Re: [avr-gcc-list] Question about gcc preprocessing andport/pinassignments

[Ben L. Titzer]

On Dec 9, 2004, at 10:01 AM, Jeff Barlow wrote:

> "Graham Davies" <address@hidden>  wrote:
>
> > It would be easier to write in assembly in the first place.
>
> That was pretty much my point.
>
> > Perhaps you are saying that my
> > approach to programming is not suitable for embedded systems.
>
> That's correct. Likewise low level driver code on larger systems. (I've
> been doing both since the early seventies.)
>
> > I personally believe that significant portability can be achieved
> > with proper high level language support.
>
> I agree. One can write code that's about 99% portable. So I write that
> 99% in C and the other 1% in assembly and spend the time I save doing
> something fun. Life is too short to waste time fighting with
> recalcitrant compilers.

One can wish there was a better way, and that someone at the language 
level could make your life much easier by changing the programming 
paradigm slightly to accommodate your situation. If there are recurrent 
themes in your (any many other people's) code that could be better 
implemented with a language construct, then perhaps it would save 
everyone the time and effort if there was a simple mechanism with 
simple semantics that both compilers and developers could agree on.

In this situation, I believe it goes back to the fact that at the 
hardware level you are manipulating the logical level of a pin--there 
is no corollary for this in C. If there were a way to address 
individual pins and manipulate their values, then this might be a good 
way to go.

For example:

pin OUTPUT_PIN = 27, INPUT_PIN = 31;
boolean STATE;

if ( INPUT_PIN && STATE )
  OUTPUT_PIN = true;
else
  OUTPUT_PIN = false;

It could then be the compiler's job to take the higher level concept of 
a "pin" and translate it into the correct, atomic instructions 
necessary to set the logical values, given a description of the device 
that contains information about how the state of pins is accessible 
through hardware registers.

-B

========================================================
One machine can do the work of fifty ordinary men. No machine can do 
the work of one extraordinary man.
--Elbert Hubbard