74310 2011-12-12 10:56:04 -0800 Add a layout test for 'init' feature application in Bengali (and other Indic scripts) 2012-11-29 04:39:55 -0800 1 1 1 Unclassified WebKit Layout and Rendering 528+ (Nightly build) Unspecified All NEW http://i18nl10n.com/fonts/beng3.html P2 Normal --- 1 jshin jshin ap bashi behdad tony oldest_to_newest 519346 0 jshin 2011-12-12 10:56:04 -0800 Chromium on Linux had a bug (due to a bug in harfbuzz-old) in the application of 'init' feature in Indic shaper. 'init' feature in Bengali (and perhaps in other Indic scripts) has to be applied only at the beginning of a word, but it was applied even in the middle of a word. It turned out that Harfbuzz-old had already fixed it in the upstream. Chromium's copy of harfbuzz is updated. While doing so, we'd better add a layout test to prevent a future regression (especially considering that we're switching to harfbuzz-ng before long). The Chromium CL fixing this issue is at http://codereview.chromium.org/8895014/ 758742 1 jshin 2012-11-05 11:34:04 -0800 Hmm.. it looks like this is regressed in an opposite direction after switch to harfbuzz-ng in Chromium Linux. 'Init' is not applied in both cases (whether it's the beginning of a word or not). It's my bad that I didn't add a test soon enough. For that regression, I'll file a separate bug after confirming that it's indeed regressed instead of a font issue (not having 'init'). 777461 2 bashi 2012-11-27 19:12:29 -0800 Behdad, do you have any ideas about this issue? 778914 3 behdad 2012-11-29 04:39:55 -0800 'init' is supposed to work, and I've definitely tested it. Attach test case / font please. Or file a separate bug for the regression and CC me.