Re: ada and robots

["Marin David Condic, 561.796.8997, M/S 731-93" <condicma@PWFL.COM>]

Joe Gwinn <gwinn@RES.RAY.COM> writes:
>There is tons of such code available for free on the internet.  Just look
>for I/O drivers, such as for UNIX and VxWorks. VxWorks Board Support
>Packages are another example.
>
>Even C-coded I/O drivers and kernel initialization sequences resort to
>some assembly code.
>
>As for shared memory, look into the handling of driver comm pools, queues
>shared between tasks and also with ISRs.  And so on.
>
>For UNIX drivers, there are multiple books available on exactly how to
>write such drivers, all in C.  Most include (annotated) sample driver
>code.
>
>I would submit that there is ample evidence that C can be used for direct
>hardware control, and for shared memory.
>
>I would not claim that all who have tried such things have succeeded,
>regardless of language.  Many programmers couldn't write an I/O driver if
>their life depended on it.
>
    It seems that what your saying is that C has been used to develop
    code which exercises direct control of hardware and is able to
    implement shared memory data structures between threads. I don't
    think that anyone would doubt this.

    I think what I originally took exception to was the claim that Ada
    *can't* do these things. If I was trying to implement some
    extension to Unix, (a device driver, perhaps) I probably wouldn't
    spit into the wind by trying to do it in Ada (Given that the whole
    OS is written in C, all the support tools are C oriented and I
    wouldn't want to maintain one lonely little Ada module afloat in a
    Sea of C ;-) But that's a far cry from claiming that Ada is
    incapable of being used to write device drivers.

    I've written lots of code which implements interfaces to A/D
    converters, ISR's, discrete I/O, Mil-Std-1553 buses, UARTs,
    Manchester data links, etc, etc, etc. All in Ada(83). I've shared
    memory between tasks and between processors using Ada(83) (albeit,
    not using the "pragma SHARED" - but implementing what you'd have
    to do for yourself in C anyway - a semaphore or some hardware
    support for shared memory.). So it's a little hard to read "Ada
    can't be used down at the bare metal" when you know that you've
    been doing just that for years - and that it's not at all
    any more difficult than it would be in C.

    While I won't presume to speak for Dr. Dewar (I know he's pretty
    good at clarifying himself :-) I think its fair to say we'd both
    like to see an example of something you believe that C does in
    dealing with the bare machine that you feel would be difficult in
    Ada. I'll accept that there may be some things C does better than
    Ada, but I'd like to see an example.

    MDC

Marin David Condic, Senior Computer Engineer     ATT:        561.796.8997
Pratt & Whitney GESP, M/S 731-96, P.O.B. 109600  Fax:        561.796.4669
West Palm Beach, FL, 33410-9600                  Internet:   CONDICMA@PWFL.COM
===============================================================================
    "You spend a billion here and a billion there. Sooner or later it
    adds up to real money."

        --  Everett Dirksen
===============================================================================