[omniORB] Callback problem

[Duncan Grisby]

On Friday 16 February, Pletyak Attila wrote:

> Duncan, I've already wanted to ask you what do you propose if someone
> wants to get to the highest levels of CORBA. With omniORB 2.8 I am quite
> familiar as if I did not understand something I looked into it's headers
> (/usr/include/omniORB2) or generated headers (I mean stubs and
> skeletons) and I got answers to my problems. I am just planning to go
> through the CORBA 2.3.1 Spec and the IDL to C++ Mapping Spec, as I see
> in all of your documentations of omniORB that you are constantly
> referring them.  Maybe the order of my business is not right, but as a
> Hungarian proverb says: Better Late Than Never. I've already succeeded
> to write a quite big and sophisticated system with CORBA, so I am not so
> afraid, but anyway. I can always find some good things to learn.

I think the best bet is to get Henning and Vinoski's book, and read
it. I don't agree with all of their opinions, but they explain things
very well. The C++ mapping specification isn't the easiest thing in
the world to read, and it often goes into more detail than a normal
user needs to know. Other bits of the specification, like the POA
chapter, for example. can be more useful.

It can sometimes be misleading to look at omniORB's stub code (or any
other ORB's for that matter), since that reveals all the
implementation details. There's a risk that you'll accidentally make
use of an implementation detail, limiting your portability to other
ORBs. The stub code can also be pretty confusing at times, too.

Cheers,

Duncan.

-- 
 -- Duncan Grisby  \  Research Engineer  --
  -- AT&T Laboratories Cambridge          --
   -- http://www.uk.research.att.com/~dpg1 --