> >><>I think lint was just doing its job, alerting about potential pro=
blems =

> >>that could arise from accuracy lossage and possibly misaligned poin=
ter =

> >>casts/arithmetic, for pointed-to variables of different size.
>
> I've looked at some of the errors, not all.  While I agree that =

> complaining about constant arguments to vsnprintf() may be too pedant=
ic, =

> I remember at least one less obvious situation like
>
> padded_len =3D h->pass_len =3D=3D 0 ? 16 : (h->pass_len+15) & ~0xf;
>
> where padded_len is int and the field pass_len in struct rad_handle i=
s =

> size_t, which is long.

I think the modifications I committed in revision 1.4 of radlib.c
fix the build problems for amd64 -- I added three explicit casts
to int.

Regards,

- H=E5vard