31804 2009-11-23 06:33:27 -0800 Safari/Chromium for Windows fails to load many CJK WebFonts and some Latin WebFonts 2009-12-02 11:18:02 -0800 1 1 1 Unclassified WebKit WebCore Misc. 528+ (Nightly build) PC Windows XP RESOLVED FIXED P2 Normal --- 1 yusukes webkit-unassigned abarth agl aroben commit-queue eric hyatt mitz oldest_to_newest 165843 0 yusukes 2009-11-23 06:33:27 -0800 Safari for Windows fails to load OpenType web fonts (@font-face) that have 2 or more faces. Chromium for Windows also fails to load TrueType and OpenType web fonts that have 2 or more faces. This is because the renameAndActivateFont() function in WebCore/platform/graphics/opentype/OpenTypeUtilities.cpp, which is used by Safari for Windows and Chromium for Windows, explicitly refuses such fonts. However, since most of CJK ttf/otf fonts and some of Latin ttf/otf fonts have more than 2 faces (e.g., one is for horizontal writing and the other is for vertical writing), the restriction seems to be too strict. For example, HIRAGINO (ヒラギノ) font, which is the default Japanese OpenType font for Mac OS X, has 2 faces. I'll attach a one-line patch that allows WebKit to load such fonts. Just for reference, Safari for Mac seems not to have the restriction. I tried to load the HIRAGINO font to Safari for Mac via @font-face (on my local machine, of course), and it succeeded. A comment in WebCore/platform/graphics/mac/FontCustomPlatformData.cpp says that when a font has more than 2 faces, the first one is used for rendering, and I believe this behavior is good enough for most cases. 165845 1 43711 yusukes 2009-11-23 06:41:43 -0800 Created attachment 43711 allow_multiple_faces_v1 166624 2 eric 2009-11-25 21:18:49 -0800 mitz wrote the code, hyatt reviewed it. The change looks fine to me, but perhaps mitz had a reason for making this != 1 instead of < 1. I would suggest we let mitz comment before this is landed. 167232 3 yusukes 2009-11-30 07:15:50 -0800 Slightly modified the summary so that it states the actual circumstance of the issue. An example of load failure of a Latin webfont is: http://code.google.com/p/chromium/issues/detail?id=28567 mitz, can you please take a look at the patch? 167346 4 abarth 2009-11-30 12:35:25 -0800 style-queue successfully ran check-webkit-style on attachment 43711 without any errors 167659 5 mitz 2009-12-01 10:00:20 -0800 I am trying to understand how this is supposed to work. On my Mac OS X v10.6 system, there are multiple font files whose name begins with ヒラギノ. Does one of them contain multiple faces? Which one? 167807 6 yusukes 2009-12-01 17:39:18 -0800 (In reply to comment #5) > I am trying to understand how this is supposed to work. On my Mac OS X v10.6 > system, there are multiple font files whose name begins with ヒラギノ. Does one of > them contain multiple faces? Which one? Please try "ヒラギノ角ゴ ProN W6", for example. $ ls -l /System/Library/Fonts/ヒラギノ角ゴ\ ProN\ W6.otf -rw-r--r-- 1 root wheel 7515064 May 31 2008 /System/Library/Fonts/ヒラギノ角ゴ ProN W6.otf 168007 7 mitz 2009-12-02 08:54:38 -0800 (In reply to comment #3) > An example of load failure of a Latin webfont is: > http://code.google.com/p/chromium/issues/detail?id=28567 The URL in that bug, <http://www.alistapart.com/d/cssatten/stef.html>, is rendered correctly in Safari on Windows with TOT WebKit. I confirmed that even if I skip the TTLoadEmbeddedFont() call in createFontCustomPlatformData() and take the renameAndActivateFont() branch, all fonts on that page load correctly. 