Subject: RE: A question about Web history and vision To: "Henry S. Thompson" , "John A. Kunze" , "Roy T. Fielding" Date: Mon, 24 Nov 2008 09:40:04 -0800 Of course, my perspective on this is personal. The fundamental notions of network nodes as resources which have names, addresses and routes to them were firm principles of the Internet architecture, and networking in general, since John Shoch's work (his PhD thesis, I believe) and predate the web significantly. The distinction between the abstract document and its representation in a network transaction was a firm principle of System 33 (developed in the late 80s, with the ideas primarily developed by Mark Weiser). Tim visited PARC in the early 90s at my invitation, and we spent a lot of time talking about the principles and how they might apply to WWW, including Steve Putz' System 33 gateway code http://www.w3.org/Gateways/System33/gateway). I've always thought that the ideas of content negotiation and the separation of the abstract 'information resource' from its concrete representation in a file format made their way into HTTP as a result of that visit, and that the elaboration of the idea that a URI might point to a service rather than any 'document' at all was also first demonstrated by Steve Putz in the PARC map browser. Tim might have a different memory of this, and in the end, I would rely on him as the ultimate authority, since "Weaving the Web" and its possible update or elaboration should be (or at least become) the definitive history. I suggest you forward this conversation to him for his feedback. Larry ---------------------------- From: Larry Masinter Subject: RE: A question about Web history and vision To: "Henry S. Thompson" , "John A. Kunze" , "Roy T. Fielding" Date: Mon, 24 Nov 2008 09:56:40 -0800 I should note a couple of players also who I didn't mention before: the extension of URIs to identify abstractions that were neither documents nor even network accessible entities seemed to me to have been promoted by a few other players. I remember arguing with Michael Mealling in his insistence that a URI could identify *anything* (and not just a communication endpoint) and finally giving in, since I believe he was representing the viewpoint of a significant community and not just his personal opinion. I don't know the origins of that opinion, but folks who might have contributed included Leslie Daigle, Karen Sollins, and Ron Daniel. Regards, Larry ------------------------------ From: "Roy T. Fielding" Subject: Re: A question about Web history and vision To: "Henry S. Thompson" Cc: John Kunze , Larry Masinter Date: Mon, 24 Nov 2008 14:55:16 -0800 On Nov 24, 2008, at 9:40 AM, Larry Masinter wrote: > Of course, my perspective on this is personal. Likewise. > The fundamental notions of network nodes as resources which have > names, addresses and routes to them were firm principles of the > Internet architecture, and networking in general, since John > Shoch's work (his PhD thesis, I believe) and predate the > web significantly. That was probably in the IETF background, though I don't think I ever read it. The word "resource" is also used in Tim's ENRAP paper in a general way for things to be found on the network http://www.w3.org/History/1992/ENRAP/Article_9202.pdf > You would have to ask TimBL why WWW address, as in http://1997.webhistory.org/www.lists/www-talk.1992/0000.html was replaced by Universal Document Identifiers http://1997.webhistory.org/www.lists/www-talk.1992/0025.html http://1997.webhistory.org/www.lists/www-talk.1992/0125.html and later by Uniform Resource Identifiers. http://1997.webhistory.org/www.lists/www-talk.1992/0246.html http://1997.webhistory.org/www.lists/www-talk.1992/0270.html Maybe this will lead to a clue http://www.w3.org/Conferences/IETF92/JillFoster.html http://www.w3.org/Conferences/IETF92/WWX_BOF.html http://www.w3.org/Conferences/IETF92/LivingDocuments.html and perhaps Larry can recall if Document -> Resource was discussed at that meeting. System33 was. Oh, here it is ... in the UDI BOF minutes: http://mirror.switch.ch/ftp/doc/ietf/92jul/udi-minutes-92jul.txt The URI and IIIR working groups were created after that meeting: http://www.ripe.net/ripe/maillists/archives/ripe-list/1992/ msg00116.html > The distinction between the abstract document and its > representation in a network transaction was a firm principle > of System 33 (developed in the late 80s, with the ideas > primarily developed by Mark Weiser). Tim visited PARC > in the early 90s at my invitation, and we spent a lot of > time talking about the principles and how they might apply > to WWW, including Steve Putz' System 33 gateway code > http://www.w3.org/Gateways/System33/gateway). Cool, though oddly enough that is the first I've heard of System33. Were there any published papers? Hmm, bad link on Larry's page to http://www.xerox.com/PARC/dlbx/other-papers/system33.ps > I've always thought that the ideas of content negotiation > and the separation of the abstract 'information resource' > from its concrete representation in a file format > made their way into HTTP as a result of that visit, > and that the elaboration of the idea that a URI might > point to a service rather than any 'document' at all > was also first demonstrated by Steve Putz in the > PARC map browser. I think those ideas were present from the very first discussion about UDI in 1992. However, I do know that none of the docs talked about resources as a service until the URI working group got under way, and even then "resource" was a term for what you could identify. It was assumed that documents are resources and that the Web would only retrieve documents. > -----Original Message----- > From: Henry S. Thompson [mailto:ht@inf.ed.ac.uk] > Sent: Monday, November 24, 2008 4:55 AM > To: John A. Kunze; Larry Masinter; Roy T. Fielding > Subject: A question about Web history and vision > > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > As you all know, I sit on the W3C TAG, where among other things I try > to make sense of this URI thing we all care about. > > In writing a layman's tutorial introduction to the subject, a question > has arisen I don't know the answer to, and a trawl through the > history, as recorded in various RFCs, only deepens the mystery. As > authors of the RFCs involved (1736, 1737, 2396), I would welcome your > insight into the following: > > When and why did the 'Resource' of URL/N/C/I become understood as > distinct from retrievable representation? In particular, when did > the possible of in-principle non-representable resources get > included in the range of URIs? Well, that's actually a different question than when identifiers became associated with resources instead of documents (one class of resource). The distinction between resource and representation was something that I added to the definitions. It came out of my early work on MOMspider, where I explored the ways in which hypertext links could be considered "broken" by loss of access or change of content. The observations demonstrated that people were not identifying documents, but rather the conceptual mapping over time -- what TimBL later called the "sameness" of resources. I probably wrote something about that in the URI WG discussions, but I don't have an archive of those any more. When Henrik and I were working on RFC1945, I used the notion of representations in our whiteboard discussions/thought experiments for the HTTP object model (what is now called REST). The distinction was already in the software; it was just a different way of thinking about HTTP interactions so that the caching model would be independent of resource type. I later defined resource in terms of that model for RFC2396. Regarding 1630 vs 1736/1737/1738 confusion: that was a combination of territorial politics and nice folks in committees trying to reach a compromise on what should have been a matter of taste (too much painting the shed). The result was something that nobody was happy with and doesn't reflect any implementation. Even 1630 was a strange hybrid of WWW addresses redefined using the artificial terminology of the URI working group. > ... > Four years later, in 2396, we have the fully articulated position -- > anything can be a resource, and representations are what is retrieved, > not resources. Resources whose representation is time-varying are > explicitly acknowledged (I'm guessing this comes from Roy's interest > in this case as discussed in his PhD). Yep. > So, any memories or reconstructions of rationales would be welcome of > why the generalisation happened from, shall we say, network-accessible > data objects, to anything with identity. > > [John, given the extract above, the particular question for you is how > or why the idea of resources which "have no electronic instantiation" > emerged. Why would you coin a network 'locator' for something which > by its nature could not have a network location?] 'locator' was a term created by the IETF discussions. I sometimes wonder what would have happened if TimBL had started with something like my definition of resources/representations and took that to the IETF. It is possible that the abstraction would have obviated the perceived need for the URL/URN split. OTOH, the split may have been inevitable, since the folks interested in persistent names were working on solutions before UDI was even proposed and were already vested in a certain POV. Just thinking about those discussions gives me a headache. It sure would be nice if we had archives for the nir and original uri mailing lists. ....Roy ----------------------------- From: "John A. Kunze" Subject: Re: A question about Web history and vision To: "Henry S. Thompson" cc: "Roy T. Fielding" , Larry Masinter Date: Mon, 24 Nov 2008 16:04:27 -0800 (PST) --- On Mon, 24 Nov 2008, Roy T. Fielding wrote: > > When and why did the 'Resource' of URL/N/C/I become understood as > > distinct from retrievable representation? In particular, when did > > the possible of in-principle non-representable resources get > > included in the range of URIs? I don't know. Many were thinking this way significantly earlier than they felt it politic to air publicly. The nature of those discussions sometimes called for restraint if the goal was consensus. There's probably a different answer inside vs outside the IETF. It's not a hard concept in the abstract (eg. pre-web world), but in the IETF the resistance came I think from those who didn't want to destabilize gains from the WWW for the sake of "unnecessary abstraction". What I believe they failed to see is the mainstream need for that kind of abstraction. > >[John, given the extract above, the particular question for you is how > >or why the idea of resources which "have no electronic instantiation" > >emerged. Why would you coin a network 'locator' for something which > >by its nature could not have a network location?] Again, strangely mainstream. I worked in a library containing lots of physical works that had electronic records. The search for a physical work was identical in all important respects to the search for a digital work. It wasn't important for identification purposes that the last step was a "click and see instantly" versus "click and receive via snail mail after an automated call request" versus "wait 18 months until this is digitized and then click and see instantly". The need was clear for an identifier system that spanned both the physical and digital worlds. As for other resources with little or no obvious representation, libraries have long been in the business of unambiguously identifying human authors (name authority), groups of humans (publishers), and vocabulary terms (Lib. of Congress subject headings, NLM disease codes). Unambiguous identification was good, and any reasonable retrievable approximate representation was a nice extra, if available. > 'locator' was a term created by the IETF discussions. I sometimes > wonder what would have happened if TimBL had started with something > like my definition of resources/representations and took that to > the IETF. It is possible that the abstraction would have obviated > the perceived need for the URL/URN split. OTOH, the split may have > been inevitable, since the folks interested in persistent names were > working on solutions before UDI was even proposed and were already > vested in a certain POV. Just thinking about those discussions > gives me a headache. Yes. I think it was innocently chosen, but people's strong tendency to spuriously equate 'locator' and 'location' has created endless confusion, unneeded URNs, unneeded companies, and roadblocks in persistent naming. -John