On 10/14/11 2:14 PM, Hugh Glaser wrote:
Thanks Rob. I agree. However, in some sense we may now have the way to climb just a bit out of the mess we have got ourselves into ("Get out of jail free card" springs to mind!). So if I am a publisher who is doing the httpRange-14 thingy (as I am), do you think doing the client-side address bar manipulation is a bad thing to do? If not, then should we not add it to the documents advising how to publish Linked Data for people doing it that way? (Which is where I have been going all along.) Which prompts the thought: Are there implications for hash-URIs too?I realise for publishers and developers it may not help much, but for consumers it is great. Which actually means for developers it is good, as they are less likely to have to worry about what the consumer is seeing being at odds with what they thought.
I don't think its great for consumers. The browser is capturing addresses. These addresses have one level of indirection re. access to actual data. That's what de-referencing an address is about. That's what WWW users are familiar with.
When browser vendors grok Linked Data indirection, and they choose to make it part of the UX, they'll change "address bar" to "resource name" or "data object name" or "data source name". Until then, I see no arguments for injecting >1 level of *indirection* into a UX pattern that's all about a single level of indirection. Of course, I am not using a zero based index re. my indirection levels :-)
Kingsley
Best Hugh On 14 Oct 2011, at 18:38, Rob Styles wrote:This redirection and the resulting address of a web page, rather than the name of the real-world thing originally requested, is one of the reasons I believe the IR/NIR distinction is problematic in practice. If we overturned httpRange-14 and allowed a non-information resource (i.e. a URI naming a real world thing) to return a 200 OK and a representation of itself then your client-side address bar manipulation would be unnecessary. In practical experience training and mentoring many developers and data hackers new to Linked Data this is one of the most problematic areas. rob Rob Styles Senior Technical Consultant http://consulting.talis.com/rob-styles On 14 October 2011 17:03, Hugh Glaser<[email protected]> wrote: Thanks, yes. It is client side JS, not a complicated server-side. It won't work for the purposes I describe if the Linked Data publisher has chosen (or had to) have the document on a different domain, but for all the cases I can quickly think of it does. Of course it won't work if window.history.replaceState is not available, but that can be tested for (Ian says). It is a bit undesirable that the "old" way, as it will become (!) should persist for the old browsers, but that is the Way of the Web. Best On 14 Oct 2011, at 16:40, David Wood wrote:Hi Hugh, Sorry, I think we are talking about different ways of accomplishing the same thing. You seem to be suggesting that the user's browser rewrite the URL in the address bar to match the URI you want to see. This is completely client-side approach. As you said, you can do that if you don't change the domain or protocol (because if you did, it would be a security nightmare). That seems fine to me. Another way to do this is to have the server rewrite the URI, using redirection and/or server-side URL rewriting. That's where my earlier comments came in: A server needs to have enough knowledge to re-write a URL properly after a redirection. Your browser tests tested your client-side scenario, but the server-side scenario is more complicated. I hope that helps. Regards, Dave On Oct 14, 2011, at 11:27, Hugh Glaser wrote:Thanks. I have just tried it on my mac with Firefox, Chrome, Safair, Flock and Opera. All seem to do it fine, although obviously I am running the latest versions. Ian Millard has modded rkbexplorer.com, so you can try: http://southampton.rkbexplorer.com/id/person-04860 It goes to http://southampton.rkbexplorer.com/description/person-04860 and then shows http://southampton.rkbexplorer.com/id/person-04860 again. (You do need to wait until the page has finished loading.) I don't know what the failure mode is on older browsers or buggy ones. But for these it looks good to me. But as you say, if there is other stuff going on too, things may break. Best Hugh On 14 Oct 2011, at 16:11, David Wood wrote:Hi Hugh, I like what you are saying and agree that this approach would be a real boon to the LOD community. Practical problems with using it are likely to be with subtleties of browser implementation. For example, Firefox resets all headers on redirect, including the Accept: header which causes us difficulty with PURL redirects. This is a known issue in Firefox and has been unfixed for four years: https://bugzilla.mozilla.org/show_bug.cgi?id=401564 The test page can be found here: http://tc.labs.opera.com/apis/XMLHttpRequest/send-redirect.htm I recall a similar bug report on Firefox where Mozilla developers were discussing how the address bar contents should be modified, and there was violent disagreement. Unfortunately, I can't put my hands on that URL at the moment, but I think you get my point: Each browser vendor will decide to handle these things differently in the absence of a standard (or in the presence of cross-site abuse concerns). The best way to proceed is probably to try it and test all major browsers. Regards, Dave On Oct 14, 2011, at 10:22, Hugh Glaser wrote:Excellent, hopefully that is out of the way. Does anyone want to express an opinion on the original question, which boils down to: "Is there a problem if going to URI http://www.myexperiment.org/workflows/158 (say, by clicking) in a browser then shows http://www.myexperiment.org/workflows/158 in the address bar for the page it displays?" I suggest not. It does bring one further question (at least): What do you display if someone goes to http://www.myexperiment.org/workflows/158.html? Possibly more controversial, as I suspect that the pragmatic answer is to display http://www.myexperiment.org/workflows/158 Best Hugh On 14 Oct 2011, at 14:23, Ian Davis wrote:On Fri, Oct 14, 2011 at 2:15 PM, Kingsley Idehen<[email protected]> wrote: Yes, opaque technology to you. Luckily, not for the rest of the computing universe. The large number of off-list messages supporting my view seems to provide evidence to the contrary. Apologies to the list for this off-topic conversation. I won't prolong it. Ian-- Hugh Glaser, Web and Internet Science Electronics and Computer Science, University of Southampton, Southampton SO17 1BJ Work: +44 23 8059 3670, Fax: +44 23 8059 3045 Mobile: +44 75 9533 4155 , Home: +44 23 8061 5652 http://www.ecs.soton.ac.uk/~hg/-- Hugh Glaser, Web and Internet Science Electronics and Computer Science, University of Southampton, Southampton SO17 1BJ Work: +44 23 8059 3670, Fax: +44 23 8059 3045 Mobile: +44 75 9533 4155 , Home: +44 23 8061 5652 http://www.ecs.soton.ac.uk/~hg/-- Hugh Glaser, Web and Internet Science Electronics and Computer Science, University of Southampton, Southampton SO17 1BJ Work: +44 23 8059 3670, Fax: +44 23 8059 3045 Mobile: +44 75 9533 4155 , Home: +44 23 8061 5652 http://www.ecs.soton.ac.uk/~hg/
-- Regards, Kingsley Idehen President& CEO OpenLink Software Web: http://www.openlinksw.com Weblog: http://www.openlinksw.com/blog/~kidehen Twitter/Identi.ca: kidehen
smime.p7s
Description: S/MIME Cryptographic Signature
