Remove ENABLE_CSS_PAINTING_API build flag

[donny-dont]

4d1720e

Remove ENABLE_CSS_PAINTING_API build flag
https://bugs.webkit.org/show_bug.cgi?id=271913

Reviewed by Said Abou-Hallawa and Michael Catanzaro.

All ports are building with `ENABLE_CSS_PAINTING_API` as `ON`. Remove the
build option and use the runtime option for controlling use.

* Source/WTF/Scripts/Preferences/UnifiedWebPreferences.yaml:
* Source/WTF/wtf/PlatformEnable.h:
* Source/WebCore/bindings/js/JSPaintRenderingContext2DCustom.cpp:
* Source/WebCore/bindings/js/JSPaintWorkletGlobalScopeCustom.cpp:
* Source/WebCore/bindings/js/WebCoreJSClientData.cpp:
* Source/WebCore/bindings/js/WebCoreJSClientData.h:
* Source/WebCore/css/CSSPaintCallback.h:
* Source/WebCore/css/CSSPaintCallback.idl:
* Source/WebCore/css/CSSPaintImageValue.cpp:
* Source/WebCore/css/CSSPaintImageValue.h:
* Source/WebCore/css/CSSPaintSize.h:
* Source/WebCore/css/CSSPaintSize.idl:
* Source/WebCore/css/CSSValue.cpp:
* Source/WebCore/css/CSSValue.h:
* Source/WebCore/css/CSSValueKeywords.in:
* Source/WebCore/css/DOMCSSNamespace+CSSPainting.idl:
* Source/WebCore/css/DOMCSSPaintWorklet.cpp:
* Source/WebCore/css/DOMCSSPaintWorklet.h:
* Source/WebCore/css/parser/CSSParserContext.cpp:
* Source/WebCore/css/parser/CSSPropertyParserHelpers.cpp:
* Source/WebCore/css/typedom/MainThreadStylePropertyMapReadOnly.cpp:
* Source/WebCore/dom/Document.cpp:
* Source/WebCore/dom/Document.h:
* Source/WebCore/html/CustomPaintCanvas.cpp:
* Source/WebCore/html/CustomPaintCanvas.h:
* Source/WebCore/html/CustomPaintImage.cpp:
* Source/WebCore/html/CustomPaintImage.h:
* Source/WebCore/html/canvas/PaintRenderingContext2D.cpp:
* Source/WebCore/html/canvas/PaintRenderingContext2D.h:
* Source/WebCore/html/canvas/PaintRenderingContext2D.idl:
* Source/WebCore/rendering/style/RenderStyle.cpp:
* Source/WebCore/rendering/style/RenderStyle.h:
* Source/WebCore/rendering/style/StyleImage.h:
* Source/WebCore/rendering/style/StylePaintImage.cpp:
* Source/WebCore/rendering/style/StylePaintImage.h:
* Source/WebCore/style/StyleBuilder.cpp:
* Source/WebCore/style/StyleBuilderState.cpp:
* Source/WebCore/workers/WorkerOrWorkletScriptController.cpp:
* Source/WebCore/worklets/PaintWorkletGlobalScope.cpp:
* Source/WebCore/worklets/PaintWorkletGlobalScope.h:
* Source/WebCore/worklets/PaintWorkletGlobalScope.idl:
* Source/WebCore/worklets/WorkletGlobalScope.h:
* Source/cmake/WebKitFeatures.cmake:
* Tools/Scripts/webkitperl/FeatureList.pm:

Canonical link: https://commits.webkit.org/277584@main

515baac

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
✅ 🛠 🧪 jsc	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
✅ 🛠 🧪 jsc-arm64	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
✅ 🛠 🧪 merge	✅ 🛠 watch		✅ 🛠 jsc-armv7
	✅ 🛠 watch-sim		✅ 🧪 jsc-armv7-tests

[webkit-early-warning-system]

EWS run on previous version of this PR (hash 12c11a6)

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
✅ 🛠 🧪 jsc	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
✅ 🛠 🧪 jsc-arm64	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		❌ 🧪 api-gtk
	✅ 🛠 watch		✅ 🛠 jsc-armv7
	✅ 🛠 watch-sim		✅ 🧪 jsc-armv7-tests

[mcatanzaro]

"All ports are building with ENABLE_CSS_PAINTING_API as ON."

Warning klaxons: looks like this is completely wrong. It's disabled in all CMake ports except macOS (which doesn't matter because production macOS builds don't use CMake). ENABLE_EXPERIMENTAL_FEATURES does not get turned on in releases and it probably hasn't been tested anywhere except buildbots.

