie. as a datasource abstraction within
Mozilla APIs. While I can certainly see the value of having a unified
view of mail, news, sitemaps, and so on, the Moz code at the time wasn't
really in a position to match up to the language in the press releases.
Not making any particular point here beyond connecting up to the MCF
heritage...
cheers,
Dan
--
http://danbri.org/