- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Fri, 1 Nov 1996 06:46:43 PST
- To: yarong@microsoft.com
- CC: w3c-dist-auth@w3.org
Danger! Here Be Dragons! # This group must define a fairly robust set of attributes. Else everyone # will invent their own and it will take forever before things sort # themselves out. Having been at the Dublin meeting that tried to define the Dublin core and watched what's happened since, I think this would be a poor choice: there's no reason to believe that _this_ group could converge faster on an identical problem. Why not just use the Dublin Core, for that matter? Or the DMA set? Or ... well, see, part of the problem is that there are so many core attribute sets around! Which to choose from? Add another? # A mechanism to set attributes is just that, a means. I am much more # interested in the ends. For that, we need to define the attributes we # defined the mechanism for. Attribute sets have a much more narrow range of applicability than do retrieval languages. # MIME types are nothing more than data formats. There is nothing mystical # about them. We have a need for well defined data formats so we decided # to use the MIME mechanism. We need a 'D'. 'M' is a 'D'. Therefore, we decided to use 'M' for our 'D'. Sorry, it doesn't fit. You could define a 'D' which included 'M', but it wouldn't be 'M'. In particular, there are systems that are defined to accept everything that is a 'M' that will resist mightily any attempt to push your 'D' needs into their 'M'. Anyway, metadata type restrictions don't really work (I mean that you can never say quite what you want in them anyway.) I'm guessing you won't take my advice that these things are ratholes and we should desist from thinking we can solve them quickly, and we'll have to have a long back and forth where I sound curmudgeonly and tell you about all the problems... Oh well, Larry
Received on Friday, 1 November 1996 12:56:19 UTC