168015 8 mitz 2009-12-02 09:20:26 -0800 (In reply to comment #6) > Please try "ヒラギノ角ゴ ProN W6", for example. I could reproduce the problem with that font, i.e. numFonts was 2. 168022 9 43711 eric 2009-12-02 09:28:45 -0800 Comment on attachment 43711 allow_multiple_faces_v1 Marking cq+ since Yusuke isn't a committer yet. 168087 10 43711 commit-queue 2009-12-02 11:17:54 -0800 Comment on attachment 43711 allow_multiple_faces_v1 Clearing flags on attachment: 43711 Committed r51605: <http://trac.webkit.org/changeset/51605> 168089 11 commit-queue 2009-12-02 11:18:02 -0800 All reviewed patches have been landed. Closing bug. 43711 2009-11-23 06:41:43 -0800 2009-12-02 11:17:54 -0800 allow_multiple_faces_v1 bug-31804-20091123234141.patch text/plain 1316 yusukes ZGlmZiAtLWdpdCBhL1dlYkNvcmUvQ2hhbmdlTG9nIGIvV2ViQ29yZS9DaGFuZ2VMb2cKaW5kZXgg NDU2MDc4Yy4uOWJiNjYzNCAxMDA2NDQKLS0tIGEvV2ViQ29yZS9DaGFuZ2VMb2cKKysrIGIvV2Vi Q29yZS9DaGFuZ2VMb2cKQEAgLTEsMyArMSwxMyBAQAorMjAwOS0xMS0yMyAgWXVzdWtlIFNhdG8g IDx5dXN1a2VzQGNocm9taXVtLm9yZz4KKworICAgICAgICBSZXZpZXdlZCBieSBOT0JPRFkgKE9P UFMhKS4KKworICAgICAgICBTYWZhcmkvQ2hyb21pdW0gZm9yIFdpbmRvd3MgZmFpbHMgdG8gbG9h ZCBDSksgV2ViRm9udHMKKyAgICAgICAgaHR0cHM6Ly9idWdzLndlYmtpdC5vcmcvc2hvd19idWcu Y2dpP2lkPTMxODA0CisgICAgICAgIAorICAgICAgICAqIHBsYXRmb3JtL2dyYXBoaWNzL29wZW50 eXBlL09wZW5UeXBlVXRpbGl0aWVzLmNwcDoKKyAgICAgICAgKFdlYkNvcmU6OnJlbmFtZUFuZEFj dGl2YXRlRm9udCk6IExvYWQgYSByZW1vdGUgZm9udCBldmVuIGlmIHRoZSBmb250IGhhcyAyIG9y IG1vcmUgZmFjZXMuCisKIDIwMDktMTEtMjMgIFNpbW9uIEhhdXNtYW5uICA8c2ltb24uaGF1c21h bm5Abm9raWEuY29tPgogCiAgICAgICAgIFJldmlld2VkIGJ5IEtlbm5ldGggUm9oZGUgQ2hyaXN0 aWFuc2VuLgpkaWZmIC0tZ2l0IGEvV2ViQ29yZS9wbGF0Zm9ybS9ncmFwaGljcy9vcGVudHlwZS9P cGVuVHlwZVV0aWxpdGllcy5jcHAgYi9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL29wZW50eXBl L09wZW5UeXBlVXRpbGl0aWVzLmNwcAppbmRleCBiMmUzZDkyLi4zYTYwMTYwIDEwMDY0NAotLS0g YS9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL29wZW50eXBlL09wZW5UeXBlVXRpbGl0aWVzLmNw cAorKysgYi9XZWJDb3JlL3BsYXRmb3JtL2dyYXBoaWNzL29wZW50eXBlL09wZW5UeXBlVXRpbGl0 aWVzLmNwcApAQCAtNDM1LDcgKzQzNSw3IEBAIEhBTkRMRSByZW5hbWVBbmRBY3RpdmF0ZUZvbnQo U2hhcmVkQnVmZmVyKiBmb250RGF0YSwgY29uc3QgU3RyaW5nJiBmb250TmFtZSkKICAgICBEV09S RCBudW1Gb250cyA9IDA7CiAgICAgSEFORExFIGZvbnRIYW5kbGUgPSBBZGRGb250TWVtUmVzb3Vy Y2VFeChyZXdyaXR0ZW5Gb250RGF0YS5kYXRhKCksIGZvbnREYXRhLT5zaXplKCkgKyBuYW1lVGFi bGVTaXplLCAwLCAmbnVtRm9udHMpOwogCi0gICAgaWYgKGZvbnRIYW5kbGUgJiYgbnVtRm9udHMg IT0gMSkgeworICAgIGlmIChmb250SGFuZGxlICYmIG51bUZvbnRzIDwgMSkgewogICAgICAgICBS ZW1vdmVGb250TWVtUmVzb3VyY2VFeChmb250SGFuZGxlKTsKICAgICAgICAgcmV0dXJuIDA7CiAg ICAgfQo=