[Bug libquadmath/114126] New: A not infinite result of tanq of M_PI_2

[Bug libquadmath/114126] New: A not infinite result of tanq of M_PI_2

[www3.spl at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114126

            Bug ID: 114126
           Summary: A not infinite result of tanq of M_PI_2
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libquadmath
          Assignee: unassigned at gcc dot gnu.org
          Reporter: www3.spl at gmail dot com
  Target Milestone: ---

Hi there. I think that this 'bug' was not sent before.

I'm obtaining an incorrect result for tanq( M_PI_2q ):
tanq( M_PI_2q ) = +2.306323558737156172766198381637374e+34
when it should be infinite.

[Bug libquadmath/114126] A not infinite result of tanq of M_PI_2

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114126

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Can you provide a full testcase? And also specify which target are you on?

[Bug libquadmath/114126] A not infinite result of tanq of M_PI_2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114126

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org
         Resolution|---                         |INVALID
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Why do you think this is a bug?
#include <math.h>
#include <stdlib.h>
#include <stdio.h>

int
main ()
{
  _Float128 f = tanf128 (M_PI_2f128);
  volatile _Float128 g = M_PI_2f128;
  g = tanf128 (g);
  char buf[128];
  strfromf128 (buf, 128, "%.34a", f);
  printf ("%s\n", buf);
  strfromf128 (buf, 128, "%.34a", g);
  printf ("%s\n", buf);
}
also prints
0x1.1c46bd57277993a2ee60193c957b000000p+114
0x1.1c46bd57277993a2ee60193c957b000000p+114

M_PI_2q or M_PI_2f128 is
1.5707963267948966192313216916397513987...
while pi/2 with larger precision is I think
1.5707963267948966192313216916397514420...
so M_PI_2{q,f128} is rounded down, not up,
so no wonder tanq/tanf128 is not inf.

[Bug libquadmath/114126] A not infinite result of tanq of M_PI_2

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114126

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Not to mention that if it would be rounded up (like it happens e.g. in the
M_PI_f32 case), you wouldn't get inf either, nor -inf, but some large negative
number.

[Bug libquadmath/114126] A not infinite result of tanq of M_PI_2

[www3.spl at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114126

--- Comment #4 from Sergio Peña <www3.spl at gmail dot com> ---
(In reply to Jakub Jelinek from comment #2)
> Why do you think this is a bug?
> #include <math.h>
> #include <stdlib.h>
> #include <stdio.h>
> 
> int
> main ()
> {
>   _Float128 f = tanf128 (M_PI_2f128);
>   volatile _Float128 g = M_PI_2f128;
>   g = tanf128 (g);
>   char buf[128];
>   strfromf128 (buf, 128, "%.34a", f);
>   printf ("%s\n", buf);
>   strfromf128 (buf, 128, "%.34a", g);
>   printf ("%s\n", buf);
> }
> also prints
> 0x1.1c46bd57277993a2ee60193c957b000000p+114
> 0x1.1c46bd57277993a2ee60193c957b000000p+114
> 
> M_PI_2q or M_PI_2f128 is
> 1.5707963267948966192313216916397513987...
> while pi/2 with larger precision is I think
> 1.5707963267948966192313216916397514420...
> so M_PI_2{q,f128} is rounded down, not up,
> so no wonder tanq/tanf128 is not inf.

Ok. It is posible I was wrong.

[Bug libquadmath/114126] A not infinite result of tanq of M_PI_2

[www3.spl at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114126

--- Comment #5 from Sergio Peña <www3.spl at gmail dot com> ---
(In reply to Jakub Jelinek from comment #2)
> Why do you think this is a bug?
> #include <math.h>
> #include <stdlib.h>
> #include <stdio.h>
> 
> int
> main ()
> {
>   _Float128 f = tanf128 (M_PI_2f128);
>   volatile _Float128 g = M_PI_2f128;
>   g = tanf128 (g);
>   char buf[128];
>   strfromf128 (buf, 128, "%.34a", f);
>   printf ("%s\n", buf);
>   strfromf128 (buf, 128, "%.34a", g);
>   printf ("%s\n", buf);
> }
> also prints
> 0x1.1c46bd57277993a2ee60193c957b000000p+114
> 0x1.1c46bd57277993a2ee60193c957b000000p+114
> 
> M_PI_2q or M_PI_2f128 is
> 1.5707963267948966192313216916397513987...
> while pi/2 with larger precision is I think
> 1.5707963267948966192313216916397514420...
> so M_PI_2{q,f128} is rounded down, not up,
> so no wonder tanq/tanf128 is not inf.

Ok. It is posible I was wrong.
I have found this question:
https://stackoverflow.com/questions/54287492/why-didnt-i-get-tanpi-2-infinty-in-c