Bug 102903 – [EFL] Wrong enabled features reported

• Component: WebKit EFL
• Status: RESOLVED DUPLICATE

[Halton Huo]

Build trunk@135354 on linux EFL port, it will dump me Enabled messages as following:
-- Enabled features:
--  ENABLE_API_TESTS ........................ ON
--  ENABLE_3D_RENDERING                       OFF
[...]
--  ENABLE_WEBGL                              OFF
[...]
--  WTF_USE_TILED_BACKING_STORE               OFF
--  WTF_USE_WTFURL .......................... OFF

But actually the EFL port is enable WTF_USE_TILED_BACKING_STORE by default in OpionsEfl.cmake. And ENABLE_3D_RENDERING and ENABLE_WEBGL is also enabled in when WTF_USE_TILED_BACKING_STORE is ON.

[Halton Huo]

Created attachment 175383 [details]
Patch

[EFL EWS Bot]

Comment on attachment 175383 [details]
Patch

Attachment 175383 [details] did not pass efl-ews (efl):
Output: http://queues.webkit.org/results/14894924

[Halton Huo]

Bug #102910 is filed to track the compiling issue on CSSComputedStyleDeclaration.cpp. I do not think my patch cause this failure.

[Halton Huo]

Created attachment 175411 [details]
Patch

[Halton Huo]

Bug 102910 is a regression with my first patch, so close that bug. The problem in 1st patch is only enabled features will be added to FEATURE_DEFINES_WITH_SPACE_SEPARATOR, this cause makeprop.pl generate wrong CSSPropertyNames.[cpp|gperf|h].

The 2nd patch fix that problem, the message will be set by using value of ${_WEBKIT_AVAILABLE_OPTIONS_INITALVALUE_${_name}} and move out from  IF (${_name}) logic. The FEATURE_DEFINES_WITH_SPACE_SEPARATOR logic keep in the IF (${_name}) logic.

Please review.

[Raphael Kubo da Costa (:rakuco)]

While this could solve the problem, this would make the list of features we enable in OptionsEFL.cmake different from the list in FeaturesList.pm, since we don't enable WebGL by default and 3D Rendering is enabled as a side-effect of tiled backing store being on by default (this one is already correctly set in both OptionsEFL.cmake and FeaturesList.pm).

I still think that a good solution to the problem would be something along the lines of what I described in comment #4 of bug 100829.

*** This bug has been marked as a duplicate of bug 100829 ***