[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[geovrml] Re: [media] Transparent URN Encoding in URLs (concept and chatexcerpt)
Hi Mitra...
> This seems to me a fundamentally bad idea
For current VRML browsers it doesn't make sense, as you suggest, since
when spec-compliant they skip over URLs (and URNs) that they don't know.
With many VRML browsers, however, they don't just skip over the URN but
instead pop up an alert box that lists them in an error message.
WorldView, for example, is so aggressive in doing this that it looks
like URNs are a problem when it's actually just a verbose reporting
mechanism.
This is what lead to the idea of transparent encoding of URNs in URLs:
the idea was to eliminate the scary dialog boxes that appear in browsers
that can't handle URNs. Along the way it became apparent that this
approach would also allow URNs to be used in standard HTML pages (or
anywhere else a URL can appear) without breaking the user agent
(browser) or requiring it to be recoded. As a URL it will load just as
any other URL. If, however, the user agent is savvy to such a
recommended practice (assuming it becomes one) then it would pick the
URN out of the URL and use it to fetch the media from the local drive.
So, we get 2 important solutions in one with this approach... 1)
elimination of verbose VRML browser warnings/alerts when URNs are
encountered and 2) full compatibility with single-URL user agent (such
as standard Web browsers).
What do you think? If this isn't a deal breaker we can easily ad a
section to our URN Recommended Practice document that shows how to do
this (I'm not suggesting that we eliminate the current solution -- not
all all; just provide content authors with a clean solution to this
problem with older VRML browser and also give HTML authors a recommended
way of integrating Universal Media into their Web pages). Alternately we
could just issue a new paper that gives guidelines for encoding URNs in
URLs for those who want to do it.
--
Regards,
Aaron
--
---------------------------------------------------------------------
Aaron E. Walsh http://www.mantiscorp.com/people/aew/ 617.350.7119
---------------------------------------------------------------------
Mitra wrote:
>
> This seems to me a fundamentally bad idea - it still requires software on
> the browser side to interpret the strange URL, so its only going to work
> with VRML browsers anyway (I lost that argument with Tim Berners-Lee when
> he was starting the web!). It won't work with any future URN schemes since
> it doesn't start "urn:".
>
> The VRML spec clearly said that a browser had to ignore URN's if it didn't
> understand them, that was the opening that allowed us to design this scheme
> in the first place. If WorldView complains about them then it is
> non-compliant and should be fixed.
>
> Placing folders in startup locations on drives is problematic, that is
> Chris and I didn't take this approach when we designed the URN -> local
> location resolution process.
> - Mitra
>
> At 07:31 PM 7/2/2000 -0400, Aaron E. Walsh wrote:
> >Hello Don and everyone, as it turns out I'm going to be burnt for time
> >most of this week so I probably won't have a chance to complete the
> >"Transparent URN Encoding in URLs" as mentioned in my reply to Don's
> >original message regarding local host URLs. I will probably complete it
> >over the next weekend so that we have something that everyone can
> >comment on before Siggraph, and also in hopes of nailing down the local
> >mapping issues that have resurfaces (I think the gap if finally closing
> >on local URLs, it's just a matter of drilling into them for a while
> >longer).
> >
> >Instead I hope that the below excerpt from our Web3D.about.com chat
> >helps. The concept is simple: embed URNs directly into URLs for
> >transparent support of URNs by software that doesn't understand them
> >already (Web browsers, for example). By mapping such a URL to a local
> >location similar to the idea Don suggested we should be able to develop
> >a single resource naming solution that travels well from web to all
> >local platform (assuming we also establish local installation default
> >locations, which shouldn't be a problem -- vendors simply place a folder
> >on the root level of the startup drive having the same domain name as
> >the URLs itself.
> >
> >For example, consider the following URL (notice the URN encoded in it):
> >
> > http://www.web3dmedia.com/urn:web3d:media:/textures/nature/grass_1.jpg
> >
> >The above would map to the local file location on Windows systems:
> >
> > C:\www.web3dmedia.com\urn\web3d\media\textures\nature\grass_1.jpg
> >
> >Colons are mapped to local (platform specific) directory separators
> >(backward slashes, in the case of Windows above). NOTE THAT THIS IS NOT
> >HOSTED ON web3d.org at the moment because I was experimenting with it
> >quite a bit and pushing it live didn't make sense (instead I worked on
> >it at the above media site), although we'd do it via web3d.org
> >officially if we agree it's a good idea (to avoid multiple local
> >installations of the same textures, sounds, etc.).
> >
> >If we can come to some agreement on this I have no problem updating our
> >Universal Media URN proposal to reflect 1) Transparent URN encoding
> >guidelines and recommendations, 2) Local mapping guidelines and
> >recommendations, 3) A *new* default installation location for our system
> >based on the above.
> >
> >Please consider the following chat we had before diving in, as we
> >covered some preliminary ground that might answer questions in advance.
> >
> >Regards, Aaron
> >--------------------------------------------------------
> >NOTE: BELOW EXCERPT TAKEN FROM WEB3D.ABOUT.COM CHAT LOG:
> >
> >http://web3d.about.com/compute/web3d/library/chatlogs/2000/blcl032200a.htm
> > SUBSTANTIALLY EDITED FOR CLARITY (REORGANIZED)
> >--------------------------------------------------------
> >
> >...... Log on 031500 started at Wed Mar 15 18:30:11 PST 2000 ......
