New builds of clang triggered warnings-as-errors when compiling JavaScriptCore. All of the issues were in the form of conspicuous uses of the |this| pointer: Source/JavaScriptCore/API/OpaqueJSString.cpp:59:10: error: 'this' pointer cannot be null in well-defined C++ code; pointer may be assumed to always convert to true [-Werror,-Wundefined-bool-conversion] One use is covered by Bug 131578; fixing the rest are covered by this bug.
<rdar://problem/18020616>
Created attachment 237174 [details] Patch v1
Comment on attachment 237174 [details] Patch v1 Attachment 237174 [details] did not pass mac-wk2-ews (mac-wk2): Output: http://webkit-queues.appspot.com/results/5582077030301696 New failing tests: editing/selection/select-line.html http/tests/navigation/anchor-basic.html editing/execCommand/delete-image-in-anchor.html editing/selection/5195166-1.html http/tests/navigation/javascriptlink-basic.html http/tests/navigation/back-to-slow-frame.html editing/deleting/delete-to-end-of-paragraph.html http/tests/navigation/forward-and-cancel.html editing/execCommand/move-up-down-should-skip-hidden-elements.html editing/pasteboard/emacs-ctrl-k-with-move.html fast/images/animated-gif-restored-from-bfcache.html editing/selection/selection-actions.html editing/deleting/smart-editing-disabled-mac.html editing/deleting/smart-editing-disabled-win.html http/tests/navigation/go-back-to-error-page.html editing/selection/4947387.html editing/style/make-text-writing-direction-inline-mac.html editing/style/make-text-writing-direction-inline-win.html editing/input/editable-container-with-word-wrap-normal.html http/tests/navigation/error404-goback.html editing/deleting/delete-surrogatepair.html http/tests/navigation/error404-basic.html editing/selection/5195166-2.html editing/deleting/smart-delete-004.html fast/history/go-back-to-changed-name.html http/tests/navigation/javascriptlink-frames.html http/tests/navigation/anchor-goback.html editing/deleting/5300379.html editing/selection/extend-forward-after-set-base-and-extent.html editing/deleting/smart-delete-003.html
Created attachment 237227 [details] Archive of layout-test-results from webkit-ews-11 for mac-mountainlion-wk2 The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews. Bot: webkit-ews-11 Port: mac-mountainlion-wk2 Platform: Mac OS X 10.8.5
Comment on attachment 237174 [details] Patch v1 Wunderbar. Need more NULL checks added.
Created attachment 237593 [details] Patch v2
(In reply to comment #6) > Created an attachment (id=237593) [details] > Patch v2 Added one change to Source/WebKit2/Shared/API/c/WKString.cpp to fix the layout test crashes.
Comment on attachment 237593 [details] Patch v2 r=me
Comment on attachment 237593 [details] Patch v2 Clearing flags on attachment: 237593 Committed r173245: <http://trac.webkit.org/changeset/173245>
All reviewed patches have been landed. Closing bug.
(In reply to comment #9) > (From update of attachment 237593 [details]) > Clearing flags on attachment: 237593 > > Committed r173245: <http://trac.webkit.org/changeset/173245> It broke the JSC testapi tests, and now JSC tests don't run on the bots at all because of this regression.
2014-09-04 01:01:29.768 testapi[53723:507] ERROR: Class ClassC exported more than one init family method via JSExport. Class ClassC will not have a callable JavaScript constructor function. 2014-09-04 01:01:29.770 testapi[53723:507] ERROR: Class ClassCPrime exported more than one init family method via JSExport. Class ClassCPrime will not have a callable JavaScript constructor function.
(In reply to comment #11) > (In reply to comment #9) > > (From update of attachment 237593 [details] [details]) > > Clearing flags on attachment: 237593 > > > > Committed r173245: <http://trac.webkit.org/changeset/173245> > > It broke the JSC testapi tests, and now JSC tests don't run on the bots at all because of this regression. Has it been rolled out? I'm now surprised that EWS doesn't run JSC tests. :(
Re-opened since this is blocked by bug 136533
(In reply to comment #13) > (In reply to comment #11) > > (In reply to comment #9) > > > (From update of attachment 237593 [details] [details] [details]) > > > Clearing flags on attachment: 237593 > > > > > > Committed r173245: <http://trac.webkit.org/changeset/173245> > > > > It broke the JSC testapi tests, and now JSC tests don't run on the bots at all because of this regression. > > Has it been rolled out? I'm now surprised that EWS doesn't run JSC tests. :( I just marked the roll-out bug as cq+ so that it will apply soon.
(In reply to comment #12) > 2014-09-04 01:01:29.768 testapi[53723:507] ERROR: Class ClassC exported more than one init family method via JSExport. Class ClassC will not have a callable JavaScript constructor function. > 2014-09-04 01:01:29.770 testapi[53723:507] ERROR: Class ClassCPrime exported more than one init family method via JSExport. Class ClassCPrime will not have a callable JavaScript constructor function. One would think that this output was the cause of the failure, but it's actually expected. :( The testapi process was crashing due to a NULL-dereference.
Results for JSC stress tests: 0 failures found. OK. Yay.
Created attachment 237698 [details] Patch v3
(In reply to comment #18) > Created an attachment (id=237698) [details] > Patch v3 All the tests in run-javascriptcore-tests now pass in Debug mode. Testing Release now; will report back when completed.
(In reply to comment #19) > (In reply to comment #18) > > Created an attachment (id=237698) [details] [details] > > Patch v3 > > All the tests in run-javascriptcore-tests now pass in Debug mode. > Testing Release now; will report back when completed. Results for JSC stress tests: 0 failures found. OK. No issues in Release mode either.
Comment on attachment 237698 [details] Patch v3 r=me
Comment on attachment 237698 [details] Patch v3 Clearing flags on attachment: 237698 Committed r173326: <http://trac.webkit.org/changeset/173326>