Mailing-List: contact dev-help@openejb.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@openejb.apache.org
Received-SPF: pass (athena.apache.org: domain of david.blevins@visi.com
 designates 208.42.176.212 as permitted sender)
Message-Id: <00216768-56D3-4A7E-9F67-7190613FED31@visi.com>
From: David Blevins <david.blevins@visi.com>
To: dev@openejb.apache.org
In-Reply-To: <1f3854d50910151610j1019dbb3sac7902c8e3cb91fe@mail.gmail.com>
Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
Content-Transfer-Encoding: 7bit
Mime-Version: 1.0 (Apple Message framework v936)
Subject: Re: TimerService from EJB 3.1
Date: Thu, 15 Oct 2009 17:03:13 -0700
References: <1f3854d50910130658vbd1573al86b4a0d0826f97dd@mail.gmail.com>
  <25889028.post@talk.nabble.com>
 <E3CB9855-D191-4572-A8E1-2405A198F9DF@visi.com>
  <1f3854d50910150106q6b15abe5wef9a2253e4386b07@mail.gmail.com>
  <4ad6e5e0.9653f10a.74f2.6782@mx.google.com>
 <3B554C31-6260-4355-B142-CB980DBA91C6@visi.com>
 <1f3854d50910151610j1019dbb3sac7902c8e3cb91fe@mail.gmail.com>


On Oct 15, 2009, at 4:10 PM, Quintin Beukes wrote:

> I'd love to have a look at this. Developing code is the ultimate
> "course" (in the academic sense) in the software being developed.
>
> How exactly would you want it to change? How I understand it something
> similar to anacron, in that tasks should somehow realize that weren't
> run and then run one after the other when OpenEJB becomes live? Some
> thought would have to go into how to keep track of these tasks. But
> this is the general idea right?

That's the general idea.  We have a DatabaseTimerStore class with most  
of the logic, but that class isn't getting used anywhere currently and  
I don't think it's quite complete.  I didn't work on that code so  
can't say offhand what is missing.

I recommend starting there.

-David

> On Fri, Oct 16, 2009 at 12:22 AM, David Blevins <david.blevins@visi.com 
> > wrote:
>>
>> On Oct 15, 2009, at 2:05 AM, John D. Ament wrote:
>>
>>> Well cron style works as well, with an interesting little caveat  
>>> due to
>>> the fact that any timers that went off while the app server were  
>>> down should
>>> be run when restarted during a failure.  I'm working on a project  
>>> around
>>> this principal actually, converting a bunch of old perl scripts to  
>>> something
>>> a bit more modern.  We ended up using a standalone process w/  
>>> openejb and
>>> created a scheduler based on the scripts.  It works good, but when  
>>> I start
>>> the container all of the jobs kick off, but then resume typical  
>>> operation.
>>>
>>> Ironically, it's exactly what the spec describes to be what should  
>>> happen.
>>
>> Right, we're definitely open to some contributions there if anyone  
>> wanted to
>> take a stab at it.
>>
>> We always have an open door policy regardless, but this area in  
>> particular
>> would be a nice spot to dig in.
>>
>> -David
>>
>>
>