[Quickfix-developers] Fwd: C# QuickFIX

[Oren M. <or...@qu...>]

Here are more details on Reuters analysis on the .NET vs. C++ API's.   
To my knowledge the java API does not have a similar instability, but  
performance will probably be comparable.

--oren

Begin forwarded message:

> From: Yihu Fang <Yih...@re...>
> Date: May 7, 2004 8:44:51 AM CDT
> To: or...@qu..., jc...@co...
> Subject: FW: C# QuickFIX
>
> Oren,
>
> I redid the C# test. The code is a simple stress test - basically there
> is a while loop in Application::run() which reads each FIX message from
> the MSMQ and then Session.sendToTarget().
>
> When we have large number of messages (50000) in the queue, the test
> code starts to choke and throws NULL exceptions (mostly) as well as
> other exceptions. If we change the directive [STAThread] to [MTAThread]
> for better multithread garbage collection, it still happens. And if we
> use ThreadedSocketInitiator instead of the single thread class, it
> sometimes got worse. It happens on both Release and Debug code. It is
> QuickFIX 1.6 (it was done 2 months ago and since we have moved C++ base
> to 1.7).
>
> As for performance, I ran both QF endpoints in C++ code, it can pump
> 50000 messages in 30+ seconds on my laptop (Thinkpad T40). If I choose
> one end as C# code and one end as C++, it at least takes twice as long
> to complete (with some exceptions thrown in C#), and the C# process
> takes 3 times as many as resources as the C++ code (memory and CPU).
>
> I include some typical exception stack as following. Hope it helps. The
> problem for us is its unpredictability in C#.
>
> Regards,
>
> -Yihu
> ----------------------------------------------------------------------- 
> -
> -
> Exception(System.NullReferenceException): Object reference not set to  
> an
> instance of an object.   at FIX.FieldMap.=(FieldMap* , FieldMap* )
>    at FIX.Message.=(Message* , Message* )
>    at Application.toApp(Application* , Message* message, SessionID*
> sessionID)
>    at FIX.Session.sendToTarget(Message* )
>    at QuickFix.Session.sendToTarget(Message message)
>    at tradeclient_csharp.Application.run() in
> c:\csharp\tradeclient_csharp\application.cs:line 132
> 8=FIX.4.0?9=78?35=D?49=FIXTEST?56=TW?11=123?21=1?38=456?40=2?44=789? 
> 54=1
> ?55=IBM?59=0?128=FID?10=101?
>
> Exception(System.Runtime.InteropServices.SEHException): External
> component has thrown an exception.
>    at FIX.FieldMap.=(FieldMap* , FieldMap* )
>    at FIX.Message.=(Message* , Message* )
>    at Application.toApp(Application* , Message* message, SessionID*
> sessionID)
>    at FIX.Session.sendToTarget(Message* )
>    at QuickFix.Session.sendToTarget(Message message)
>    at tradeclient_csharp.Application.run() in
> c:\csharp\tradeclient_csharp\application.cs:line 132
> 8=FIX.4.0?9=78?35=D?49=FIXTEST?56=TW?11=123?21=1?38=456?40=2?44=789? 
> 54=1
> ?55=IBM?59=0?128=FID?10=101?
>
> Exception(System.Runtime.InteropServices.SEHException): External
> component has thrown an exception.
>    at std.basic_string<char,std::char_traits<char>,std::allocator<char>
>> .__ctor(basic_string<char,std::char_traits<char>,std::allocator<char>
>> * , basic_string<char,std::char_traits<char>,std::allocator<char> >* )
>    at FIX.FieldBase.__ctor(FieldBase* , FieldBase* )
>    at
> std._Tree_nod<std::_Tmap_traits<int,FIX::FieldBase,FIX:: 
> message_order,st
> d::allocator<std::pair<int const ,FIX::FieldBase> >,1>
>> ._Node.__ctor(_Node* , _Node* _Larg, _Node* _Parg, _Node* _Rarg,
> pair<int const ,FIX::FieldBase>* _Val, SByte _Carg)
>    at
> std._Tree<std::_Tmap_traits<int,FIX::FieldBase,FIX:: 
> message_order,std::a
> llocator<std::pair<int const ,FIX::FieldBase> >,1>
>> ._Buynode(_Tree<std::_Tmap_traits<int,FIX::FieldBase,FIX:: 
>> age_order,std
> ::allocator<std::pair<int const ,FIX::FieldBase> >,1> >* , _Node*  
> _Larg,
> _Node* _Parg, _Node* _Rarg, pair<int const ,FIX::FieldBase>* _Val,  
> SByte
> _Carg)
>    at FIX.FieldMap.=(FieldMap* , FieldMap* )
>    at FIX.Message.=(Message* , Message* )
>    at Application.toApp(Application* , Message* message, SessionID*
> sessionID)
>    at FIX.Session.sendToTarget(Message* )
>    at QuickFix.Session.sendToTarget(Message message)
>    at tradeclient_csharp.Application.run() in
> c:\csharp\tradeclient_csharp\application.cs:line 132
> 8=FIX.4.0?9=78?35=D?49=FIXTEST?56=TW?11=123?21=1?38=456?40=2?44=789? 
> 54=1
> ?55=IBM?59=0?128=FID?10=101?
>
>
>
>
> -----Original Message-----
> From: Yihu Fang
> Sent: Thursday, May 06, 2004 3:14 PM
> To: 'or...@qu...'
> Cc: jc...@co...
> Subject: RE: C# QuickFIX
>
> Oren,
>
> What we saw is that the QuickFix .NET throws null exceptions
> occasionally on some of the messages when we pump large amount of
> message (like 10000) without throttling through the engine. The message
> is one identical single order message. We were using VS.NET 2003.
>
> We are moving to QuickFIX C++ implementation with VC7 compiler. I will
> try to reproduce the problems when I get chance.
>
> Regards,
>
> -Yihu
>
> -----Original Message-----
> From: or...@qu... [mailto:or...@qu...]
> Sent: Thursday, May 06, 2004 2:04 PM
> To: Yihu Fang
> Cc: jc...@co...
> Subject: C# QuickFIX
>
> Yihu,
>
> At the SIMC conference you mentioned that you guys had problems with  
> the
> QuickFIX .NET API.  I was wondering if
> you can share any more details about this.  Were the problems you had
> encountered just stability problems or
> were there also performance issues?  We are getting more and more
> inquiries about the .NET interface and I
> would just like to gather as much useful information as I can.  Also
> when you were doing your testing was that
> with VS.NET 2002, or 2003?  Thanks.
>
> --oren
>
>
>
> -----------------------------------------------------------------
>         Visit our Internet site at http://www.reuters.com
>
> Get closer to the financial markets with Reuters Messaging - for more
> information and to register, visit http://www.reuters.com/messaging
>
> Any views expressed in this message are those of  the  individual
> sender,  except  where  the sender specifically states them to be
> the views of Reuters Ltd.
>
>