Re: [avr-gcc-list] Re: [avrdude-dev] [RFC] avrdude Feature RequestandCall for Volunteers

[David Brownell]

On Tuesday 13 March 2007 1:51 pm, Volkmar_WebDE wrote:

> > - Fuse bytes are particular to a chip for a given set of features to 
> > activate/disactivate and used when flashing a chip.  ...
> 
> I do not agree at this point. The source code depends on features
> which you can enable/disable with the fuses. The integration of
> defining the fuses in the source code ensures that you have one
> location where all the needed features are defined.

Keep in mind that it's common to stick such critical info in
ELF sections; that's what they were designed for.  Linux kernels
use them for all kinds of stuff.

It may also be useful to tag the ELF files with the chip, to
help prevent glitches like flashing an Mega168 with a file
that was intended for an Mega169, etc.  I know there are many
variants of that particular problem (e.g. in libraries), so
it may be worth thinking about how build systems may want to
make such sanity checks.  Folk not using avrdude could also
benefit from such checks, but it's hard to argue that avrdude
shouldn't notice a mismatch and abort.

Seems to me that what pushback this idea gets is coming from
what one might call "current generation tools", and that this
idea is a good basis for "next generation tools" which are less
error prone and easier to use ... initially for developers and
small-scale production, then eventually for other tools.

- Dave