Re: [avr-gcc-list] Improving in leaps and skips!

[Rolf Ebert]

On Wed, 02 Mar 2005 12:56:56 -0700, E. Weddington <address@hidden> wrote:

> Rolf Ebert wrote:
>
> > The mere availablity of nested functions in GNU C made it possible to  
> > base  GNU compilers for Ada and Pascal on the existing GNU C compiler.
> If that's the case, does the fact that nested functions don't seem to  
> work for the AVR (according to Björn's view of working with the test  
> suite) affect how AVR-Ada will behave? Will there be problems?

That depends on the meaning of the word "work"...

I never had problems up to now with nested functions in AVR-Ada.  But then  
I have mostly small local functions that don't do heavy tasks and that  
have short parameter lists.

Before posting this answer I made up a test case with heavily nested  
functions of three levels with local variables at all levels partly hiding  
each other.  The test case works correctly when I simulate it in AVR  
Studio!

The example which started this thread passes a pointer to a nested  
function as a parameter to another function.  I did not test that case  
(will do shortly).  That seems to be a very contrived example. I had never  
seen code like this.

The point is: nested functions *do work* in AVR-Ada for what they are  
typically used in Ada (limiting visibilty, enhancing code readability).

    Rolf