[Om] CD directories once more: Where to place official vs. experimental CDs?

[Christoph LANGE]

2010-02-13 22:03 David Carlisle <davidc at nag.co.uk>:
> On 13/02/2010 18:20, Christoph LANGE wrote:
> > So what is the purpose of having two directories?
> 
> Historically, and perhaps still? the main difference was who had write
> access. Anyone could add files to contrib just by submitting a CD via a
> web form, whereas the core directory needed CVS (Now svn) write access.

Aha, I see.

> > Should they be cleaned up?
> 
> perhaps, a bit, but Cool uris's don't change
> [ http://www.w3.org/Provider/Style/URI ]
> which limits how much cleanup you can do.

Sure, cleaning up need not mean removing, but it can also mean setting
redirects or marking some of those CDs as "obsolete".

Secondly, I don't believe that _these_ URLs (I wouldn't call them URIs here,
see below) are the cool ones that we should pay attention to.  They are not
widely used by external users and should not be used for referring to CDs or
symbols.  I did a quick Google search for "cdfiles2" and found some old
publications, but those papers mentioned even older and less cool predecessors
of these URLs, e.g. the old ones that contained "cocoon" and thus reflected
the web server architecture we had at that time.

> One thing that I had thought we should set up is that the short
> openmath.org/cd/name version should use some smarter logic to redirect to
> whatever file drrectory is being used at the back, whetheer cdfiles2/cd or
> contrib/cd or somewhere else.

Good point, I fully support that.  _These_ are the cool URIs that should
officially be used for referring to CDs and symbols (now even recommended by
the MathML 3 spec), and indeed there should be such a setup as you suggest.  A
pre-smart setup is already in effect, in that for those CDs that are in the
cdfiles2/cd directory the bare CD name redirects to the XHTML file.  An
inclusion of cdfiles2/contrib/cd into that should be added.

And I'd suggest offering content negotiation, as described in more detail in
the extended Cool URIs paper (http://www.w3.org/TR/cooluris/).  XHTML should
only be served when it is requested (HTTP header "Accept: text/html"), e.g.
when accessing a CD/symbol URI/URL with a browser.  Alternatively, the CD
source should be served.  We do not yet have a MIME type for that (do we?);
can we use application/openmath+xml (see
https://trac.mathweb.org/OM3/ticket/106)?  Or should there be separate ones
for OMOBJs and for CDs?

I was planning to say more about that when receiving the notification about
that Web Science submission on the (mathematical) semantics of statistical
government data I mentioned two weeks ago.  There, we want to annotate data
exposed as RDF with URIs of OpenMath symbols.  For that to work well we'd need
to take Cool URIs and Linked Data best practices into account (see
http://www.w3.org/DesignIssues/LinkedData.html), i.e. that content negotiation
additionally serves RDF/XML when application/rdf+xml is requested.  We do have
all that RDF, but currently it's hidden somewhere in the OpenMath wiki and
only used internally there.

Cheers,

Christoph

-- 
Christoph Lange, Jacobs Univ. Bremen, http://kwarc.info/clange, Skype duke4701
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part.
Url : http://openmath.org/pipermail/om/attachments/20100213/f1de8363/attachment.pgp