Bug 238443 – [GLIB] Build error in GraphicsContextState.cpp (error: no match for 'operator!=') in Ubuntu LTS after r291696

• Component: New Bugs
• Status: RESOLVED FIXED

[Diego Pino]

[GLIB] Build error in GraphicsContextState.cpp (error: no match for 'operator!=') in Ubuntu LTS after r291696

[Diego Pino]

Created attachment 455896 [details]
Patch

[Adrian Perez]

Comment on attachment 455896 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=455896&action=review

> Source/WebCore/platform/graphics/GraphicsTypes.h:90
> +}

This cannot work, moving the function outside of the “class” block means
that now there is no implicit “this”, and that no operator== is defined
at all for CompositeMode. You need to leave this one as it was.

[Diego Pino]

Created attachment 455903 [details]
Patch

[Diego Pino]

The build error from the Ubuntu LTS build bot was:

```
Source/WebCore/platform/graphics/GraphicsContextState.cpp:60:84: error: no match for ‘operator!=’ (operand types are ‘const WebCore::CompositeMode’ and ‘WebCore::CompositeMode’)
   60 |         changeFlags.set(change, !lastDrawingState || (*lastDrawingState).*property != this->*property);
```

https://build.webkit.org/#/builders/68/builds/12304/steps/8/logs/stdio

[EWS]

Committed r291970 (248935@main): <https://commits.webkit.org/248935@main>

All reviewed patches have been landed. Closing bug and clearing flags on attachment 455903 [details].

[Radar WebKit Bug Importer]

<rdar://problem/90929679>

[Lauro Moura]

FTR, this seems to be related to changes to the C++20 equality/spaceship operators in GCC 10

Sample godbolt: https://godbolt.org/z/GWen331rM

GCC C++20 support: https://gcc.gnu.org/projects/cxx-status.html#cxx20

Especially the following spec, first supported in gcc10: http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2019/p1185r2.html#wording

> A defaulted != operator function for a class C with parameters x and y is defined as deleted if

> (snip list of cases where != would be deleted.)

> Otherwise, the operator function yields (x == y) ? false : true if an operator == with the original order of parameters was selected, or (y == x) ? false : true otherwise.

So, GCC >10 correctly defined the defaulted `!=` operator, while older GCC's needed them to be explicitly defined (like the patch did), as it did not implement this behavior.