Hi Josh,
I'll try to give you an idea of how QuickFIX handles these scenarios.
So we are running along and processing messages, sequence numbers are in
order, everything is great, then all of a sudden we get a sequence
number higher than expected.
Ok, so something has been missed. And it is a requirement of FIX that
we process all messages in order. What QuickFIX will do in this
scenario is send our a resend request for the missing range, and stick
the out of order message into an in memory queue. Any new messages that
come in will also be placed at the end of this queue.
At some point the other side will begin resending messages. Once all
missing messages have been processed, QuickFIX will immediately process
everything that is sitting in the queue. At this point we should be all
caught up and will be able to process new messages.
No, the other part of your question is for when you are resending
messages. Messages are stored in the MessageStore (you provide a
MessageStoreFactory in your initialization). There is a one to one
relationship between a session and a message store. This store will
store all messages that have been sent on that session until the session
expires. This gives you the capability to be able to resend any message
if it is requested. This is important if you are sending trade reports
and someone needs to re-request the days trades.
The scenario you pointed out is different. What if I sent out an order
2 minutes ago and they re-requests that. I do NOT want that order to go
out anymore. We thought of that case as well. For this you need to
override the toApp call in your Application object. Here you can
inspect all the outgoing messages. A common thing to do here would be
to check either for a MsgType of D or an out of date SendingTime and
throw a DoNotSend exception. This will signal the engine to replace
your message with a gap fill.
This way the other side is satisfied and your message will not be sent
out.
I think that covers the scenarios you were interested in.
--oren
> -------- Original Message --------
> Subject: [Quickfix-developers] How does Resending messages work?
> From: Fervus <JL...@ge...>
> Date: Thu, March 04, 2010 12:24 pm
> To: qui...@li...
>
>
> QuickFIX Documentation: http://www.quickfixengine.org/quickfix/doc/html/index.html
> QuickFIX Support: http://www.quickfixengine.org/services.html
>
>
> I understand that quickfix will automatically handle requesting missing
> messages, what I want to know is how does it handle this?
>
> Does it store the messages after the missing message and then process them
> after receiving the missing message? Or does it just throw it away and
> request all the messages again? Or some other way entirely? How long back
> will go looking for lost messages or if I was sending out messages and I get
> a resend request but it was a market order is there a way to intercept that
> message and decide whether I want to really resend that message or not?
>
> I have looked through the code and these forums and can not find anything
> that tells me this. I am only curious because I do not want to just use some
> system and then if it fails for some reason or I need to adjust how I handle
> missing/lost messages I understand how my adjustments will effect QuickFix
> and my FIX messaging system.
>
> Thanks
> Josh
> --
> View this message in context: http://old.nabble.com/How-does-Resending-messages-work--tp27784661p27784661.html
> Sent from the QuickFIX - Dev mailing list archive at Nabble.com.
>
>
> ------------------------------------------------------------------------------
> Download Intel® Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> Quickfix-developers mailing list
> Qui...@li...
> https://lists.sourceforge.net/lists/listinfo/quickfix-developers
|
