W3C home > Mailing lists > Public > public-xg-webid@w3.org > April 2011

Re: WebID-ISSUE-55 (schema-openess): explore WebID schema agnosticims

From: Henry Story <henry.story@bblfish.net>
Date: Tue, 19 Apr 2011 20:40:07 +0200
Message-Id: <0C6CD29C-4455-44C0-B5FF-C4A2BB4A5298@bblfish.net>
To: WebID Incubator Group WG <public-xg-webid@w3.org>
Kinglsey wrote in a different thread that was way too long but can be found here
> Henry,
> ftp: scheme is done re. ODS. It was done the day http: scheme was done since our architecture is based on the URI abstraction and its application to data access by reference. That lies at the core of our platform.

Great. But is it in the spec? And do we have tests for it? The point of W3C is to make sure things are documented, non patented, free and tested.

> ftp: scheme doesn't solve the intuitiveness problem. Again "deceptively simple" is better than "simply simple". Who is the target audience for the beyond http: scheme meme? There very folks WebID needs the most, the ones that already have applications that need a security tweak. Not the ones looking for a new programming adventure.

Can you explain the intuitiveness problem? I have no idea what you are on about. 

> WebID should be delivered in an unobtrusive manner.
> Toby's silence is borderline deafening, he is a co-creator of WebID who (like us at OpenLink) has WebID authentication that works beyond HTTP scheme URIs. Now you have 2 sources of what you've veered towards "ftp:" using the same old "its simple" excuse.

So what was all that thread about then talking of people having double standards?  Do you want us to make sure the spec is written so as to allow new URI schemes to come in or not? Is the specification broken by the way? Where? How can we fix it? FTP is good for solving those problems.

> What's the entire justification for the myopia inherent in all Web 2.0 applications re. Identity and as a result privacy? The same old "its simple" excuse.
> Why does the InterWeb actually work? IMHO because URIs are "deceptively simple". Ditto why the WWW part of the InterWeb works courtesy of the inherent ingenuity of HTTP which is also "deceptively simple".

> Henry Story wrote:
>> As you say ftp urls lack content negotiation, which means of course a small but interesting problem.
> An unnecessary problem since ftp: scheme is not the choice, if anything it opens a can of worms re. strange biases.

So you don't want ftp scheme in there? Should we make sure the spec excludes it? (I would be against that, btw)

>> The receiver will have to do content inspection just for that to guess what the format is.
>> It is pretty easy to see here how by solving those issues and testing them ftp urls would become a well documented part of webid. (I would nearly suggest it is so easy to see I am surprised anyone would doubt that we are not schema agnostic) Now it might be a good stimulus to help us change the specificiation somewhat so that we can spot those parts that are too URI specific.
> It took ftp: scheme for that to be clear?

That is what spec writing is about. Finding those small unclarities and clarifying them. 

>> I don't really see the use case for this, since anyone who has an ftp server can get an http server.
> Exactly! As per my opening comments.

You opening comment Kingsley is that ODS has implemented FTP. Is that compatible with the spec as it is written now? http://webid.info/spec/ Or do we need to change something. If it is ok, then what was all that talk about WebID not being schema agnostic enough.

>>  FTP was really important when the web came to be because then the difficulty of acquiring http was a lot greater comparatively. On the other hand the amount of work to do ftp should be pretty minimal.
>> If people are for that please +1 and I'll add it as an issue.
> -1000 . Solve nothing.

Not even making your ODS server implementation standard? Why did ODS implement the FTP protocol?

>>  When done we can have a vote to open it too, the idea being to look at the spec and see how it needs to be rewritten for ftp (and hence made generic enough for other existing or yet to be URI schemes)
> Why don't you simply define what it takes to have an implementation instead.
> mailto:, acct: schemes already exist in ODS. Why doesn't it count?

Never said it does not count. What needs to change in the spec? I think starting with FTP is good because we can all implement that in an hour I am pretty sure. 

> I am a horrible speculator (don't have the patience for it), but I think I have a pretty good track record re. implementation. We (OpenLink) implemented this stuff a long time ago, and the new set of WebID tools are telling us: your stuff is invalid when we do interop tests.

That is great. I just think we should also build tests for what we implement. It makes it much easier to test that our spec is good and our implementations are good. It saves us a lot of time trying this out on a case by case basis. Remeber the whole thread "self-signed" started because our test suites permitted 
Joerg Anders to find a problem on one server, that he did not find on another. 


Once that thread is finished we can build a new test to make sure we understand when this happens next time.

> Then when I flag the dangerous trend, you respond (subjectively) with comments like: "You shouted at XYZ..." .
> I could be more aggravated than Peter right now, but I sorta very used to a lot of this re. standards implementation :-)
> Kingsley
Social Web Architect
Received on Tuesday, 19 April 2011 18:40:39 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:39:44 UTC