> >
> >AaronWalsh said:
> >
> >Anybody seen one of these:
> > http://www.web3dmedia.com/urn:web3d:media:/textures/nature/grass_1.jpg
> >?
> >
> >It's a URN encoded in the body of a valid URL -- best of both worlds.
> >Here's the hook -- 1 identifier that acts like a URL and an URN at
> >once...
> >
> >Braden: I'd have to check RFC1738, but I'm not entirely sure that's
> >kosher with the definition of the http scheme.
> >
> >AaronWalsh: Last week I ran it by Leslie Dangle (URN chair at IETF). It
> >may fly.
> >
> >Braden: What was the reaction?
> >JEM: What does it accomplish?
> >
> >AaronWalsh: Universal Media in standard Web browsers! Web browsers don't
> >support multiple URLs, soooo....we give them 1, only 1, that has a URN
> >and URL. If it's UMedia savvy it grabs the resource locally, otherwise
> >it fetches it off the net without a problem. It also resolves ALL OF THE
> >NASTY dialog boxes that some browsers toss up with they encounter a URL
> >(URN) that they can't handle (Worldview, for example). It also allows us
> >to migrate into single URL environments where we aren't already...
> >
> >Guy: that's a great solution to the single line urn/url
> >
> >web3dADM: would this turn out to be a new method for caching?
> >AaronWalshL: Sandy (vweb3dADM) -- yes!!!!
> >
> >Braden: I'm still not clear on the problem this is solving. Web browsers
> >don't know about multiple URIs, but VRML browsers do. Why can't they
> >check the URN (if they understand it) and then hand off a URL
> >(subsequent in the list) if that doesn't work out?
> >
> >AaronWalsh: UMedia is all about global caching, one that ALL programs
> >has access to.
> >
> >JEM: You know, it's possible for 3rd parties to write protocol handlers
> >(such as URN:) for Internet Explorer. I bet it could be done pretty
> >easily. Mozilla probably has a similar capability..
> >AaronWalsh: Braden, how would you hand a URN to a Web browser via HTML
> >today?
> >
> >Braden: Aaron, Why would I need to?
> >AaronWalsh: Braden, suppose that you wanted to use a UMedia texture for
> >the background image of your Web page (HTML). Or as a IMG image.By
> >handing it a URL/URN combo we could support future-ready browsers
> >without breaking the millions of browsers already out there. Future
> >browsers can pick out the URN and resolve locally, while others fetch
> >from the fully specified URL. Seamless backwards compatibillity with
> >HTML.
> >
> >web3dADM: right it's UMEL as applied to any Web media application
> >AaronWalsh: Yes, waaaay beyond Web3D years from now.
> >web3dADM: I could see UMEL flash content
> >Guy: sure. the content is very useful for many tools and apps
> >web3dADM: with the main benfit I think, of increased reliability and
> >backward compatability for old browsers
> >
> >AaronWalsh: We're building media building blocks and a global, shared
> >cache that's valuable to any application that needs images, sounds, or
> >3D objects.
> >
> >Braden: And how do current browsers resolve the URL-URN hybrid? What
> >conventional URL would your hybrid map to?
> >AaronWalsh: Hold on....let me get one:
> >
> > http://www.web3dmedia.com/urn:web3d:media:/textures/nature/grass_1.jpg
> >
> >Load that into your Web browser now... does it resolve?
> >
> >Laurent: it even works
> >AaronWalsh: yep!
> >Braden: Sho' 'nuff.
> >web3dADM: hey it worked!
> >keithv: Cool!
> >Guy: well pinch my toes and call me a jelly donut. It works..!
> >
> >AaronWalsh: If you were a savvy tool, like Spazz3D or Contact, you'd see
> >this substring:
> >
> > urn:web3d:media:/textures/nature/grass_1.jpg
> >
> >and grab it from the hard drive instead. 2 in one: URL for backwards
> >compatiblity, URN for the future...
> >
> >web3dADM: why does it work?
> >
> >AaronWalsh: It's a URL!
> >AaronWalsh: What you get when you load this into your Web browser is a
> >real, honest to God URL.
> >
> >Guy: it's a URN..!
> >AaronWalsh: No, it's a URL with a URN in it! Cool, huh?
> >
> >Guy it's two.. two.. two links in ONE
> >web3dADM: yeah way cool
> >
> >AaronWalsh: Yes, so if we do this instead of making our URNs separate,
> >we handle everything without breaking current standards.
> >
> >--------------------------------------------------------
> >NOTE: ABOVE EXCERPT TAKEN FROM WEB3D.ABOUT.COM CHAT LOG:
> >
> >http://web3d.about.com/compute/web3d/library/chatlogs/2000/blcl032200a.htm
> > SUBSTANTIALLY EDITED FOR CLARITY (REORGANIZED)
> >--------------------------------------------------------
> >
> >Regards,
> >Aaron
> >--
> >---------------------------------------------------------------------
> >Aaron E. Walsh http://www.mantiscorp.com/people/aew/ 617.350.7119
> >---------------------------------------------------------------------
>
> --------------------------------------------------------------------------
> * Mitra Internet Consulting
> * Chief Technology Officer - AppleSoup.com
> * mitra@earth.path.net http://earth.path.net/mitra
> * International fax/voicemail: +1-415-704-3167
> * mobile +61-414-648-722
> * now in the US +1-415-931-8796
> * Or find me: http://earth.path.net/mitra/contact.htm
> *
> **** Life is a mystery to be lived, not a problem to be solved ******>