Cyrille Giquello wrote:

> hello Marcel, 

olleh elliryC,

>
>
> A long time ago, I did some test with big messages, and the problem 
> was that xmlBlaster blocking while message transfering, so no new 
> messages can arrive before the big message transfert was finished ...
>
> What do  you think ?? 

I think it is not very nice!

I think (i have not tried) that this should not
happen with corba, as we have the 'one Thread per request'
policy switched on.
I don't know how XmlRpc is implemented, but the
current SOCKET protocol is 'one thread per client'
so here your observation should be correct.
(Probably i should add a thread pool to the socket protocol as well)

Could you please take a copy of

   xmlBlaster/testsuite/org/xmlBlaster/stress/BigMessage.java

name it something like CyrilleBigMessageAndASmallOne.java

and try to add a CORBA testcase showing this problem?

I'm currently trying to build a release 0.80 and probably
your case should be addressed in this release.

thanks,

Marcel

>
>
> Cyrille
>
>
> Marcel Ruff wrote:
>
>> Date:    Wednesday September 11, 2002 @ 20:34
>> Author:    ruff
>>
>> Update of /opt/cvsroot/xmlBlaster/testsuite/org/xmlBlaster/stress
>> In directory server:/tmp/cvs-serv15995
>>
>> Added Files:
>>     BigMessage.java Makefile Log Message:
>> Added testcase for a 2MByte message
>>
>>  
>>
>
>
>



Reply via email to