[Om] Going real with the OpenMath CD Linked Data

[Paul Libbrecht]

Le 27 juil. 2011 à 11:44, Christoph LANGE a écrit :

> Hi Paul,
> 
> I'm back working on the linked data publication.
> 
> On 26/07/2011 13:12, Paul Libbrecht wrote:
>> one version is active now. It was manually contributed by our sysadmin and worked for the xhtml case.
>> Another version is coming up in a few minutes which is almost the
>> same as your file.
> 
> Thanks, the redirects seem to work.  I have just tried "cdname" with
> different Accept headers and it looks all fine.

ok

> BTW, contrary to the original .htaccess file that I sent you, I think
> it would make sense to redirect requests to "cdname" without an Accept
> header to the *.ocd files instead of *.xhtml.  That mainly addresses
> users of command-line clients such as wget, and it would help to make
> them aware of the original implementation of the CDs in the OCD
> format.

ok for me. Can you do it?

>> thanks for feedback
>> There's an amount of URLs that need to be checked, including:
>> - http://www.openmath.org/cd/cdname.xhtml
>> - http://www.openmath.org/cd/cdname (with a diversity of accept
>> headers)
> 
> these work
> 
>> - http://www.openmath.org/cd/cdname.html (historical)
> 
> That ultimately redirects to cdname.xhtml, as desired, but for some
> strange reason the redirect goes via the intermediate step
> cdname.html.xhtml (in case of "Accept: text/html") or via
> cdname.html.rdf (if there is no Accept header):
> --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< ---
> 
> $ wget --header="Accept: text/html" -O /dev/null -S http://www.openmath.org/cd/a
> lg1.html
> --2011-07-27 11:33:57--  http://www.openmath.org/cd/alg1.html
> Resolving www.openmath.org (www.openmath.org)... 134.96.184.220
> Connecting to www.openmath.org (www.openmath.org)|134.96.184.220|:80... connecte
> d.
> HTTP request sent, awaiting response...
>  HTTP/1.1 303 See Other
>  Date: Wed, 27 Jul 2011 09:33:56 GMT
>  Server: Apache/2.0.50 (Linux/SUSE)
>  Location: http://www.openmath.org/cd/alg1.html.xhtml
>  Content-Length: 337
>  Keep-Alive: timeout=15, max=100
>  Connection: Keep-Alive
>  Content-Type: text/html; charset=iso-8859-1
> Location: http://www.openmath.org/cd/alg1.html.xhtml [following]
> --2011-07-27 11:33:58--  http://www.openmath.org/cd/alg1.html.xhtml
> Reusing existing connection to www.openmath.org:80.
> HTTP request sent, awaiting response...
>  HTTP/1.1 301 Moved Permanently
>  Date: Wed, 27 Jul 2011 09:33:56 GMT
>  Server: Apache/2.0.50 (Linux/SUSE)
>  Location: http://www.openmath.org/cd/alg1.xhtml
>  Content-Length: 333
>  Keep-Alive: timeout=15, max=99
>  Connection: Keep-Alive
>  Content-Type: text/html; charset=iso-8859-1
> Location: http://www.openmath.org/cd/alg1.xhtml [following]
> --2011-07-27 11:33:58--  http://www.openmath.org/cd/alg1.xhtml
> Reusing existing connection to www.openmath.org:80.
> HTTP request sent, awaiting response...
>  HTTP/1.1 200 OK
>  Date: Wed, 27 Jul 2011 09:33:56 GMT
>  Server: Apache/2.0.50 (Linux/SUSE)
>  Last-Modified: Mon, 25 Apr 2011 22:18:18 GMT
>  ETag: "73daa-878e-95a39e80"
>  Accept-Ranges: bytes
>  Content-Length: 34702
>  Keep-Alive: timeout=15, max=98
>  Connection: Keep-Alive
>  Content-Type: application/xhtml+xml
> Length: 34702 (34K) [application/xhtml+xml]
> Saving to: `/dev/null'
> 
> 100%[======================================>] 34,702      --.-K/s   in 0.1s
> 
> 2011-07-27 11:33:58 (319 KB/s) - `/dev/null' saved [34702/34702]
> 
> --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< ---

that's ok for me.

> 
>> - http://www.openmath.org/cd/cdname.xml (historical)
> 
> That one does not yet work. BTW, what should *.xml redirect to? To
> *.xhtml, or to *.ocd, or to *.rdf?  Note that all of these are XML
> formats.
> 
>> Maybe we need to add more redirects.
> 
>> I am wondering if redirect (303 in your case) is the right thing to do.
>> I believe it is ok to actually server the resource, which gives all the right headers, under
>>   http://www.openmath.org/cd/cdname
>> or?
> 
> For answering that I have to dig into linked data specifications.  The
> short answer is that 303 redirects are part of the linked data best
> practices, and that most linked data clients won't work otherwise.
> One can test that in a very nice way using the
> http://validator.linkeddata.org/vapour validator.

that most validators follow 303 is clear.
My problem with 303 is with browsers: they contain a history and that history's results (the target of the redirects) is what is exchanged with other people (copy and pasting URLs). This practice suffers from redirects because it lets the URL that is exchanged witness the representational nature (used to be .xml now is .xhtml) while it should not.

> The long version, quoting from
> http://www.w3.org/TR/cooluris/#r303gendocument
> 
> --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< ---
> … to use a special HTTP status code, 303 See Other, to give an indication that the requested resource is not a regular Web document. Web architecture tells you that for a thing resource (URI) it is inappropriate to return a 200 because there is, in fact, no suitable representation for those resources. However, it is useful to provide information about those resources. The W3C's Technical Architecture Group proposes in its httpRange-14 resolution [httpRange] document a solution that is to direct you to a document which has information about the thing you asked about. By doing this we avoid ambiguity between the original, real-world object and the resource that represents it.
> --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< --- %< ---
> 
> In short: The cdname.rdf RDF/XML file, which is the target of the
> redirect, is not the same as "the 'cdname' OpenMath CD". E.g. the
> cdname.rdf file may have metadata such as dc:creator = "Christoph
> Lange", dc:date = "2011-07-27" (because I created it), whereas the
> actual OpenMath CD, i.e. the thing described by the URI
> <http://www.openmath.org/cd/cdname> (and not by the URI
> <http://www.openmath.org/cd/cdname.rdf>) would rather have metadata
> such as dc:creator = "James Davenport", dc:date = "2004-06-30".  For
> our day-to-day practice this distinction is rather philosophical, but,
> as I said above, linked data tools expect 303 redirects nevertheless.

I am not disputing that 303 should redirect.
I am disputing that we should use 303 instead of just serving.
My point is that browsers should not receive redirects.
I am fine with other tools to receive redirects.

> 
>> Robin Berjon had a particularly nice post here:
>>    http://berjon.com/blog/2009/12/xmlbp-non-http-namespaces.html
>> where he says... don't name the technology of your namespace document! Another day it's going to be another technology... I understand this as the fact that we should share URLs without the nitty gritty details that made us use .xml one day and .xhtml another day... just use one name and do the right thing (of today and in the future).
> 
> I completely agree with that, and I think that with the improved
> redirects we achieved exactly that.

Except if people exchange the URL arith1.xhtml...

paul