Re: [avr-libc-dev] pgmspace.h compatibility with -mint8

[Theodore A. Roth]

On Tue, 16 Sep 2003, E.Weddington wrote:

> > As E. Weddington wrote:
> >
> > > > There is a small patch to make pgm_read_*() work
> with -mint8
> > > > switch.
> >
> > > However, I don't know if this will be put into the
> library,
> > > mainly because there are large portions of the library
> that
> > > are not -mint8 comptabile, and I don't know if there
> will
> > > ever be plans to make it so.
> >
> > I think consensus was that we don't want to spend time on
> making it
> > -mint8 clean, but we don't want to artificially prevent
> those parts to
> > be -mint8 clean where this can be had without additional
> costs, in
> > particular without any negative side effects on the !-
> mint8 case.
> >
> > I didn't look at the patch so far, but if it fits into
> this, I don't
> > see a reason to not accept it.  The only nit is that it
> might be worth
> > documenting when some header file is believed to be -
> mint8 compatible
> > (because we made a global statement that »avr-libc is
> basically not
> > compatible with -mint8«).  We did some minor reorg on
> <inttypes.h>
> > before to make it -mint8 compatible.  This is documented
> as well.
> >
>
>
> Ah, it's all coming back to me.....
> Ok, looked at the patch, and IMHO it should be committed as
> obvious on both head and branch. Opinion, Ted?

I think using -int8 will break the pgm_*_far() macros. Might need to
remove those if -int8 is used.

Also, should also change 'unsigned long' to uint32_t in the pgm_*_far
macros.

Does -int8 break the prog_long and prog_long_long typedefs too? I'm
not sure I fully understand the purpose of the prog_* typedefs.

I think the changes to uint*_t are obvious, but proper -int8 support
may need more work.

Ted Roth