fts extension is not expanding stuff correctly in queries

Bug #614284 reported by Markus Korn
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Zeitgeist Extensions
Invalid
Undecided
Unassigned

Bug Description

Michal reported a bug in the fts extension on irc today:

11:06 < mhr3> second, once there's a bit more complex query, it no longer
              expands stuff
11:07 < mhr3> ie searching for "v*" and "(v*) AND (())" returns different
              results
11:08 < mhr3> i'm now not sure if "(v) AND (())" vs "v" also returns different
              stuff
11:09 < mhr3> but it's possible

Revision history for this message
Seif Lotfy (seif) wrote : Re: [Bug 614284] [NEW] fts extension is not expanding stuff correctly in queries

Are you sure this is fts and not wildcards ?

On Fri, Aug 6, 2010 at 11:43 AM, Markus Korn <email address hidden> wrote:

> Public bug reported:
>
> Michal reported a bug in the fts extension on irc today:
>
> 11:06 < mhr3> second, once there's a bit more complex query, it no longer
> expands stuff
> 11:07 < mhr3> ie searching for "v*" and "(v*) AND (())" returns different
> results
> 11:08 < mhr3> i'm now not sure if "(v) AND (())" vs "v" also returns
> different
> stuff
> 11:09 < mhr3> but it's possible
>
> ** Affects: zeitgeist-dataproviders
> Importance: Undecided
> Status: New
>
> --
> fts extension is not expanding stuff correctly in queries
> https://bugs.launchpad.net/bugs/614284
> You received this bug notification because you are subscribed to The
> Zeitgeist Project.
>
> Status in Zeitgeist Data-Sources: New
>
> Bug description:
> Michal reported a bug in the fts extension on irc today:
>
> 11:06 < mhr3> second, once there's a bit more complex query, it no longer
> expands stuff
> 11:07 < mhr3> ie searching for "v*" and "(v*) AND (())" returns different
> results
> 11:08 < mhr3> i'm now not sure if "(v) AND (())" vs "v" also returns
> different
> stuff
> 11:09 < mhr3> but it's possible
>
>
>

--
This is me doing some advertisement for my blog http://seilo.geekyogre.com

Markus Korn (thekorn)
affects: zeitgeist-dataproviders → zeitgeist-extensions
Revision history for this message
Mikkel Kamstrup Erlandsen (kamstrup) wrote :

fts does it's own wildcard handling, so it may well be there...

however I am not sure I really understand what the bug is about, can someone elaborate a bit?

Revision history for this message
Michal Hruby (mhr3) wrote :

I have a libzg query that searches for "v*" with an empty event as a template, this is the result:
DEBUG:zeitgeist.fts:Search '(v*) AND (())' gave 8 hits in 34.8780155182ms

The same query by using `./talis 'v*'`:
DEBUG:zeitgeist.fts:Search 'v*' gave 8845 hits in 12.4750137329ms

Now libzg search for "v" (again with empty event as a template)
DEBUG:zeitgeist.fts:Search '(v) AND (())' gave 8 hits in 0.739097595215ms

Result of `./talis 'v'`:
DEBUG:zeitgeist.fts:Search 'v' gave 70 hits in 0.261068344116ms

All of this was run on the same DB, why do the hits pairs differ?

Revision history for this message
Michal Hruby (mhr3) wrote :

And one more with Timestamp.anytime(): (libzg)
DEBUG:zeitgeist.fts:Search '(v*) AND (0..9223372036854775807ms)' gave 4422 hits in 33.2410335541ms

Talis:
DEBUG:zeitgeist.fts:Search 'v*' gave 8845 hits in 14.8799419403ms

And yes, I'm quite sure I don't have 4000 events with Timestamp < 0 in my DB.

Revision history for this message
Mikkel Kamstrup Erlandsen (kamstrup) wrote :

Can you try executing the generated queries directly in talis? Ie.

./talis '(v*) AND (())'

etc... Maybe you're not seeing the entire generated query or something...

Revision history for this message
Michal Hruby (mhr3) wrote :

It seems that this issue was caused by the empty Event (the "(())"), if I pass instead empty array of Events, everything is fine.

Michal Hruby (mhr3)
Changed in zeitgeist-extensions:
status: New → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.