Re: A C++ API for the pcas

[Tony Cox - (415)926-3105 <tcox@ssrl01.slac.stanford.edu>]

Michael writes:-

>
>At an object oriented programming class, I talked to a couple of high level
>engineers from Talegent (they [soon] will be selling a C++ "environment").
>They said C++ is inappropriate for realtime uses--among other things
>garbage collection is still not solved.  They noted it has been used in
>realtime, but the examples they gave had rather slow response bounds (~1
>sec).  They said it would be a good choice for user interface work.

That sounds backwards to me. The fact that garbage collection is not part of 
the C++ language is a _good thing_ for real-time programming. The time taken to
garbage collect is essentially unbounded, and whilst in progress you can't get
at your objects (because the GC is moving them around). Better to use smart
pointers and do your own GCing, not let some language compiler decide when is a
good time to clean house.

Further C++ has compile-time binding, unlike some other languages. This means
that you can pass messages (call virtual functions) in a way which is quite
deterministic. Most other languages use some form of run-time binding, which
involves deciding at execution time who the message is for - extend your object
hierarchy, and suddenly you find the timing for your message dispatching is
completely different from what it was.

C++ certainly has its faults, but inapplicability for RT software development
isn't one of them.

Tony

--------------------------------------------------------------------------------
Dr Anthony D Cox
Computer Systems Specialist
Stanford Synchrotron Radiation Laboratory
Stanford Linear Accelerator Center
MS 69, Box 4349
Stanford CA 94305
tcox@ssrl01.slac.stanford.edu
--------------------------------------------------------------------------------