Bug 72924 – CSSValue: reorder ClassType enum to allow faster comparisons, add COMPILE_ASSERT on class size.

• Component: CSS
• Status: RESOLVED FIXED

[Luke Macpherson]

CSSValue: reorder ClassType enum to allow faster comparisons, add COMPILE_ASSERT on class size.

[Luke Macpherson]

Created attachment 116162 [details]
Patch

[Andreas Kling]

Comment on attachment 116162 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=116162&action=review

Good idea, I'd like one more iteration though.

> Source/WebCore/css/CSSValue.h:97
> +        /* Primitive class types must appear before PrimitiveClass */

Coding style, we use C++ comments (//), and end them in periods.

> Source/WebCore/css/CSSValue.h:103
> +        /* Image generator classes */

Ditto.

> Source/WebCore/css/CSSValue.h:109
> +        /* Timing function classes */

Ditto.

> Source/WebCore/css/CSSValue.h:114
> +        /* Other class types */

Ditto.

> Source/WebCore/css/CSSValue.h:139
> +        /* List class types must appear after ValueListClass */

Ditto.

> Source/WebCore/css/CSSValue.h:145
> +        /* Do not append non-list class types here */

Ditto.

> Source/WebCore/css/CSSValue.h:157
> +        COMPILE_ASSERT((sizeof(CSSValue) - sizeof(RefCounted<CSSValue>)) <= 4, CSS_value_packs_into_four_bytes);

This should be in CSSValue.cpp instead.

> Source/WebCore/css/CSSValue.h:173
> -    unsigned m_primitiveUnitType : 7; // CSSPrimitiveValue::UnitTypes
> +    unsigned char m_primitiveUnitType : 7; // CSSPrimitiveValue::UnitTypes

I suppose you're making this change because of MSVC? At least note it in the ChangeLog as it's unrelated to the rest of the patch.

[Luke Macpherson]

Created attachment 116285 [details]
Patch

[Luke Macpherson]

Patch addresses those issues. I've changed back from unsigned char to unsigned - since we are only trying to pack 15 bits into 32 (behavior due to extending the 32-bit RefPtr), we probably don't care too much that MSVC doesn't pack it tightly.

[Luke Macpherson]

(In reply to comment #4)
> Patch addresses those issues. I've changed back from unsigned char to unsigned - since we are only trying to pack 15 bits into 32 (behavior due to extending the 32-bit RefPtr), we probably don't care too much that MSVC doesn't pack it tightly.

s/RefPtr/RefCounted

[Luke Macpherson]

Created attachment 116317 [details]
Patch

[Andreas Kling]

Comment on attachment 116317 [details]
Patch

r=me

[WebKit Review Bot]

Comment on attachment 116317 [details]
Patch

Rejecting attachment 116317 [details] from commit-queue.

Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 2

Last 500 characters of output:
Value.o
  CXX(target) out/Debug/obj.target/webcore_remaining/Source/WebCore/css/CSSValue.o
  CXX(target) out/Debug/obj.target/webcore_remaining/Source/WebCore/css/CSSValueList.o
Source/WebCore/css/CSSValue.cpp:62: error: size of array 'CSS_value_packs_into_four_bytes' is negative
  CXX(target) out/Debug/obj.target/webcore_remaining/Source/WebCore/css/CSSWrapShapes.o
make: *** [out/Debug/obj.target/webcore_remaining/Source/WebCore/css/CSSValue.o] Error 1
make: *** Waiting for unfinished jobs....

Full output: http://queues.webkit.org/results/10645027

[Luke Macpherson]

Created attachment 116472 [details]
Patch for landing

[WebKit Review Bot]

Comment on attachment 116472 [details]
Patch for landing

Clearing flags on attachment: 116472

Committed r101119: <http://trac.webkit.org/changeset/101119>

[WebKit Review Bot]

All reviewed patches have been landed.  Closing bug.