Another day, another 20 commits got us this far yesterday: - Fixed more Khmer stuff. Multiple signs in the syllable are positioned correctly now. There remain 525, ie. 0.175% failures.
- Tamil, after a few quick fixes, was the first script to reach 0 failures! - We then routed Tibetan to go through the default shaper (non-Indic), but turned 4 additional global features for it (abvs, blws, abvm, blwm), and that single change fixed all the failures. All 90% of them. So Tibetan is also at 0 failures now. The change was a net deletion of code even. - We then looked at Sinhala. There were 15% failures initially. After a few fixes for Al-Lakuna-specific behavior, ie. forming reph only if a joiner is present, and allowing rakars in weirdly encoded sequences (eg. Consonant,Matra,ZWJ,Al-Lakuna,ZWJ,Ra), Sinhala is now at 622 failures, ie. 0.229%. - In the process Devanagari failures went down from 560 to 552 (0.079%) - The following scripts we have never looked at, but here's their stats given current state of affairs: * Gurmukhi: 8.1% failures, * Malayalam: 3.3% failures, * Bengali: 2.5% failures, * Gujarati: 0.67% failures, So hopefully those are the ones we will be looking at today. That was it for yesterday. More to come tonight. Cheers, behdad _______________________________________________ HarfBuzz mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/harfbuzz
