W3C home > Mailing lists > Public > uri@w3.org > March 2004

Re: media URI design

From: Al Gilman <al.gilman@comcast.net>
Date: Tue, 30 Mar 2004 11:42:24 -0500
Message-Id: <p06020419bc8f4eef9374@[10.0.1.2]>
To: Kai Hendry <hendry@cs.helsinki.fi>
Cc: uri@w3.org

At 1:20 PM +0300 3/30/04, Kai Hendry wrote:
>Could someone please give me pointers/tips to a URI "scheme" for
>images/media I should be using, instead of making up my own.

Free advice is worth what you pay for it.

Mostly what you want is best practice.  Personally, I don't think
that this list is a good place to find that.

Try to find out what Black Star and other image re-sellers are doing
for a better 'benchmarking' step.

That said, here's my advice:

Some will say you should hide the file type; but use 'cleansed' URIs 
without filetypes very carefully, and not exclusively.

You can support negotiable URIs without filetypes if you wish; you 
have to decide what services you are going to support on your server 
to decide how much information disappears in these URIs. But also 
take responsibility for accepting and responding appropriately to 
reliable fileName+fileType URIs for the specific variants.

There is no golden rule about how you order the Dublin Core properties
of the image to construct a key, or how mnemonic vs. short you make
that URI.

BUT start with a Dublin Core record for the information object and use
those properties in the key FIRST before adding any funnybusiness.

If indeed you have a serialization scheme that will provide a flat
serial index of images across the 13 cameras you are using on a given
job, more power to you.  You will probably do better with a key
based on which device took the image and then serial within that.

Consider RDF metadata alongside the URIs to tell the whole story that
you would want recycled when some photography magazine re-publishes
one of your images.

Al

>
>Right now I use:
>http://natalian.org/pictures/2003/07/24/DCP_7071.JPG
>For the large version of the picture. And:
>http://natalian.org/thumb/2003/07/24/DCP_7071.JPG
>For the thumbnail.
>
>I am thinking of getting rid of the "pictures" bit, the filename and
>incorporating the device used to capture the media.
>
>http://natalian.org/kodak/2004/03/25/14:00:04/
>Or:
>http://natalian.org/kodak/2004/03/25/67/
>67 is the 67th image taken that day (a key)
>Or:
>http://natalian.org/kodak/2004/03/25/description/
>
>Thumbnail I could generalise to a preview(although a painfully long 7
>letter word). Example:
>http://natalian.org/preview/kodak/2004/03/25/23:45:45/drunk/
>
>I am worried that a browsers might ignore the content-type and expect
>HTML. Hmmmm...
>
>Comments please, and please CC: me. Good day,
Received on Tuesday, 30 March 2004 12:05:49 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 13 January 2011 12:15:32 GMT