[avr-libc-dev] Re: [avrdude-dev] Problems with -O2 and gcc-4.0 on Darwin

[Eric Weddington]

Joerg Wunsch wrote:

> > > So the code that's generated is equivalent to what you get with
> > > _BV() and with bitfields, only it's more type-safe.
>
>
> Well, using bit numbers is not our invention: it's Atmel's idea.  We
> are basically bound to it since it's the de-facto standard for the
> AVR.  Personally, I find it stupid as well, as any seasoned C
> programmer would have standardized the bitmasks instead of their
> numbers, but obviously, these header files are intented not only for C
> but also for assembly programmers, and these do need the bit numbers
> occasionally (like in SBI or CBI instructions).

IYHO. :-) Personally I don't find thim as stupid; there are many 
different ways to achieve the same thing. As long as the bit numbers are 
constants then the compiler will do the heavy lifting of creating the 
bit mask, so no penalty is involved in using them. Sometimes I find it 
easier to think about the bit number needed than thinking about the bit 
mask.

Bitfields have their place too, but I find it overkill when using them 
just to work with one bit. Plus, using the bitwise operators allows 
greater flexibility.


> > Really, this conversation should be taken over to the avr-libc-dev
> > mailing list, ...
>
>
> Agreed.

CCed.

--
Eric Weddington