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

[Ben L. Titzer]

On Mar 14, 2005, at 9:01 AM, E. Weddington wrote:

> Marek Michalkiewicz wrote:
>
> > On Sun, Mar 13, 2005 at 06:56:30PM +0100, Rolf Ebert wrote:
> >
> >
> > > My statement above is still true.  Nested functions per se do work 
> > > with  current gcc and the Ada front end.  It does not work to take 
> > > addresses of  nested functions, though.  Obviously the generation of 
> > > trampolines is  deferred to the actual use of nested function 
> > > pointers.
> >
> > Trampolines can't work on the AVR because it is not possible to 
> > execute
> > code from writable memory.  Well, flash is writable with SPM if you 
> > are
> > really desperate... but it is slow and erase/write cycles are limited.
> > So the best we can do is to document the limitation and live with it.
>
> Oh, well then I take back my suggestion I just sent about writing up 
> an enhancement report.
> I'm still catching up with my mail.... :-) Sorry for the noise.

Just as a side note, people in sensor networks using the AVR are doing 
"over the air" programming where the program binary is transmitted via 
the radio, the nodes download it into their flash, and begin executing 
it. There have been some major complaints about avr-gcc having trouble 
generating PIC (position-independent code) on the AVR.

So, yes there are crazy people out there :-)

-B

=================================================
"Research is what I'm doing when I don't know what I'm doing."
--Wernher Von Braun