Re: Deadlock issue

Sunday, April 22, 2012, 8:21:38 AM, Jacopo Cappellato wrote:

>
> On Apr 20, 2012, at 6:23 PM, Daniel Dekany wrote:
>
>> Friday, April 20, 2012, 2:33:44 PM, Jacopo Cappellato wrote:
>> 
>>> Hi Daniel,
>>> 
>>> On Apr 20, 2012, at 2:21 PM, Daniel Dekany wrote:
>>> 
>>>> Friday, April 20, 2012, 10:37:52 AM, Jacopo Cappellato wrote:
>>>> 
>>>>> Daniel D., Daniel L.,
>>>>> 
>>>>> thanks to both of you for the valuable help.
>>>>> 
>>>>> I was able to find and fix the OFBiz code that was causing the
>>>>> multiple reload of Groovy related classes: in OFBiz there was cache
>>>>> object containing the parsed Groovy scripts that was not managed in
>>>>> a thread safe way.
>>>> 
>>>> So then can it still occur that a Groovy-script-class is reloaded even
>>>> if the script wasn't modified, only not so often?
>>> 
>>> It should no more happen now: the OFBiz cache object is now
>>> properly synchronized and even if multiple threads will try to
>>> compile the same script only the first Groovy-script-class will be
>>> added to the cache and then used to renderthe freemarker template.
>>> After this change all "reload" messages from freemarker disappeared.
>> 
>> What I meant is that if these are in-memory-only classes, the Groovy
>> script cache might drops some of them sometimes. If it's that kind of
>> cache.
>
> Oh yes, when you clear the cache in OFBiz

Which, I hope, doesn't happen often.

> then Freemarker will reload its caches; however, with the fix I did
> in OFBiz, Freemarker will now reload the cache only once while
> before it could reload it several times (we had a bug in OFBiz that
> prevented properly handling of concurrency and the same class was
> compiled and added to the cache several times).
>
> We are testing now the fixes in production and they seem to have solved the problem, so thank you!
>
> As a side note: I would like to propose/submit some patches to
> improve the implementation of synchronized code in BeansWrapper and
> so I am wondering if it would make sense to sign the CLA; I am an
> active committer (and I am in the OFBiz PMC) of OFBiz and we are
> using Freemarker extensively so it could be useful also in the
> future (if we will ever find a bug). In the same time I don't want
> to put on your shoulders unnecessary paperwork, so please disregard
> my proposal if you feel that a CLA is an overkill.

This paper work is not a big deal for us. I will email the CLA form.

> Kind regards,
>
> Jacopo

-- 
Best regards,
 Daniel Dekany


------------------------------------------------------------------------------
For Developers, A Lot Can Happen In A Second.
Boundary is the first to Know...and Tell You.
Monitor Your Applications in Ultra-Fine Resolution. Try it FREE!
http://p.sf.net/sfu/Boundary-d2dvs2