Re: [Nmh-workers] Nmh-commits Digest, Vol 108, Issue 11

[Ralph Corderoy]

Hi Paul and Bakul,

> > Paul wrote:
> > > Ralph Corderoy (7):
> > >     fmttest.c: Avoid `++' with bools, silencing compiler warnings.
> > 
> > i hate that perfectly reasonable, traditional idioms have to be
> > avoided for this reason.
>
> No strong reason to use type bool in the first place. It didn’t show
> up till c99.

You're both wrong.  :-)

It's always been confusing that MH uses `foo++' to process `-foo'
because it's not clear from that location whether multiple `-foo' are
significant, e.g. they might increase the fooness.  By changing foo's
definition to a bool it's is very clear that only its truthness matters.
Those left as non-bool, once the move to bool is completed, suggest that
the non-zero quantity matters.

Further, elsewhere there are quite a few eight-bit ints that are happily
`++'d without care that this incrementing might happen to stop at the
point it has wrapped to zero.  IIRC mhparam used to try and exit(3) with
the number of missing components.  I readily use `++' instead of `+= 1',
who wouldn't?  But not without considering overflow in both cases.

As for C99, POSIX mandates it these days so it's time to upgrade from
pcc(1)!

-- 
Cheers, Ralph.
https://plus.google.com/+RalphCorderoy