WebKit Bugzilla
New
Browse
Search+
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
150700
[Win] build-jsc and run-javascriptcore-tests do not work
https://bugs.webkit.org/show_bug.cgi?id=150700
Summary
[Win] build-jsc and run-javascriptcore-tests do not work
Brent Fulgham
Reported
2015-10-29 16:41:49 PDT
The "run-javascript-core" script does not work following our switch to CMake on Windows, because it attempts to build using the increasingly-out-of-date "Source/JavaScriptCore/JavaScriptCore.vcxproj/JavaScriptCore.sln" solution to build the various pieces of JSC. The actual failure was in the "build-jsc" script, which does not use the new CMake build system for Windows. This patch corrects this oversight, allowing JSC developers to build only the JavaScript-related files. (820 files instead of 3500 files!)
Attachments
Patch
(5.43 KB, patch)
2015-10-29 16:49 PDT
,
Brent Fulgham
mark.lam
: review+
Details
Formatted Diff
Diff
View All
Add attachment
proposed patch, testcase, etc.
Brent Fulgham
Comment 1
2015-10-29 16:49:08 PDT
Created
attachment 264365
[details]
Patch
Mark Lam
Comment 2
2015-10-29 16:52:00 PDT
Comment on
attachment 264365
[details]
Patch r=me
Alex Christensen
Comment 3
2015-10-29 16:55:04 PDT
lgtm, too
Brent Fulgham
Comment 4
2015-10-29 17:09:48 PDT
The windows EWS failure looks like a bot problem. I'm investigating.
Brent Fulgham
Comment 5
2015-10-29 17:20:55 PDT
Committed
r191766
: <
http://trac.webkit.org/changeset/191766
>
Michael Saboff
Comment 6
2015-10-29 18:16:32 PDT
After this change, I now can't build JavaScriptCore or WebKit. I get: $ build-webkit --release --64-bit Use of uninitialized value $expectedVersion in scalar chomp at Tools/Scripts/update-webkit-support-libs line 113. Falling back to existing version of WebKitSupportLibrary. Error: /home/msaboff/src/OpenSource/WebKitBuild/Release is not a directory The directory does exist. $ ls /home/msaboff/src/OpenSource/WebKitBuild/Release ALL_BUILD.vcxproj* cmakeconfig.h* INSTALL.vcxproj* RUN_TESTS.vcxproj* x64/ ALL_BUILD.vcxproj.filters* CMakeFiles/ INSTALL.vcxproj.filters* RUN_TESTS.vcxproj.filters* ZERO_CHECK.vcxproj* bin64/ CPackConfig.cmake* lib32/ Source/ ZERO_CHECK.vcxproj.filters* buildfailed* CPackSourceConfig.cmake* lib64/ Tools/ build-webkit-options.txt CTestTestfile.cmake* obj64/ WebKit.sln* cmake_install.cmake* DerivedSources/ PACKAGE.vcxproj* webkit_errors.log* CMakeCache.txt* include/ PACKAGE.vcxproj.filters* webkit_warnings.log* If I roll back to
r191765
, I can build.
Brent Fulgham
Comment 7
2015-10-29 19:42:48 PDT
I'll fix ASAP!
Csaba Osztrogonác
Comment 8
2015-10-30 03:23:45 PDT
(In reply to
comment #5
)
> Committed
r191766
: <
http://trac.webkit.org/changeset/191766
>
Apple Windows and WinCairo build is still broken.
Brent Fulgham
Comment 9
2015-10-30 10:06:46 PDT
Committed
r191793
: <
http://trac.webkit.org/changeset/191793
>
Csaba Osztrogonác
Comment 10
2015-11-10 03:52:55 PST
(In reply to
comment #9
)
> Committed
r191793
: <
http://trac.webkit.org/changeset/191793
>
Originally you landed this change in
r191766
: - buildCMakeProjectOrExit(0, cmakeBasedPortName(), undef, "jsc $makeArgs", (cmakeBasedPortArguments(), $cmakeArgs)); + my $testapi = (isAnyWindows() ? "testapi" : ""); + buildCMakeProjectOrExit(0, cmakeBasedPortName(), undef, "jsc $testapi $makeArgs", (cmakeBasedPortArguments(), $cmakeArgs)); And this one in
r191793
: - # By default we build using all of the available CPUs - $makeArgs .= ($makeArgs ? " " : "") . "-j" . numberOfCPUs() if $makeArgs !~ /-j\s*\d+/; + + my $buildTarget = ""; + unless (isAnyWindows()) { + # By default we build using all of the available CPUs + $makeArgs .= ($makeArgs ? " " : "") . "-j" . numberOfCPUs() if $makeArgs !~ /-j\s*\d+/; + $buildTarget = "jsc $makeArgs"; + } elsif (canUseNinja()) { + $buildTarget .= "jsc testapi"; + } Could you explain why -jN causes problem for Windows-ninja ?
Csaba Osztrogonác
Comment 11
2015-11-12 03:58:26 PST
Brent?
Csaba Osztrogonác
Comment 12
2015-11-17 05:23:19 PST
(In reply to
comment #11
)
> Brent?
ping?
Csaba Osztrogonác
Comment 13
2015-11-18 02:36:26 PST
(In reply to
comment #11
)
> Brent?
ping?
Csaba Osztrogonác
Comment 14
2015-11-19 04:42:00 PST
(In reply to
comment #11
)
> Brent?
ping?
Csaba Osztrogonác
Comment 15
2015-11-24 10:09:37 PST
(In reply to
comment #11
)
> Brent?
ping?
Csaba Osztrogonác
Comment 16
2015-11-25 08:54:28 PST
(In reply to
comment #11
)
> Brent?
ping?
Csaba Osztrogonác
Comment 17
2015-11-26 04:42:25 PST
Brent, it's not fair that you landed an unreviewed hack 2 weeks before and you don't answer my question. Please answer as soon as possible.
Csaba Osztrogonác
Comment 18
2015-12-01 01:09:30 PST
ping?
Alex Christensen
Comment 19
2015-12-01 11:02:02 PST
The unless isAnyWindows is not necessary. It works without it. The canUseNinja is necessary because our bots use CMake and Visual Studio without Ninja. If we change it, though, the testb3 that is there now shouldn't be a target on Windows because b3 isn't enabled on Windows.
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug