Ideas and discussion about tracking behavior

[jjb]

Ideas and discussion about tracking behavior

[jmettraux]

Is there some kind of fallback scenario where that value might be true instead of the job instance?

Hello,

It's not a fallback, it's a temporary true value, to pre-reserve before assignation. IIRC.

https://github.com/jmettraux/rufus-scheduler/blob/master/lib/rufus/scheduler/jobs_core.rb#L311-L312

[jmettraux]

"losing data" sounds scary.

There is also the case where the thread use by the job just ended and it pops in your loop because when the .threads(:active) array was established it serviced a job, but when you consider it, it might have ended. I would do something like

  schedule.work_threads(:active).each do |t|
    i = t[:rufus_scheduler_job].identifier rescue nil
    thread_usage[i] += 1 if i
  end

That should cover the nil and the true extra cases.

[jmettraux]

Job#name

rufus-scheduler/spec/job_spec.rb

Lines 389 to 397 in 64591be

	describe '#name' do
	it 'returns the job name' do
	j = @scheduler.schedule_in '10d', name: 'alice' do; end
	expect(j.name).to eq('alice')
	end
	end

[jmettraux]

Hello,

I have the impression the hacks are better at the level of your ruby-clock gem.

Regarding the name, how about

class Rufus::Scheduler::Job
  def full_name
    [ name, source_location.map(&:to_s).join(':') ].join(' ')
  end
end

which returns something like "alice /home/jmettraux/w/rufus-scheduler/spec/job_spec.rb:408" ?

It uses the #source_location method I added to Job.

Best regards.

[jjb]

wow cool! thanks!

do you know if there's a way to access the job from schedule.work_threads(:active).first?

[jmettraux]

Hello,

yes, you can do

scheduler.work_threads(:active).each_with_index do |t, i|
  puts "#{i}. work thread #{t.id} is active for job #{t[:rufus_scheduler_job].source_location}"
end

Is that sufficient?

[jjb]

this is perfect - this code gives me exactly what i wanted

  job_code = schedule.work_threads(:active).map do |t|
    t[:rufus_scheduler_job].handler.source.split("\n")[1].strip
  end

[jmettraux]

OK, give me one or two days to release with Job#source_location, then I will close this issue.

Best regards.

[jjb]

Thanks, for now I don't need it for what I'm working on, so no time pressure here. I'll let you know what I come up with.

[jjb]

Okay, I've made Rufus::Scheduler::Job#identifier jjb/ruby-clock#2

I'm going to implement this at work first to see if I want to update anything, so it will get merged and released in the next few days. Any feedback welcome!

---

edit: I did actually end up depending on the new source_location feature!

[jjb]

just got a confusing error message from this code:

  schedule.work_threads(:active).each do |t|
    thread_usage[t[:rufus_scheduler_job].identifier] += 1
  end

undefined method `identifier' for true:TrueClass

it happened exactly once when my clock process started. Now that loop, which runs every 10 seconds, seems to be happy and functional.

Is there some kind of fallback scenario where that value might be true instead of the job instance?

[jjb]

Gotcha, thanks! So it's only when the thread is starting up. so i think i can safely just next along when i see that and not risk losing any data

[jmettraux]

3.8.0 released

Thanks!

[jjb]

great! and thanks for the extra info about missing jobs