Bug 147815 – [GTK] webkitgtk 2.8.5 fails to build in mips, powerpc and others using GCC 5

• Status: RESOLVED
• Resolution: FIXED
• Component: WebKitGTK

[Alberto Garcia]

Here's the error message:

CMakeFiles/JavaScriptCore.dir/runtime/JSArray.cpp.o: In function `JSC::JSArray::push(JSC::ExecState*, JSC::JSValue)':
JSArray.cpp:(.text+0x4170): undefined reference to `void JSC::JSObject::putByIndexBeyondVectorLengthWithoutAttributes<(unsigned char)20>(JSC::ExecState*, unsigned int, JSC::JSValue)'
collect2: error: ld returned 1 exit status

This may be related to the usage of GCC 5. Here's what appears to be the same error in QtWebKit:

https://bugreports.qt.io/browse/QTBUG-44829

[Zack Weinberg]

Created attachment 259741 [details]
patch (from QtWebkit bug tracker)

Patch, by Khem Raj, hoisted from https://codereview.qt-project.org/#/c/107921/ & https://bugreports.qt.io/browse/QTBUG-44829 .  Line numbers may be off.

This doesn't appear to have anything to do with GTK (or Qt) -- it's completely generic JSC code.

I can confirm that this fixes a failure to link the 'jsc' executable, but I have not verified that there are no other such problems in the rest of the current Webkit development sources.

[Michael Catanzaro]

FWIW we've been carrying this patch in Fedora for a while

[Alberto Garcia]

I tested it with webkitgtk 2.8.5 in s390x and I confirm that it works. Thanks!

[Carlos Garcia Campos]

Thanks for the patch, we need a proper patch that applies on current trunk and  with a ChangeLog entry, though. See http://www.webkit.org/coding/contributing.html

[Zack Weinberg]

Created attachment 260169 [details]
Patch

[Zack Weinberg]

(In reply to comment #4)
> Thanks for the patch, we need a proper patch that applies on current trunk
> and  with a ChangeLog entry, though. See
> http://www.webkit.org/coding/contributing.html

Sorry for the delay, I don't seem to be getting bugmail from this tracker for some reason.  Anyway, the current patch should apply cleanly to trunk, has a changelog entry, and passes 'webkit-patch upload''s linter.  Note that I corrected the bug title in the changelog entry; I would do that here too but I don't have the necessary privileges.  I also made the changelog entry reflect the original author of the fix (Khem Raj).

[WebKit Commit Bot]

Attachment 260169 [details] did not pass style-queue:


ERROR: Source/JavaScriptCore/ChangeLog:9:  Line contains tab character.  [whitespace/tab] [5]
Total errors found: 1 in 2 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Zack Weinberg]

(In reply to comment #7)
> Attachment 260169 [details] did not pass style-queue:
> ERROR: Source/JavaScriptCore/ChangeLog:9:  Line contains tab character. 
> [whitespace/tab] [5]
> Total errors found: 1 in 2 files

This is the first project I have ever tripped over in which a GNU-style ChangeLog file was *not* to be indented with a single hard tab?

Anyway I would ask whoever commits this patch to fix up this trivial error for me; it does not seem worth respinning the whole thing over.  You're going to have to edit the changelog to put yourself on the "Reviewed by" line anyway ;-)

[Filip Pizlo]

(In reply to comment #8)
> (In reply to comment #7)
> > Attachment 260169 [details] did not pass style-queue:
> > ERROR: Source/JavaScriptCore/ChangeLog:9:  Line contains tab character. 
> > [whitespace/tab] [5]
> > Total errors found: 1 in 2 files
> 
> This is the first project I have ever tripped over in which a GNU-style
> ChangeLog file was *not* to be indented with a single hard tab?
> 
> Anyway I would ask whoever commits this patch to fix up this trivial error
> for me; it does not seem worth respinning the whole thing over.  You're
> going to have to edit the changelog to put yourself on the "Reviewed by"
> line anyway ;-)

If you would like this change landed, then you should submit a patch that obeys our style.

Also, patches by non-committers are usually landed by the commit-queue after a committer grants the cq+ flag.  The commit-queue does not automatically fix style errors for you.  So, what you're asking is for is not only that some reviewer do some work for you, but that they land the patch for you rather than using the commit-queue.  Most reviewers have more important things to do than fix other people's style errors for them.

[Filip Pizlo]

Comment on attachment 260169 [details]
Patch

R- because you should fix the ChangeLog style.

[Zack Weinberg]

Created attachment 260184 [details]
patch v3 (tab character removed)

Downloaded patch, edited out offending tab character, put it back.

[Filip Pizlo]

Comment on attachment 260184 [details]
patch v3 (tab character removed)

Thanks!  You can mark the cq flag to "?" if you want someone to land it for you.  Otherwise, we assume that you will land it yourself.

[WebKit Commit Bot]

Comment on attachment 260184 [details]
patch v3 (tab character removed)

Clearing flags on attachment: 260184

Committed r189133: <http://trac.webkit.org/changeset/189133>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.