> We are in a similar situation. We have an EPICS and CDEV
> based telescope control system upgrade that is as yet
> unproven while another group at the observatory is starting
> to investigate CORBA as a part of a new detector controller.
> Would be nice to not use two simiiar tools where one would suffice.
> CORBA is in use at least one site within the EPICS collaboration
> that I know of - DESY.
> Aloha,
> Peregrine
Yes, it doesn't make sense to duplicate the effords of learning, using ,
testing and maintaning two different distributed middleware
implementations, and even trying to interoperate both.
Some important steps have been taken permiting the consideration
of CORBA as a serious alternative:
- ORB vendors have ported their products to RTOS platforms
(Orbix-VxWorkx, Chorus-Cool,..)
- It seems that the DoD have adopted CORBA as a standard for
ditributed systems, so several projects have started in the RT area:
(this will
provide long term support)
- The people from ACE (recently included with cdev) are
working in TAO (The Ace Orb). They will provide RT features
(RT Event Services,10 microseconds latency in combination with
Gigabit ATM..).
- There is a group in OMG working in CORBA and RT
In the other hand EPICS and cdev are a proben solution, and I think they
provide an excelent driver suppport. In addition several telescopes
have built their control system using EPICS, but none
(let me know if this is correct) have tried a full CORBA implementation.
So,
I don't see a clear choice.
josema.