[ts-gen] Experiments with Java API

[R P Herrold]

On Tue, 6 Oct 2009, Ken Feng wrote:

> One interesting thing I have found after browsing the IB documentation
> is that OrderStatus seems to have a whole bunch of states, many of
> which are ignored/filtered-out by the Shim:  Cancelled, Inactive,
> Pending, Filled, Partial, PendingCancel, PendingSubmit, PreSubmitted,
> and Submitted.  Some of these I have yet to see in paper trading -
> like Inactive.  But I have definitely seen PreSubmitted and
> PendingSubmit.  In fact, I previously had to look for these in the
> 3/3/6 filters through the logs to capture them;  I actually prefer to
> interpret PendingSubmit as Submitted, but this could depend on the user.

As I recall, the state machine we maintain is the one relevant 
to the non-GUI implementation, and based on the Java code, 
rather than the textual documentation (which textual 
documentation is of course a secondary source to the code 
itself)

> I will post more later as I learn more about the Java API...

I am all in favor of people reading and understandig the 
upstream's approach as documented in teh Java code.  In 
addition to:  EWrapper.java you may wish to also consider: 
EClientSocket.java

As to direct experimentation with the Java API, the shim, of 
course, is not a derivative work representing yet another 
layer on top of IB's code, but a true new implementation 
speaking the wire protocol of the TWS, quite intentionally on 
our part.

Others have that Java extensions space covered and as 
reinventing that wheel is less interesting, we leave that to 
them

-- Russ herrold