ws-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Andreas Veithen <andreas.veit...@gmail.com>
Subject Re: [VOTE] Release Axiom 1.2.13
Date Wed, 04 Apr 2012 21:14:17 GMT
I found one other report [1] of that issue, but it seems to be very
sporadic and not reproducible. That was more than one year ago which
means that it is not a regression in 1.2.13. I regularly build with
Java 6 and also sometimes without Internet connection, so it's not
related to these two factors.

After reviewing the code I think that the problem comes from a piece
of code that was introduced in 2008 by IBM people and that is supposed
to limit the concurrency in one of the code paths and to dynamically
adjust the file buffering threshold based on current memory usage. I
think that in some cases this gives a negative threshold which
actually forces buffering in memory. Since the algorithm uses heap
memory metrics (max, allocated, free) as variables it is highly
environment dependent and not deterministic. All this is compatible
with the observations made above.

Anyway, I think that the code was tuned for a specific product
(WebSphere) and this kind of code should not exist in a library such
as Axiom that can be used in various different contexts. Actually I
was already looking for some time for an excuse to get rid of that
code and now I have one ;-)

Finally, since it is not a regression, I think it should not block the
1.2.13 release.

Thanks for bringing this to my attention!

Andreas

[1] http://markmail.org/thread/eclcydt6kq3ygcot

On Tue, Apr 3, 2012 at 21:09, Daniel Kulp <dkulp@apache.org> wrote:
>
> +1
>
> Everything looks OK from a Legal standpoint.
>
> Everything builds and passes with Java 5.  I'm seeing 3 test failures in api
> with Java 6:
>
> Failed tests:
> testHeaderGetSet(org.apache.axiom.attachments.PartOnFileTest): Expected
> FileDataSource, but got org.apache.axiom.attachments.PartDataSource
>  testPurgeDataSource(org.apache.axiom.attachments.AttachmentsTest):
> expected:<1> but was:<0>
>  testReadOnceOnBufferedPart(org.apache.axiom.attachments.AttachmentsTest):
> expected:<1> but was:<0>
>
> I'm not sure how big of an issue that is to you.  It could be something
> related to the office I'm in right now.   Having major network issues so if
> the tests are doing anything network related, that could easily explain it.
>
>
> (note:  we don't have java 6 builds in Jenkins for Axiom, should we?)
>
>
> Dan
>
>
>
>
> On Monday, April 02, 2012 10:48:22 PM Andreas Veithen wrote:
>> Here is the vote for the Axiom 1.2.13 release.
>>
>>
>> Release notes:
>>
>> https://svn.apache.org/repos/asf/webservices/commons/tags/axiom/1.2.13/REL
>> EASE-NOTE.txt
>>
>> The Maven staging repository is at:
>>
>> https://repository.apache.org/content/repositories/orgapachews-003/
>>
>> Source and binary packages can be found at:
>>
>> https://repository.apache.org/content/repositories/orgapachews-003/org/apa
>> che/ws/commons/axiom/axiom/1.2.13/
>>
>> Preview of the Maven site:
>>
>> http://people.apache.org/~veithen/axiom/
>>
>>
>> The vote will be open for 72 hours.
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@ws.apache.org
>> For additional commands, e-mail: dev-help@ws.apache.org
> --
> Daniel Kulp
> dkulp@apache.org - http://dankulp.com/blog
> Talend Community Coder - http://coders.talend.com
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@ws.apache.org
For additional commands, e-mail: dev-help@ws.apache.org


Mime
View raw message