>A belated reply:
>
>>That would be good. What would *really* be good was if you could have
>>words indexed several ways, i.e.
>>
>>Smith-Watson
>>
>>would be found by searching for
>>
>>"Smith"
>>or
>>"Watson"
>>or
>>"Smith-Watson"
>
>
>If the future SET OF WORD DELIMITERS really is a set, and if it's
>applied BOTH to the index AND to the user's query, the above will
>work.
>
>Meanwhile, just create a second field that converts "-" (and any
>other chars that should behave the same way) into " ", and index
>that. (I can't remember if you can do this with a method; if not,
>just do it using RB/C/AppleScript/Frontier/whatever.) And
>, apply the same transformation on the user's text when you build the SQL.
>
That is in fact what I'm doing. But it is still not ideal. Using this
strategy, searching for
Smith-Watson
will find records containing the words
Smith AND Watson
(not necessarily together), when I want it to find the specific name
Smith-Watson
(and if there's no hyphen in the entry in the database, it shouldn't
find that).
Right now I'm offering two search options, one indexed (fast, not as
"accurate"), the other for a given string (slow, but will find
Smith-Watson as a unique hit).
Jon
-------------------------------------------------------------
The Valentina mailing list is brought to you by MacServe.net
For info on lists services, see http://www.macserve.net/lists.html
©2001 jda |
