Indexing associations

[Jim Ruther Nill]

Hi!

in my define_index block inside the Task model i have a where clause

where 'project.status != "closed"'

but everytime I change the project status, I have to reindex else the

search results will be wrong.  Is there any better way to do this?  I have

delayed delta indexing enabled for tasks.  I wonder if I also need to do

this for projects.

Thanks!

[Pat Allan]

Hi Jim

I guess what you'll want to do is have a after_commit hook on your project model that updates the delta flags on all tasks for that project (which will in turn fire the delta callback for them). This will ensure Sphinx's data is close to up-to-date.

How many tasks are generally tied to a project?

--
Pat

> --
> You received this message because you are subscribed to the Google Groups "Thinking Sphinx" group.
> To post to this group, send email to thinkin...@googlegroups.com.
> To unsubscribe from this group, send email to thinking-sphi...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/thinking-sphinx?hl=en.

[Jim Ruther Nill]

Hi Pat,

Thanks for getting back to me.  There's no limit to the number of tasks per project.

A good estimate would be around 50 tasks per project.  Does this mean that

everytime the project changes status, i have to update those 50 tasks? Sorry but

does that also mean that the delayed delta will queue 50 jobs to delayed_job?

Thanks!

Jim

--
-------------------------------------------------------------
visit my blog at http://jimlabs.heroku.com

[Pat Allan]

At a basic level, you'd need to update all of those tasks, but jobs will not be duplicated (however, if a job is completed before all 50 tasks are updated, then another would be added).

There is a better way, of course - it's just a little more fiddly.

Firstly, update the delta flags in one SQL statement (assuming this is done within the context of a project object):

Task.update_all({:delta => true}, {:project_id => id})

And then, let's just manually create the delta job:

ThinkingSphinx::Deltas::Job.enqueue(
ThinkingSphinx::Deltas::DeltaJob.new(Task.delta_index_names),
ThinkingSphinx::Configuration.instance.delayed_job_priority
)

That should be the best approach from a performance perspective. Let us know how it goes :)

--
Pat

[Jim Ruther Nill]

On Sun, Nov 20, 2011 at 1:47 PM, Pat Allan <p...@freelancing-gods.com> wrote:

At a basic level, you'd need to update all of those tasks, but jobs will not be duplicated (however, if a job is completed before all 50 tasks are updated, then another would be added).

There is a better way, of course - it's just a little more fiddly.

Firstly, update the delta flags in one SQL statement (assuming this is done within the context of a project object):

 Task.update_all({:delta => true}, {:project_id => id})

And then, let's just manually create the delta job:

 ThinkingSphinx::Deltas::Job.enqueue(
   ThinkingSphinx::Deltas::DeltaJob.new(Task.delta_index_names),
   ThinkingSphinx::Configuration.instance.delayed_job_priority
 )

That should be the best approach from a performance perspective. Let us know how it goes :)

cool. Thanks Pat! I'll let you know how it goes when I try it tomorrow. Thanks!

[Jim Ruther Nill]

On Sun, Nov 20, 2011 at 6:33 PM, Jim Ruther Nill <jvn...@gmail.com> wrote:

On Sun, Nov 20, 2011 at 1:47 PM, Pat Allan <p...@freelancing-gods.com> wrote:

At a basic level, you'd need to update all of those tasks, but jobs will not be duplicated (however, if a job is completed before all 50 tasks are updated, then another would be added).

There is a better way, of course - it's just a little more fiddly.

Firstly, update the delta flags in one SQL statement (assuming this is done within the context of a project object):

 Task.update_all({:delta => true}, {:project_id => id})

And then, let's just manually create the delta job:

 ThinkingSphinx::Deltas::Job.enqueue(
   ThinkingSphinx::Deltas::DeltaJob.new(Task.delta_index_names),
   ThinkingSphinx::Configuration.instance.delayed_job_priority
 )

That should be the best approach from a performance perspective. Let us know how it goes :)

cool. Thanks Pat! I'll let you know how it goes when I try it tomorrow. Thanks!

This is working perfectly Pat. Thanks again!