Remove the upper limit on extracted text
Bug #1365493 reported by
Paul Everitt
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
KARL4 |
New
|
Wishlist
|
Chris Rossi |
Bug Description
As Chris noted in lp:1364383 the contextual summary expects to have the extracted text around. We just removed the extracted text in some cases in lp:1340295.
I'm putting this as a low priority, put in October, as it isn't clear that many search results will hit the small number of objects that no longer have extracted text. This conclusion presumes that LiveSearch does not call contextual summaries. If I'm wrong about that, or you disagree that this won't be triggered much, tell me.
Once we do decide to do it, perhaps we can consider some mild zlib compression on the extracted text, as a defense against database bloat.
affects: | karl3 → karl4 |
Changed in karl4: | |
milestone: | m142 → none |
Changed in karl4: | |
milestone: | none → 003 |
Changed in karl4: | |
milestone: | 003 → 999 |
To post a comment you must log in.
FWIW, we're already doing the zlib compression. I thought the contextual
summaries were in the LiveSearch but just looked and it looks like they
aren't, so that's good.
Generally speaking, this will just mean that in the somewhat rare cases
that a document without the extracted text cache shows up in a search
results listing, the text will be re-extracted--so it will contribute to
slowness but not breakage, except in pathological cases like lp:1364383
where the extractor hangs on a particular document.
On Thu, Sep 4, 2014 at 9:02 AM, Paul Everitt <email address hidden> wrote:
> Public bug reported: archimedeanco) /bugs.launchpad .net/bugs/ 1365493 /bugs.launchpad .net/karl3/ +bug/1365493/ +subscriptions
>
> As Chris noted in lp:1364383 the contextual summary expects to have the
> extracted text around. We just removed the extracted text in some cases
> in lp:1340295.
>
> I'm putting this as a low priority, put in October, as it isn't clear
> that many search results will hit the small number of objects that no
> longer have extracted text. This conclusion presumes that LiveSearch
> does not call contextual summaries. If I'm wrong about that, or you
> disagree that this won't be triggered much, tell me.
>
> Once we do decide to do it, perhaps we can consider some mild zlib
> compression on the extracted text, as a defense against database bloat.
>
> ** Affects: karl3
> Importance: Low
> Assignee: Chris Rossi (chris-
> Status: New
>
> --
> You received this bug notification because you are a bug assignee.
> https:/
>
> Title:
> Remove the upper limit on extracted text
>
> Status in KARL3:
> New
>
> Bug description:
> As Chris noted in lp:1364383 the contextual summary expects to have
> the extracted text around. We just removed the extracted text in some
> cases in lp:1340295.
>
> I'm putting this as a low priority, put in October, as it isn't clear
> that many search results will hit the small number of objects that no
> longer have extracted text. This conclusion presumes that LiveSearch
> does not call contextual summaries. If I'm wrong about that, or you
> disagree that this won't be triggered much, tell me.
>
> Once we do decide to do it, perhaps we can consider some mild zlib
> compression on the extracted text, as a defense against database
> bloat.
>
> To manage notifications about this bug go to:
> https:/
>