FTS doesn't work properly with *Subject groupings

Bug #947835 reported by Michal Hruby on 2012-03-06
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Zeitgeist Framework
Fix Released
High
Michal Hruby

Bug Description

Currently FTS is only approximating results when any of the *Subject groupings are used. This means that asking FTS for 20 most recent subjects containing "a" can right now return anything between 1 and 60 different subjects all depending on the order of recent events.

The idea here is to use xapian's collapse key functionality to be able to get rid of this approximation while doing performance and memory requirement measurements of its impact.

Related branches

Michal Hruby (mhr3) on 2012-03-06
Changed in zeitgeist:
status: New → Triaged
importance: Undecided → High
assignee: nobody → Michal Hruby (mhr3)
Michal Hruby (mhr3) wrote :

Adding massif profiles where one is current FTS and the other does the collapsing (using md5-hashes the uris, see RainCT's branch)

Michal Hruby (mhr3) wrote :
Michal Hruby (mhr3) wrote :

Although I didn't do real benchmarking, but I don't really much difference in query times.

Michal Hruby (mhr3) wrote :

Disk space usage (with ~60k events):
- original: 77.8MB
- with hashes: 79.7MB

Michal Hruby (mhr3) wrote :

All in all I think this is good to go, I'm not really sure whether to trust the massif profiles (basically they are the same both with and without hashes), because gnome-system-monitor shows something different, but perhaps it's the sysmon that's not to be trusted. :)

Changed in zeitgeist:
status: Triaged → In Progress
milestone: none → 0.9.0

Let's hack all zeitgeist in c++

On Tue, Mar 6, 2012 at 11:27 AM, Michal Hruby <email address hidden>wrote:

> All in all I think this is good to go, I'm not really sure whether to
> trust the massif profiles (basically they are the same both with and
> without hashes), because gnome-system-monitor shows something different,
> but perhaps it's the sysmon that's not to be trusted. :)
>
> ** Changed in: zeitgeist
> Status: Triaged => In Progress
>
> ** Changed in: zeitgeist
> Milestone: None => 0.9.0
>
> --
> You received this bug notification because you are subscribed to The
> Zeitgeist Project.
> https://bugs.launchpad.net/bugs/947835
>
> Title:
> FTS doesn't work properly with *Subject groupings
>
> Status in Zeitgeist Framework:
> In Progress
>
> Bug description:
> Currently FTS is only approximating results when any of the *Subject
> groupings are used. This means that asking FTS for 20 most recent
> subjects containing "a" can right now return anything between 1 and 60
> different subjects all depending on the order of recent events.
>
> The idea here is to use xapian's collapse key functionality to be able
> to get rid of this approximation while doing performance and memory
> requirement measurements of its impact.
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/zeitgeist/+bug/947835/+subscriptions
>

Michal Hruby (mhr3) on 2012-03-15
Changed in zeitgeist:
status: In Progress → Fix Committed
Seif Lotfy (seif) on 2012-03-20
Changed in zeitgeist:
status: Fix Committed → Fix Released
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers