RESOLVED FIXED 177364
REGRESSION(r221909): Failing fast/text/international/iso-8859-8.html
https://bugs.webkit.org/show_bug.cgi?id=177364
Summary REGRESSION(r221909): Failing fast/text/international/iso-8859-8.html
Zan Dobersek
Reported 2017-09-22 07:27:13 PDT
Created attachment 321545 [details] Diff image for fast/text/word-space-between-inlines.html failure The following two layout tests fail after changes in r221909: - fast/text/international/iso-8859-8.html - fast/text/word-space-between-inlines.html https://trac.webkit.org/changeset/221909/webkit The first one fails with a text failure: --- /home/slave/webkitgtk/gtk-linux-64-release-tests/build/layout-test-results/fast/text/international/iso-8859-8-expected.txt +++ /home/slave/webkitgtk/gtk-linux-64-release-tests/build/layout-test-results/fast/text/international/iso-8859-8-actual.txt @@ -1,4 +1,4 @@ Test rendering of ISO-8859-8. -PASS Test completed successfully +FAIL Test 1 LTR: character should be rendered right of כ The second one is a reference test, but fails with a minimum difference between the actual and expected renderings (0.01%). Diff image is attached.
Attachments
Diff image for fast/text/word-space-between-inlines.html failure (750 bytes, image/png)
2017-09-22 07:27 PDT, Zan Dobersek
no flags
A test content to check the position of the trainling space (559 bytes, text/html)
2017-10-02 21:13 PDT, Fujii Hironori
no flags
Patch (5.33 KB, patch)
2017-10-02 23:53 PDT, Fujii Hironori
no flags
Fujii Hironori
Comment 1 2017-09-26 20:16:53 PDT
fast/text/word-space-between-inlines.html fails in ports enabling the advance text rendering mode, Cocoa and FreeType ports. Bug 142086 tracks the failure.
Fujii Hironori
Comment 2 2017-10-02 03:11:50 PDT
fast/text/international/iso-8859-8.html outputs the following error message to stderr in Debug build: > ERROR: Shaper couldn't shape text run. > /webkit/Source/WebCore/platform/graphics/cairo/FontCairoHarfbuzzNG.cpp(97) : void WebCore::FontCascade::adjustSelectionRectForComplexText(const WebCore::TextRun&, WebCore::LayoutRect&, unsigned int, unsigned int) const
Fujii Hironori
Comment 3 2017-10-02 04:12:18 PDT
The second div element has 8 chracters. > <div class="test">שנב גקכ </div> Its RenderText removes the last trailing space. Then, it has 7 characters, "שנב גקכ". getClientRects tried get the position of the 7th character (the removed space). Then, HarfBuzzShaper::shape failed.
Fujii Hironori
Comment 4 2017-10-02 21:13:11 PDT
Created attachment 322490 [details] A test content to check the position of the trainling space I checked the position of the trainling space. Google Chrome 61: left:78 width:0 Firefox 56 : left:63.98 width:0 Edge : left:56.79 width:4 Mac port : left:68 width:0 GTK+ port : left:11 width:57 Screenshots: https://photos.app.goo.gl/HpfoE56etUyYj2092 GTK+ port returns the rect of the whole sentence.
Fujii Hironori
Comment 5 2017-10-02 23:53:43 PDT
WebKit Commit Bot
Comment 6 2017-10-03 10:27:17 PDT
The commit-queue encountered the following flaky tests while processing attachment 322495 [details]: fast/events/popup-blocking-timers3.html bug 177822 (authors: cdumez@apple.com, jochen@chromium.org, and mark.lam@apple.com) http/tests/security/window-properties-clear-domain.html bug 177823 (authors: koivisto@iki.fi and rniwa@webkit.org) The commit-queue is continuing to process your patch.
WebKit Commit Bot
Comment 7 2017-10-03 11:39:34 PDT
Comment on attachment 322495 [details] Patch Clearing flags on attachment: 322495 Committed r222792: <http://trac.webkit.org/changeset/222792>
WebKit Commit Bot
Comment 8 2017-10-03 11:39:35 PDT
All reviewed patches have been landed. Closing bug.
Note You need to log in before you can comment on or make changes to this bug.