W3C home > Mailing lists > Public > www-webont-wg@w3.org > August 2003

Re: FW: OWL guide note?

From: Dan Connolly <connolly@w3.org>
Date: Thu, 28 Aug 2003 12:22:44 -0500
To: Charles White <Charles.White@networkinference.com>
Cc: www-webont-wg@w3.org, Jack Berkowitz <Jack.Berkowitz@networkinference.com>, Rob Shearer <Rob.Shearer@networkinference.com>
Message-Id: <1062091364.3433.550.camel@dirk.dm93.org>

On Fri, 2003-08-22 at 18:42, Charles White wrote:
> All,
>  
> A note from one our engineers.
>  
> charles
>         -----Original Message-----
>         From: Rob Shearer 
>         Sent: 22 August 2003 07:46
>         To: Jack Berkowitz; Engineering; Paul Turner; Matthew Quinlan
>         Subject: RE: OWL guide note?
>         
>         
>         My main concern is not in what the file extension is, but that
>         in making the change from .owl to .rdf the authors of the OWL
>         Guide have tried to hedge their bets and try going without any
>         extension at all. The "imports" statements in the ontologies
>         now reference a URLs that do not include *any* extensions.
>          
>         However, the Guide continues to reference the food and wine
>         files *with* an extension, now ".rdf".

That seems like the source of the problem.

I hope to investigate further.

>          So if we load "wine.rdf", it imports "food", which in turn
>         imports "wine", a completely *different* URL. They've added
>         xml:base attributes such that the new statements from "wine"
>         should be exact duplicates of everything in "wine.rdf", but
>         the fact that we're now dealing with three files instead of
>         two just seems silly.
>          
>         Much worse, this "two URLs for the wine file" thing is a huge
>         pain to configure. Maybe the W3C server guys managed to set up
>         two URLs for the same file, and made sure that the MIME type
>         mapping for the file without an extension must be RDF (since
>         the server can no longer automatically infer it from the
>         extension), but are server admins going to want to go through
>         such shenanigans for every ontology they publish?

One doesn't need to configure every ontology, or even every
MIME type. We use apache, and it's just a one-line config
file change, I think, to have requests for file_foo
serviced with the contents of file_foo.ext .

Details to follow...

>          And if you want to load an ontology not from a web server but
>         from a local file this sequence gets even worse, particular on
>         filesystems which don't allow symbolic links (which,
>         incidentally, are exactly those file systems on which
>         extensions are most important). How will Cerebra be able to
>         load the food and wine ontologies from local files on a
>         Windows machine?

If you're changing the address from http://www.w3.org...
to file://... you can change the extension while you're at it.

But the smart thing to do is to build caching mechanisms into
your product so that it can map http://... names to local
files.

Hmm... I ought to write this up as a best practice note/FAQ
in the ESW wiki... http://esw.w3.org/topic/ .

>          
>         Getting these ".rdf" extensions back into the "imports" lines
>         is probably by far the easiest solution.

While that might be the locally easiest solution,
the point of doing work in W3C is to optimize globally.

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Thursday, 28 August 2003 13:25:13 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:02 GMT