[Webkit-unassigned] [Bug 182757] [CMake][Ninja] Fails to compile TestWebCore due to missing WebCore's derived headers

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sat Feb 17 07:51:18 PST 2018


https://bugs.webkit.org/show_bug.cgi?id=182757

--- Comment #11 from Konstantin Tokarev <annulen at yandex.ru> ---
Comment on attachment 334072
  --> https://bugs.webkit.org/attachment.cgi?id=334072
WIP Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=334072&action=review

>> Source/JavaScriptCore/CMakeLists.txt:15
>> +    "${CMAKE_BINARY_DIR}/JavaScriptCore.framework/PrivateHeaders"
> 
> This follows along with how Apple builds operate.
> 
> Stuff in PrivateHeaders does a #include <JavaScriptCore/Foo.h> so this would likely catch any compile errors. Also LLInt needs these headers installed.

I don't think it makes any sense to copy layout of macOS framework when there is no actual framework built.

Unless we are 100% sure that we don't want to use built-in cmake's support for frameworks on macOS, and even then having .framework on other platforms may be confusing

>> Source/cmake/WebKitMacros.cmake:265
>> +            add_custom_command(TARGET ${framework} PRE_LINK
> 
> So this moves the DerivedSources to being copied within the original ${framework} as a PRE_LINK step which happens before a POST_BUILD. Seems like a safe bet that things are going to be ok at this point.
> 
> I need to see how things work with WebCore to really determine if this will work across the board.
> 
> It seems to not cause rebuilds within Visual Studio so I think it solves things. Placing them within PrivateFrameworkHeaders causes the rebuilds in visual studio.

Did you check with Ninja generator? In general it's better to stay away from PRE_LINK, POST_BUILD etc., these things don't have well-defined behavior outside of Visual Studio generator

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20180217/dd2b5952/attachment.html>


More information about the webkit-unassigned mailing list