The Summary of Recent Important Changes

[Trustin Lee <tr...@gmail.com>]

Hi all,

I made a few important changes recently.  There's the summary:

* ByteBufferAllocator - http://issues.apache.org/jira/browse/DIRMINA-176

Now you can implement your own ByteBuffer pool.  The default pool
implementation has a timeout so it decreases the total size of the pool when
the buffers are not used for a long time.

* CompressionFilter -
http://issues.apache.org/jira/browse/DIRMINA-179(Thanks to Vinod)

Vinod contributed a compression filter based on JZlib.  He says LZW support
will follow.

* ThreadModel - http://issues.apache.org/jira/browse/DIRMINA-190

You don't have to add a ThreadPoolFilter anymore. The default thread model
is PooledThreadModel, and you can change it as you want.  To disable this
feature, set the thread model to ThreadModel.MANUAL.  Unless you change your
code, there will be inefficiency that events propagates via two thread pool
filters; one from PooledThreadModel, the other from your code.

Now we don't miss ServiceRegistry at all, right?  :D

Although I fixed quite many issues today, there are two more issues left
till we release 0.9.3:

* ByteBuffer.duplicate() problem -
http://issues.apache.org/jira/browse/DIRMINA-165
* SocketBindTest problem - http://issues.apache.org/jira/browse/DIRMINA-163

Please post your idea to the JIRA so we can accelerate the schedule.  Of
course, patches are welcome!

Cheers,
Trustin
--
what we call human nature is actually human habit
--
http://gleamynode.net/
--
PGP key fingerprints:
* E167 E6AF E73A CBCE EE41  4A29 544D DE48 FE95 4E7E
* B693 628E 6047 4F8F CFA4  455E 1C62 A7DC 0255 ECA6

Re: Strategies for managing write queue size in 0.8.2?

[Trustin Lee <tr...@gmail.com>]

Hi Steve,

On 3/14/06, Steve Bate <st...@technoetic.com> wrote:
>
> Hello,
>
> Again, this is version 0.8.2.
>
> I'm currently having a problem with a fast writer and
> slower reader causing the writer's memory consumption to
> grow rapidly. The memory consumption appears to be caused
> by the unbounded write buffer queue in the SocketSession.
>
> What strategies do you recommend to control the writer's
> memory usage in these scenarios?


If you know the average (or maximum) buffer size per message, then you can
calculate the total size of data from IoSession.getScheduledWriteRequests().
So you can delay writing messages considering it.  That's all for now.  For
more accurate calculation, we probably need to add
IoSession.getScheduledWriteBytes() in 0.9 stream.  You can track this issue
here:

http://issues.apache.org/jira/browse/DIRMINA-191

Trustin
--
what we call human nature is actually human habit
--
http://gleamynode.net/
--
PGP key fingerprints:
* E167 E6AF E73A CBCE EE41  4A29 544D DE48 FE95 4E7E
* B693 628E 6047 4F8F CFA4  455E 1C62 A7DC 0255 ECA6

Strategies for managing write queue size in 0.8.2?

[Steve Bate <st...@technoetic.com>]

Hello,

Again, this is version 0.8.2.

I'm currently having a problem with a fast writer and
slower reader causing the writer's memory consumption to
grow rapidly. The memory consumption appears to be caused
by the unbounded write buffer queue in the SocketSession.

What strategies do you recommend to control the writer's
memory usage in these scenarios?

Thanks,

Steve