Path: news.easynews.com!newsfeed1.easynews.com!easynews.com!easynews!uunet!dfw.uu.net!an02.austin.ibm.com!ausnews.austin.ibm.com!sp15ce20.hursley.ibm.com!not-for-mail From: "Paul Vernon" Newsgroups: comp.databases.theory Subject: Re: database design method Date: Mon, 11 Nov 2002 16:53:51 -0000 Organization: IBM Hursley Lines: 45 Message-ID: References: <3dcda833$1@news.uia.ac.be> <3dcfc941$1@news.uia.ac.be> NNTP-Posting-Host: dyn-9-140-177-100.warwick.uk.ibm.com X-Trace: sp15at20.hursley.ibm.com 1037033682 20732 9.140.177.100 (11 Nov 2002 16:54:42 GMT) X-Complaints-To: news@hursley.ibm.com NNTP-Posting-Date: Mon, 11 Nov 2002 16:54:42 +0000 (UTC) X-Priority: 3 X-MSMail-Priority: Normal X-Newsreader: Microsoft Outlook Express 6.00.2800.1106 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2800.1106 Xref: newsfeed1.easynews.com comp.databases.theory:23557 X-Received-Date: Mon, 11 Nov 2002 10:26:31 MST (news.easynews.com) "Jan Hidders" wrote in message news:3dcfc941$1@news.uia.ac.be... > >> This is of course their good right but I hope the authors realize how much > >> better for the acceptance of their model it would be if it were on line. > >> Think of ODMG/OQL and W3C/XML. > > > >How much difference do you think it would make? I agree that their model > >could certainly be 'marketed' with more, err pazazz. > > Indeed, and if the tone at dbdebunk became a bit less aggressive people > might actually start to listen again. You do have a point there - I might try to pass it on. > >Are there arguments in favour of recursively defined types in the relational > >model? I don't recall D&D explaining why they disallow them. > > Trees are a very common data structure. Also think of semistructured data. > If you don't support that then people will add their own domains to store > them in and then they will be out of reach of the query optimizer. True, but would your tree tuple example be the best way of implementing a tree data-structure anyway? > >What would a valid value of Tree look like? > > > >{"NodeA:Value1", ("NodeB:Value2", ("", {}) )} > > Yeah, something like that. > So ("", {}) and therefore {} would be valid values of that type? Doesn't look quite right to me. Regards Paul Vernon Business Intelligence, IBM Global Services