This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
3.5.2 Window Selection [1] para 1 "A window selection consist of" s/consist/consists/ [2] example 4 'The following expression returns the empty sequence, because in the selected book element, there is no occurrence of "efficient" within a window of 3 tokens which would not also contain an occurrence of "and": /book[@number="1" and . ftcontains "efficient" ftand ftnot "and" window 3 words]' [2a] But don't the occurrences of "and" give rise to StringExcludes? If the window selection only cares about whether StringIncludes fall within a window, the position of the "and" tokens shouldn't matter. [2b] If the "3" were changed to "2" or "1", would it return true?
Further to [2a]: Currently, the description of Window Selection, and the semantic function fts:ApplyFTWordWindow in 4.2.7.8 FTWindow, say that StringExcludes in a match for the operand full-text selection don't affect whether that match is also a match for the Window Selection (only the StringIncludes matter), but I'm pretty sure that's incorrect. Instead, I think the Window Selection must also require that any StringExcludes in an operand match fall outside the window. (It's not clear how it should treat a StringExclude that is only *partly* outside the window.)
[1] is fixed in CVS.
The Task Force considered item [2] at its F2F meeting on 2008-01-24 and adopted the proposal in member-only FTTF email http://lists.w3.org/Archives/Member/member-query-fttf/2007Dec/0004.html. The solution was accepted as written and the changes applied to the document. Michael, if you agree with the changes made, please close this bug.
Looks good.