- From: Alan Kent <ajk@mds.rmit.edu.au>
- Date: Mon, 21 Jul 2003 19:54:30 +1000
- To: Adam Dickmeiss <adam@indexdata.dk>
- Cc: www-zig@w3.org
On Mon, Jul 21, 2003 at 11:25:36AM +0200, Adam Dickmeiss wrote: > > > > http://www.mds.rmit.edu.au/~ajk/z39.50/util.html > > > > I've read it and I'd like to reduce the size of the spec if possible:) Always a good thing. > The spec says that comparison should only be one of "all", "any", "adj" if > structure is word. If that is the case, why not make a new comparion=string > compare (14) and skip structure type entirely? > > I guess one would not: there will be other structure types in the future > (date?). I want to support other values for this type personally. The format/structure is also necessary for scanning where you would not say 'all', 'any' or 'adj'. > Since, all,any,ajd,_relevance_ are now all of type comparison, does that > mean we can no longer say "relevance search - all words must match"? > Admitted, if the comparison type was repeatable it's would be a > different story > (but that is a sign of an insufficient model). Or move all,any,adj into expansion where repeats are allowed. I originally proposed expansion then moved it to comparison based on other feedback. (Actually, I think I started with a new type! ;-) I personally have no preference. All choices can be justified (which might been the spec is a bit loose...) Alan
Received on Monday, 21 July 2003 05:54:39 UTC