I would do a separate commit first to enable the build flag, land that first, then remove the build flag. I think it should be safe because the runtime feature is also disabled everywhere except if experimental features are enabled, so in theory it shouldn't matter. But still safer to do in two stages.

[webkit-early-warning-system]

EWS run on current version of this PR (hash 515baac)

Misc	iOS, tvOS & watchOS	macOS	Linux	Windows
✅ 🧪 style	✅ 🛠 ios	✅ 🛠 mac	✅ 🛠 wpe	✅ 🛠 wincairo
✅ 🧪 bindings	✅ 🛠 ios-sim	✅ 🛠 mac-AS-debug	✅ 🧪 wpe-wk2
✅ 🧪 webkitperl	✅ 🧪 ios-wk2	✅ 🧪 api-mac	✅ 🧪 api-wpe
	✅ 🧪 ios-wk2-wpt	✅ 🧪 mac-wk1	✅ 🛠 wpe-skia
✅ 🛠 🧪 jsc	✅ 🧪 api-ios	✅ 🧪 mac-wk2	✅ 🛠 gtk
✅ 🛠 🧪 jsc-arm64	✅ 🛠 tv	✅ 🧪 mac-AS-debug-wk2	✅ 🧪 gtk-wk2
	✅ 🛠 tv-sim		✅ 🧪 api-gtk
✅ 🛠 🧪 merge	✅ 🛠 watch		✅ 🛠 jsc-armv7
	✅ 🛠 watch-sim		✅ 🧪 jsc-armv7-tests

[shallawa]

What is the rule for removing an ENABLE() flag if it is unconditionally set to true? Currently we have all these flags set to true. This list includes ENABLE_MATHML and the ENABLE_OFFSCREEN_CANVAS.

#if !defined(ENABLE_CONTEXT_MENUS)
#define ENABLE_CONTEXT_MENUS 1
#endif

#if !defined(ENABLE_CONTEXT_MENU_EVENT)
#define ENABLE_CONTEXT_MENU_EVENT 1
#endif

#if !defined(ENABLE_CSS_PAINTING_API)
#define ENABLE_CSS_PAINTING_API 1
#endif

#if !defined(ENABLE_CUSTOM_CURSOR_SUPPORT)
#define ENABLE_CUSTOM_CURSOR_SUPPORT 1
#endif

#if !defined(ENABLE_DRAG_SUPPORT)
#define ENABLE_DRAG_SUPPORT 1
#endif

#if !defined(ENABLE_FTPDIR)
#define ENABLE_FTPDIR 1
#endif

#if !defined(ENABLE_INPUT_TYPE_COLOR)
#define ENABLE_INPUT_TYPE_COLOR 1
#endif

#if !defined(ENABLE_MATHML)
#define ENABLE_MATHML 1
#endif

#if !defined(ENABLE_OFFSCREEN_CANVAS)
#define ENABLE_OFFSCREEN_CANVAS 1
#endif

#if !defined(ENABLE_OFFSCREEN_CANVAS_IN_WORKERS)
#define ENABLE_OFFSCREEN_CANVAS_IN_WORKERS 1
#endif

#if !defined(ENABLE_POINTER_LOCK)
#define ENABLE_POINTER_LOCK 1
#endif

#if !defined(ENABLE_TEXT_CARET)
#define ENABLE_TEXT_CARET 1
#endif

#if !defined(ENABLE_TEXT_SELECTION)
#define ENABLE_TEXT_SELECTION 1
#endif

#if !defined(ENABLE_XSLT)
#define ENABLE_XSLT 1
#endif

So why do we care about ENABLE_CSS_PAINTING_API only and ignore the reset? Or are we planning to remove all of these flags?

[donny-dont]

What is the rule for removing an ENABLE() flag if it is unconditionally set to true? Currently we have all these flags set to true. This list includes ENABLE_MATHML and the ENABLE_OFFSCREEN_CANVAS.
....

So why do we care about ENABLE_CSS_PAINTING_API only and ignore the reset? Or are we planning to remove all of these flags?

@shallawa myself and @annevk have been going through the list and removing ENABLEs one by one.

Algorithmically it boils down to. Does the ENABLE flag require port specific code? Something like ENABLE_XSLT does not pass that because it requires an implementation of libxslt, while the CSS Painting API is port agnostic so it passes this check. Does the ENABLE flag have a runtime option? Yes for CSS Painting API. Since it passes those checks its up for removal.

[webkit-commit-queue]

Committed 277584@main (4d1720e): https://commits.webkit.org/277584@main

Reviewed commits have been landed. Closing PR #26627 and removing active labels.

[annevk]

PlatformEnable.h is unfortunately not sufficient to determine if a flag is always true for all ports. You also have to look at the CMake code. And yeah, generally the less branches we have to (pretend to) maintain the better.