lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Doron Cohen (JIRA)" <>
Subject [jira] Commented: (LUCENE-903) FilteredQuery explanation inaccuracy with boost
Date Mon, 04 Jun 2007 01:05:15 GMT


Doron Cohen commented on LUCENE-903:

> holding them to the same standard (when they can't easily modify 
> the string pattern rules in CheckHits) doesn't seem fair. 

> lemme try to refactor the current patch a bit so that the "deep" 
> Explanation testing is optional (and used by the core tests) 

I agree (and pointed that) the "deep" explanation test is imperfect in this regard. But it
seems to me that it is the benefit of new query writers to have their new queries explanations
tested as thorough as possible. So I am not sure that disabling this test by default (for
non core queries) is the right thing to do.

How do you feel about adding an escape way, allowing one to specify that a certain (part of
an) explanation is not to be checked in this manner?

I see two immediate ways to add this: 
- by the explanation text: if the description starts with "other:" the explanation is not
subject to this "deep" check
- by api: adding a property to Explanation class - disabledDeepCheck - false by default -
allows new unique queries to deliberately disable this "deep" test.

> FilteredQuery explanation inaccuracy with boost
> -----------------------------------------------
>                 Key: LUCENE-903
>                 URL:
>             Project: Lucene - Java
>          Issue Type: Bug
>          Components: Search
>    Affects Versions: 2.2
>            Reporter: Doron Cohen
>            Assignee: Doron Cohen
>            Priority: Minor
>             Fix For: 2.2
>         Attachments: lucene-903.patch
> The value of explanation is different than the product of its part if boost > 1.
> This is exposed after tightening the explanation check (part of LUCENE-446).

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message