Hello OFP Developers,
Nice looking project, congratulations. We currently develop using a customized version of the Quanthouse (Quantfactory) codebase, but are always on the lookout for a decent OSS platform like OFP to take the lead in the future (as only a OSS project could .

A question on the latest news titled "New Release(s) Coming" - are there any more details/debate on the new design and direction? Here the move has been over to Event Sourcing/Command query separation design (I'll put some public references below for convenience). Interested to see if the new OFP design has gone this direction given its natural fit for multi-"Expert" trading systems - or if you guys have come up with something different entirely.

Kalori

Ref:
http://en.wikipedia.org/wiki/Command-query_separation
http://martinfowler.com/eaaDev/EventSourcing.html

Related Discussion/Application to trading platforms:
http://qconsf.com/sf2007/file?path=/QCo ... Design.pdf
http://www.infoq.com/interviews/greg-young-ddd
http://elegantcode.com/2009/11/11/cqrs-la-greg-young/

Hey Kalori,

interesting topic this

What we have done is build a new module we call a "super pool" that brings together a message bus, a "remoting" framework, a thread (management) and a component control. It sounds like a mess, but it is amazingly simple to use and the implementation is also very clear. It is a new pattern (or sort of combination of many) and it has many many advantages, and is hella fast to write code with and execute code upon.

This will be part of a separate new platform we shall soon release (expected release date 1.Mar), and it will be aimed at general programming, not finance, with the next phaze a (new) finance platform based on that.

The direction we are moving is putting more action on automated trading and optimization, as well as simplified more pragmatic structure and architecture. The current platform is amazingly flexible, but a bit hard to work with so the new thing will be a big step in that direction...

Thanks sounds very interesting, look forward to seeing how it all fits together.

Just a quick question: Will the new OFP design allow easy scaling "ExpertHosts" out across many physical/virtual servers (all coordinating on the same trading account) - or is the design more towards one or two servers handling multiple experts? Just though I'd ask as scalability was one of the first reasons we started eyeing CQRS design (no choice when each "expert" system starts to take up a lot of resources , so be interesting to know where OFP is heading.


Thanks!

Scaling the solution is way fun on the new system (in a good way)!

Since all the communication (inter component invocation) is inherently message based, and we have provided a very fast default built in TCP.IP module, you can move your components around as you wish. Placing them on as many separate machines as needed.

The only requirement of the model is it requires one machine to be assigned "server" and essentially it brings them all together, acting as a server everyone connects to. However a distributed (multiple servers) model is also easily achievable, though probably rarely needed.

So this all goes to ExpertHosts as well, no problem to have a server with a single trading connection and a 100 different expert hosts running on 100 separate machines.

And all of this comes standard, and takes 3 lines of code to set up.

Great, looking forwarde to trying it out!