Actually the any argument about TRIS changing somehow (due to a bug) makes
it clear that there is no one solution that fits all cases when your
talking about bidirectional I/O.

Because what if you have an input to the pic that cannot tolerate a
pull-up/pull down resistor.
Without the assumption that the software is without bugs that will alter
TRIS there is no solution, simply because the pins are I/O pins

For instance it is a bad idea for me to put pull-up/down resistors on an
analog input.  It effects the analog value.  But a software bug could
cause that pin to be a digital input which might be somewhere
indeterminate because of the device giving the signal and cause a problem.

Also the pin could *ACCIDENTALLY* become an output in which case I might
damage the device the analog input is connected to.





On Wed, 26 May 1999, Benjamin Petersen wrote:

> > Just as an aside, with the talk about what one should do with
> > unused IO
> > pins. The consensus seemed to be that that the TRIS register may
> > accidently, or otherwise, change state during operation and therefore
> > the pins should be set and left as unconnected outputs.
> >
> > What if the TRIS state somehow changes to an input? Floating
> > inputs can
> > cause a bit of havoc as well. Maybe they should be set as
> > outputs with a
> > high value pullup/pulldown on them as well just to be on the
> > safe side.
>
> Unused pins ? Not in my project !
>
> I'd say that you could put a C demo version and a JAL (FULL) version and so
> on on your CD.
>
> Regards
> Benjamin Petersen
>