Using NMS to connect to OpenMQ via stomp

[matt walsh <ma...@gmail.com>]

Hi All,

I have had a problem connecting to OpenMQ via stomp using Apache NMS. It
seems the NMS code terminates stomp frames using a 'NULL' character (as per
the stomp spec), which is defined in the class StompFrameStream as follows

public const char NULL = (char) 0;

OpenMQ was giving me errors because it seems from my investigations that
this 'NULL' is infact a double byte character and is therefore sending 2
zeros at the end of a frame. This leaves an extra NULL lying around in the
stream that OpenMQ is unhappy with, breaking further commands sent for a
session.

I have patched the NMS source to define NULL as a byte, ie

public const byte NULL = (byte) 0;

This makes the problem go away for OpenMQ.

I would like some advice on this. Is it acceptable for a Stomp frame to have
additional NULL characters lying around as NMS seems to be doing (and thus
is this a problem that should be dealt with by OpenMQ) or should NMS be
patched in a way similar to what I have done here?

I have attached my patched version of the 1.0 version of StompFrameStream.
http://www.nabble.com/file/p24196808/StompFrameStream.cs StompFrameStream.cs 
-- 
View this message in context: http://www.nabble.com/Using-NMS-to-connect-to-OpenMQ-via-stomp-tp24196808p24196808.html
Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.

Re: Using NMS to connect to OpenMQ via stomp

[Hiram Chirino <ch...@gmail.com>]

Seems to me that 2 nulls is an error.  The normal case, is just 1 null being
sent.  I would open it up as an issue.

On Thu, Jun 25, 2009 at 12:29 AM, matt walsh <ma...@gmail.com>wrote:

>
> Hi All,
>
> I have had a problem connecting to OpenMQ via stomp using Apache NMS. It
> seems the NMS code terminates stomp frames using a 'NULL' character (as per
> the stomp spec), which is defined in the class StompFrameStream as follows
>
> public const char NULL = (char) 0;
>
> OpenMQ was giving me errors because it seems from my investigations that
> this 'NULL' is infact a double byte character and is therefore sending 2
> zeros at the end of a frame. This leaves an extra NULL lying around in the
> stream that OpenMQ is unhappy with, breaking further commands sent for a
> session.
>
> I have patched the NMS source to define NULL as a byte, ie
>
> public const byte NULL = (byte) 0;
>
> This makes the problem go away for OpenMQ.
>
> I would like some advice on this. Is it acceptable for a Stomp frame to
> have
> additional NULL characters lying around as NMS seems to be doing (and thus
> is this a problem that should be dealt with by OpenMQ) or should NMS be
> patched in a way similar to what I have done here?
>
> I have attached my patched version of the 1.0 version of StompFrameStream.
> http://www.nabble.com/file/p24196808/StompFrameStream.csStompFrameStream.cs
> --
> View this message in context:
> http://www.nabble.com/Using-NMS-to-connect-to-OpenMQ-via-stomp-tp24196808p24196808.html
> Sent from the ActiveMQ - Dev mailing list archive at Nabble.com.
>
>


-- 
Regards,
Hiram

Blog: http://hiramchirino.com

Open Source SOA
http://fusesource.com/