On Wed, Mar 3, 2010 at 10:34, Charles Lesire-Cabaniols
<Charles [dot] Lesire [..] ...> wrote:
> Hi Falks,
>
> The app I am developing is fully dependent on StateMachine
> implementation: my root state machine controls the autonomous mission of
> my robot, executing sub state machines depending on my robot's state.
>
> The pb I encounters is that I have to define all my state machines (root
> and subs) on the same osd file, while it would be really great for me to
> define sub machines in independent files (usefull for unit testing and
> of course reusability).
>
> As anyone an idea on how to do that we the current 1.10 Orocos scripting
> service?
>
> Could this feature be added to the next (2.0) release?

We actually had this discussion when implementing the state machines.
The idea back then was that if you needed a definition of another
state machine, you should be able to #include it and use the cpp
preprocessor to do the work for you. Since keeping every state machine
definition into memory is eventually a waste of resources (gcc doesn't
do this either), we sticked to this.

The current architecture for parsing scripts is to create a parser on
the stack and thus clean up everything when parsing is done. We'd need
to create a component local repository to keep track of definitions
before we could implement what you propose. Not a big patch, but I'm
not 100% convinced either.

Maybe we should build-in the support of calling a preprocessor ?

Peter