Re: Address Bar URI

On 10/18/11 1:49 PM, Jonathan Rees wrote:
> On Tue, Oct 18, 2011 at 12:42 PM, Kingsley Idehen
> <kidehen@openlinksw.com>  wrote:
>> On 10/18/11 11:20 AM, Jonathan Rees wrote:
>>> Wow, this is new information for me that the redirect-to-hash issue
>>> would bear on this question, so this is interesting.
>>>
>>> However I must be dense since I don't see how it applies. The scenario
>>> I'm talking about is: I want an RDF URI for something. I mint a URI
>>> A#it, publish a document at A explaining what A#it means, and I think
>>> I'm done. Then something really bad happens, and I attribute the
>>> evilness to the use of hash, and I swear off ever using hash again.
>>> Luckily there is 303 as a backup, and I'm willing to pay the extra
>>> round trip overhead to avoid the hash badness, since the badness is so
>>> bad.
>>>
>>> I know about Dublin Core's use of redirect-to-hash, but it's being
>>> used to implement hashless URIs, not hash URIs. My question is why the
>>> need to use hashless URIs for this use is felt in the first place. (In
>>> DC's case it was legacy, but few people on this list are dealing with
>>> pre-2005 URIs.)
>>>
>>> What is the "really bad" thing that happened? (And what could it
>>> possibly have to do with redirects?)
>> The issue was, at the time, choosing a URI style for DBpedia that would work
>> with all browsers, including IE 6. Hash URIs would have been problematic
>> since the # crossed the wire. DBpedia modulo IE wasn't an option. The goal
>> was to deliver a Linked Data showcase that worked with all browsers starting
>> with IE 6.
> Do you mean to say: If an HTML document had
> href="http://example.net/doc#frag", and the link were followed, then
> IE6 would send a GET request to example.net, with a request-URI of
> doc#frag?  I find that very hard to believe. I googled around a bit
> and there seem to be lots of fragment id issues with IE6, but I didn't
> see which might apply to linked data.
>
> I'm not trying to be difficult, I just really don't get what you're saying.
>
>> I believe the your quests was about a case for 303's. Which is basically
>> another way of seeking a case for slash terminated URIs re. Linked Data
>> deployment.
> Not exactly - I'm trying to build a case against hash URIs. The only
> reason 303s exist is because people for reasons still not clear to me
> don't want to use hash URIs.

That cannot logically be the reason for 303s existence.

More specifically, I think you mean, in the context of Linked Data: use 
of slash terminated URIs as Object IDs (function: Generic Name).

As I've already stated, and this is really important. DBpedia (first 
significant Linked Data showcase) opted for slash terminated URIs 
because it sought to be browser agnostic, and then meant a solution that 
worked for IE 6 users.

The arguments about Slash vs Hash URIs are just another variation of the 
Syntax arguments re. directed graph construction. In both cases, 
implementation details detract from core concept comprehension.

Slash and Hash terminated URIs are options that people will have to live 
with forever re. Linked Data. Each approach has its implementation pros 
and cons.

>   To relieve the general suffering, one
> possibility would be to advocate for hash URIs.

Why?

Folks should grok the core concept and then choose whatever works best 
for their publication circumstances. User Agents could (and should) care 
less.

> I'm trying to figure
> out whether that's a bad idea.

Advocating a particular style of Identifier is a bad idea, as per my 
comments above :-)
> I have no idea what you mean about slash terminated URIs, sorry.

http://kingsley.idehen.net/dataspace/person/kidehen#this -- hash 
terminated (maybe I should have said: hash style of URI)

http://dbpedia.org/resource/Linked_Data -- slash style of URI .

Kingsley
> Best
> Jonathan
>
>> Kingsley
>>
>>
>>> Thanks
>>> Jonathan
>>>
>>> On Tue, Oct 18, 2011 at 9:41 AM, Kingsley Idehen<kidehen@openlinksw.com>
>>>   wrote:
>>>> On 10/18/11 7:54 AM, Jonathan Rees wrote:
>>>>> Can someone remind me why people are using 303 at all, as opposed to
>>>>> hash URIs in the #_ or #it pattern?
>>>>>
>>>>> I've been trying to make a compelling case for 303 over hash, without
>>>>> much success.
>>>>>
>>>>> What would be most valuable is war stories, especially ones that
>>>>> answer questions that have been left unanswered in the previous thread
>>>>> http://lists.w3.org/Archives/Public/www-tag/2011Sep/0003.html and in
>>>>> the writeup http://www.w3.org/2001/tag/awwsw/issue57/latest/#hash
>>>>>
>>>>> Thanks
>>>>> Jonathan
>>>>>
>>>>>
>>>> I think that had a lot to do with IE and the desire to boostrap Linked
>>>> Data
>>>> in a manner that worked across all browsers.
>>>>
>>>> Links:
>>>>
>>>> 1.
>>>>
>>>> http://jamespreston.co.uk/Articles/RedirectingIE6ToUrlWithFragmentIdentifier.html
>>>> -- here is a 2007 post about the problem
>>>>
>>>> --
>>>>
>>>> Regards,
>>>>
>>>> Kingsley Idehen
>>>> President&      CEO
>>>> OpenLink Software
>>>> Web: http://www.openlinksw.com
>>>> Weblog: http://www.openlinksw.com/blog/~kidehen
>>>> Twitter/Identi.ca: kidehen
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>
>> --
>>
>> Regards,
>>
>> Kingsley Idehen
>> President&    CEO
>> OpenLink Software
>> Web: http://www.openlinksw.com
>> Weblog: http://www.openlinksw.com/blog/~kidehen
>> Twitter/Identi.ca: kidehen
>>
>>
>>
>>
>>
>>
>>
>


-- 

Regards,

Kingsley Idehen 
President&  CEO
OpenLink Software
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen