Re: long long unsigned on 32-bit machines

[Kai Harrekilde-Petersen]

My guess is that the problem lies in the bit-selection expresion: (1 << i).
The 1 is a signed int by default, so forcing i to be an uint64_t 
promotes the result to be 64 bit as well, so testing for bit 31 works 
correctly. If this is indeed the problem, use 1UL instead.

/Kai, shooting from the hip here

On 26 Jan 2020 15.21, Bernd Zeimetz wrote:
> On 1/26/20 2:21 PM, Sanjeev Gupta wrote:> I think we should wait for
> Bernd to chime in.
>
> That is just the fix for the change before - unsigned long is still not
> 64bit on 32bit. I can't remember how I stumbled across it, but I guess
> some tool pointed it out.
>
> After looking at the code again - I have no idea why there is something
> else than a simple int needed.
>
>
>
> commit b1b555e0cb37382655a3a7cb37f2f89cb89eda7a
> Author: Gary E. Miller <address@hidden>
> Date:   Thu Oct 10 12:19:28 2019 -0700
>
>      monitor_italk.c: 32-bit unsigned too small by one.
>
>      Problem found by cpplint
>
> diff --git a/monitor_italk.c b/monitor_italk.c
> index 8da7e17ea..6ee79cbf0 100644
> --- a/monitor_italk.c
> +++ b/monitor_italk.c
> @@ -160,11 +160,11 @@ static void display_itk_navfix(unsigned char *buf,
> size_t len)
>       (void)wmove(navfixwin, 11, 6);
>       {
>          char prn[4], satlist[38];
> -       unsigned int i;
> +       unsigned long i;         // unsigned long just in case ints are
> 32-bit
>          satlist[0] = '\0';
>          for (i = 0; i < 32; i++) {
>              if (svlist & (1 << i)) {
> -               (void)snprintf(prn, 4, "%u ", i + 1);
> +               (void)snprintf(prn, 4, "%lu ", i + 1);
>                  (void)strlcat(satlist, prn, sizeof(satlist));
>              }
>          }
>
>
> > --
> > Sanjeev