- From: Pieter Van Lierop <pvanlierop@geac.fr>
- Date: Mon, 17 Sep 2001 09:16:23 +0200
- To: Ian Ibbotson <ian.ibbotson@k-int.com>
- Cc: ZIG <www-zig@w3.org>
You don't want to execute the same SQL request twice. Here is a simple solution, maybe it helps: The server does an UNLOAD of the SQL search to a file that includes the name of the ResultSet. Whenever the server receives a Sort request on that ResultSet it uses the file on disk. By the way, you will have the same problem with the Present, I guess? So this is a solution for Present, too... Pieter van Lierop Geac > -----Original Message----- > From: Robert Sanderson [mailto:azaroth@liverpool.ac.uk] > Sent: dimanche 16 septembre 2001 14:47 > To: Ian Ibbotson > Cc: ZIG > Subject: Re: Sort criteria in Search Request > > > > > instructions are bundled with the search criteria. As I > said at the time, I > > suspect this comes from a long tradition of SQL where the > order by is an > > integral part of the whole statment. > > Right. But in Z39.50, it isn't explicit. Equally, you can't (as far as > I'm aware) do something like: > > search for X and scan the results by Y. > > (compared to 'search for X and sort the results by Y') > (eg: 'Search for books written by Tolkien, and browse their titles' > vs: 'Search for books written by Tolkien, and sort based on > their titles') > > > > Could someone please explain the actual point of doing > this as opposed to > > > just sending multiple requests as per the status quo? > > > The real need for sort to be close to search-request is, as > the doc says, for > > targets that need to be able to optimise the way they work. > I have a few SQL > > backends, and it's really inconvienient to execute what can > be a heavy SQL > > statement, only to have to immediately re-evaluate it > because a sort-request > > has just come in and now we need to tag some order by > clause on the end. > > So the point is just to make it easier for using SQL as a > backend then? > > I don't have a great problem with doing this sort of thing, > but it seems > to be getting away from what makes Z a good protocol for IR. > See also ZNG. > Like Unix, you have small units that are put together in > various ways to > get to the results you want, rather than the Microsoft one > thing-does-all > approach. > > On the other hand, look at Microsoft for what happens when > you 'extend' a > protocol to make it more integrated with currently existing > applications. > > Our (generic) Z client makes use of Explain. It's -extremely- > inconvenient > that this simply isn't implemented at all in most servers, which means > that most of the time we need to try to search -all- of the > possible USE > attributes for supported indexes. With the (very good) > proposed Bib2, as I > understand it, even this won't really be a feasable approach to > discovering how its possible to get records from the target > as the index > types are qualified semantically rather than having a separate set of > attributes. > > A lot of targets, according to their Supported Requests info > in Init don't > support sort. And a lot of the ones that claim it, don't in > truth. Things > like encapsulating sort in search tends to beg the questions > of is sort > supported at all, and being able to find out that it is with > reliability. > > Rob > -- > ,'/:. Rob Sanderson (azaroth@liverpool.ac.uk) > ,'-/::::. http://www.o-r-g.org/~azaroth/ > ,'--/::(@)::. Special Collections and Archives, extension 3142 > ,'---/::::::::::. Syrinnia: telnet: syrinnia.o-r-g.org 7777 > ____/:::::::::::::. WWW: > http://syrinnia.o-r-g.org:8000/ > I L L U M I N A T I >
Received on Monday, 17 September 2001 03:26:21 UTC