Bug 204834 – Clang 10 reports -Wimplicit-int-float-conversion compilation warnings

• Status: RESOLVED
• Resolution: FIXED
• Component: Web Template Framework

[Fujii Hironori]

Clang 10 reports a compilation warning:

> wtf/MathExtras.h(401,43): warning: implicit conversion from 'unsigned long long' to 'double' changes value from 18446744073709551615 to 18446744073709551616 [-Wimplicit-int-float-conversion]
>         double fmodValue = fmod(trunc(d), std::numeric_limits<unsigned long long>::max() + 1.0);

doubleToInteger was added in Bug 67980.

[Fujii Hironori]

More warnings: 

> [20/129] Building CXX object Source\WTF\wtf\CMakeFiles\WTF.dir\MediaTime.cpp.obj
> ..\..\Source\WTF\wtf\MediaTime.cpp(117,21): warning: implicit conversion from 'long long' to 'float' changes value from 9223372036854775807 to 9223372036854775808 [-Wimplicit-int-float-conversion]
>     if (floatTime > std::numeric_limits<int64_t>::max())
>                   ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> ..\..\Source\WTF\wtf\MediaTime.cpp(124,36): warning: implicit conversion from 'long long' to 'float' changes value from 9223372036854775807 to 9223372036854775808 [-Wimplicit-int-float-conversion]
>     while (floatTime * timeScale > std::numeric_limits<int64_t>::max())
>                                  ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> ..\..\Source\WTF\wtf\MediaTime.cpp(147,22): warning: implicit conversion from 'long long' to 'double' changes value from 9223372036854775807 to 9223372036854775808 [-Wimplicit-int-float-conversion]
>     if (doubleTime > std::numeric_limits<int64_t>::max())
>                    ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> ..\..\Source\WTF\wtf\MediaTime.cpp(154,37): warning: implicit conversion from 'long long' to 'double' changes value from 9223372036854775807 to 9223372036854775808 [-Wimplicit-int-float-conversion]
>     while (doubleTime * timeScale > std::numeric_limits<int64_t>::max())
>                                   ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 4 warnings generated.

[Fujii Hironori]

Created attachment 384784 [details]
WIP patch

[Fujii Hironori]

Umm, there are more. Here is full compilation log after applying the WIP patch.
https://gist.github.com/fujii/160ae6e66138d6b48dc4e342f58837b6

[Don Olmstead]

https://bugzilla.mozilla.org/show_bug.cgi?id=1577236

Here's Mozilla's work to fix this warning in their codebase

[Fujii Hironori]

Created attachment 395215 [details]
Patch

[Darin Adler]

Comment on attachment 395215 [details]
Patch

We don’t use typically function style casts in WebKit. We should look for other ways to do the conversion.

These warnings suggest there are mistakes in some of these places. I am not sure that just quieting the warning without a behavior change is good.

[Fujii Hironori]

https://stackoverflow.com/a/17822304

This is interesting.
They uses an idea using ldexp which is similar approach with my WIP patch (comment 2).

[Fujii Hironori]

Filed: Bug 209955 – [Clang 10] Fix -Wimplicit-int-float-conversion compilation warnings in WTF

[Fujii Hironori]

No more tasks. Closed.