Re: Ada 83 - Sometimes still chosen

[Marin David Condic <condicma@bogon.pwfl.com>]

Richard D Riehle wrote:
> 
> In article <7dbvd6$56q$1@nnrp1.dejanews.com>,
>         robert_dewar@my-dejanews.com wrote:
> 
> in response to my observation about the continuation of Ada 83,
> 
> >And almost certainly it will still be Ada 95 code, seeing
> >as Ada 95 is almost exactly upwards compatible with Ada 83
> 
>  On some projects some programming managers feel it is appropriate
>  to freeze even the version of the compiler used to develop some
>  long-lived software product.  That is, once the satellite is
>  deployed, all updates will be made using exactly the same language
>  and version of the compiler used for the original development.
> 
I'd say this is true of most mission critical applications which require
any significant amount of verification. And the freezing point will come
far sooner than system deployment. The usual question goes like this:
"If you change the compiler version, host platform, etc. and feed it the
source code you gave me last week, can you guarantee that I'll get
exactly the same bits out this time as I did then?" The answer is
obviously "No." (It gets worse that that when you consider that it is
possible to have the same compiler version, OS version, etc. and run on
two different days with different loads on the system and because of
dynamic memory situations, the optimizer might give you different object
code. This actually happened to us once.)

Once you go to the effort of expensive verification testing, you need to
be able to reproduce the image precisely without disturbing anything you
did not intend to modify.

-- 
Marin David Condic
Real Time & Embedded Systems, Propulsion Systems Analysis
United Technologies, Pratt & Whitney, Large Military Engines
M/S 731-95, P.O.B. 109600, West Palm Beach, FL, 33410-9600
***To reply, remove "bogon" from the domain name.***