[Bug c/103531] New: Prpose compiler warning when ceil functions used on integral value

[Bug c/103531] New: Prpose compiler warning when ceil functions used on integral value

[eyalroz1 at gmx dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103531

            Bug ID: 103531
           Summary: Prpose compiler warning when ceil functions used on
                    integral value
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: eyalroz1 at gmx dot com
  Target Milestone: ---

The ceil functions (ceil() and ceilf() in C) take floating-point arguments:
double or float.

If the programmer passes an integer value to one of these functions, it'll get
converted to float/double, and the function will be idempotent. It is thus
typically useless to do so, but also harmless in itself.

But this is _not_ harmless when the programmer assumes they are running ceil()
on what they believe is a floating-point value, but actually isn't. A "typical"
example:

forgot_whether_this_type_is_integral_t x = get_x();
int covers_half = ceil(x / 2); 

If x is a floating-point value, this codes will act as expected; but if it's
integral, it will actually place the _floor_ of x / 2.0 in covers_half.

I propose, therefore, that invoking ceil()/ceilf() on an integral value trigger
a warning. Alternatively, that some more complicated conditions trigger the
warning, e.g. invoking ceil() on the result of an integer division or integer
multiplication.

[Bug c/103531] Prpose compiler warning when ceil functions used on integral value

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103531

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement
     Ever confirmed|0                           |1
           Keywords|                            |diagnostic
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2021-12-02

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
Confirmed.  Also applies to the C++ frontend.

[Bug c/103531] Propose compiler warning when ceil/ceilf used on integral value

[eyalroz1 at gmx dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103531

Eyal Rozenberg <eyalroz1 at gmx dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Prpose compiler warning     |Propose compiler warning
                   |when ceil functions used on |when ceil/ceilf used on
                   |integral value              |integral value

--- Comment #2 from Eyal Rozenberg <eyalroz1 at gmx dot com> ---
This also applies to C++, although in C++ you might be doing this in a
templated context, e.g. 

T x = get_x();
auto y = ceil(x); 

which is fine for floating-point T's and fishy for integral T's.

[Bug c/103531] Propose compiler warning when ceil/ceilf used on integral value

[egallager at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103531

Eric Gallager <egallager at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |egallager at gcc dot gnu.org
             Blocks|                            |87403

--- Comment #3 from Eric Gallager <egallager at gcc dot gnu.org> ---
-Wtraditional-conversion catches this:

$ /usr/local/bin/gcc -c -Wall -Wextra -pedantic -Wconversion -Wdouble-promotion
-Wtraditional-conversion 103531.c
103531.c: In function 'foo':
103531.c:5:30: warning: passing argument 1 of 'ceil' as floating rather than
integer due to prototype [-Wtraditional-conversion]
    5 |     int covers_half = ceil(x / 2);
      |                            ~~^~~
103531.c:5:23: warning: conversion from 'double' to 'int' may change value
[-Wfloat-conversion]
    5 |     int covers_half = ceil(x / 2);
      |                       ^~~~
$


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87403
[Bug 87403] [Meta-bug] Issues that suggest a new warning

[Bug c/103531] Propose compiler warning when ceil/ceilf used on integral value

[eyalroz1 at gmx dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103531

--- Comment #4 from Eyal Rozenberg <eyalroz1 at gmx dot com> ---
(In reply to Eric Gallager from comment #3)
> -Wtraditional-conversion catches this:

Well... you're technically right, but:

1. That is a much wider warning. If someone were to turn this on they would get
a gazillion warnings which would drown out this one.
2. (Weak point) Few people would consider turning this on given the
description.
3. This doesn't work for C++.