Re: [Jepp-users] Simple test which shows memory leak in Jepp.

[Mike J. <mi...@pu...>]

Hi,

Sorry for the delayed response.

So, I didn't see a memory leak when I ran it on x64 Linux with Java6u18.
Which JVM and Python are you using?

I see:
Total data in memory after closing = 0

Thanks,
Mike

On Fri, Jan 7, 2011 at 5:33 AM, Matthew GERRING <mat...@es...>wrote:

>  Hello jepp-users,
>
> There is a simple test here:
> http://www.edna-site.org/svn/trunk/dawb/org.dawb.common.util/test/org/edna/common/util/test/JepTest.java
>
> Which shows memory leaks when using Jepp. The test
>
> testMemoryLeakJep
>
>  is the main one, it is using the set(String, Object) method from Java to
> python.
>
> I wondered if it is possible to release the jarray object in the JNI as it
> seems that Java cannot release the array once passed into Python jepp using
> set(...).
>
> Has anyone seen this too? Is there already a workaround or do I have to
> recompile the JNI some how?
>
> Thanks,
>
> Matt
>
> On 07/01/2011 11:31, Jon Wright wrote:
>
> Hi Matt,
>
> The best person is probably Mike Johnson who is the author of Jepp. He was
> usually very helpful with queries in the past when we posted on the jepp
> users mailing list (jep...@li...). The C code is
> relatively straightforward in case he doesn't come up with a quick fix, even
> I've modified it in the past...
>
> Cheers
>
> Jon
>
> On 07/01/2011 11:15, Matthew GERRING wrote:
>
> Jon,
>
> I need to check this further but I think that it is possible to release
> the jarray in the JNI, thus removing the leak. Whom should I talk to
> about doing this? I am not an expert in C/JNI.
>
> Best Regards,
>
> Matt
>
> On 07/01/2011 09:45, Jon Wright wrote:
>
> Hi Matthew,
>
> Prior to jepp we tried using xmlrpc to connect python with java, but
> this had a lot of disadvantages too. Shared memory was previously
> ruled out. Is there some other option nowadays?
>
> The tests look great for tracking down and fixing the leak :-)
>
> See you later
>
> Jon
>
>
> On 07/01/2011 09:10, Matthew GERRING wrote:
>
> Hi,
>
> Recently I found a memory leak with Jepp. The very specific way that
> fabio uses Jepp is probably ok (based on some fable GUI tests I wrote
> comparing diamond with fable plotting) but some of the other method
> calls which I need for the workflow have a memory leak.
>
> Also did some thread testing and testing with Numpy. Due to the memory
> leak I did not get as far as checking the threading more fully.
>
> All the tests are here:
>
> http://www.edna-site.org/svn/trunk/dawb/org.dawb.common.util/test/org/edna/common/util/test
>
>
>
> You can check out
> http://www.edna-site.org/svn/trunk/dawb/org.dawb.common.util as a
> project.
>
> I have yet to draw firm conclusions on these tests but at the moment it
> looks as if the memory leak will mean that jepp is not reliable.
>
> Best Regards,
>
>
> Matthew Gerring
>
>
> On 06/01/2011 17:35, Jon Wright wrote:
>
> Hi Andy, Matthew,
>
> Here's where things were last time I looked at Jepp and windows 64 bit:
>
>
> http://sourceforge.net/mailarchive/forum.php?forum_name=jepp-users&max_rows=25&style=nested&viewmonth=200907
>
>
>
> It might turn out to be resolved now, but there was some hairy stuff
> going on related to threading and multiple interpreter instances.
>
> See you tomorrow,
>
> Jon
>
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> Gaining the trust of online customers is vital for the success of any
> company
> that requires sensitive data to be transmitted over the Web.   Learn how to
> best implement a security strategy that keeps consumers' information secure
> and instills the confidence they need to proceed with transactions.
> http://p.sf.net/sfu/oracle-sfdevnl
> _______________________________________________
> Jepp-users mailing list
> Jep...@li...
> https://lists.sourceforge.net/lists/listinfo/jepp-users
>
>