[Info-vax] C compiler question

[John Reagan]

On Monday, July 29, 2019 at 9:05:12 AM UTC-4, John E. Malmberg wrote:
> On 7/27/2019 11:57 AM, Mark Berryman wrote:
> > Something I see all the time when porting unix code is something similar 
> > to the following:
> > 
> > unsigned int x;
> > .
> > .
> > .
> > if (x <= 0) {something}
> > 
> > This, of course, generates a QUESTCOMPARE informational.  I'm wondering 
> > what code the compiler generates in such a case.  Does it treat the 
> > variable as if it had been declared signed or does it do something else? 
> >   Is the compiler's behavior in this case defined by a standard or is it 
> > left up to the implementation?
> 
> By the standard.  "x" gets promoted to a signed type which preserves the 
> negative value.
> 
> This is unavoidable in porting code since many standard routines are 
> allowed to have a prototype return an unsigned type, yet specify a 
> negative value for errors.
> 
> > Part of what I am wondering is if I can count on the code generated by a 
> > VMS compiler to function the same as the code generated by a gcc or 
> > clang compiler.
> 
> I had the same concerns until the relevant section of the C standard was 
> shown to me.  I think it was on comp.os.vms that I may have asked about 
> it a few years ago.
> 
> Regards,
> -John
> wb8tyw at qsl.net_work

C's unsigned promotion rules are rarely what a human expects.  Go spend 6 minutes and watch

https://www.youtube.com/watch?v=wvtFGa6XJDU

If that still doesn't scare you away, go spend an hour and watch

https://www.youtube.com/watch?v=yG1OZ69H_-o