[HarfBuzz] Regression test for Khmer(0.9.5 VS 0.9.14)

[datao zhang]

Hi all: Test font: Khmer OS.ttf, please refer to attachment. After compared two version, The new version can't handle consonant register shift correctly when combine with zwnj: 1.
 
 
 

  ហ្វ‌៊ីតូអេសស្ត្រូចេន
  ha,coeng,vo,zwnj,triisap,_ii,ta,_uu,qa,_e,sa,sa,coeng,ta,coeng,ro,_uu,ca,_e,no

 

 
 
 

  ប៉ូលីហ្វ‌៊ីណូល
  ba,muusikatoan,_uu,lo,_ii,ha,coeng,vo,zwnj,triisap,_ii,nno,_uu,lo

 
 when zwnj is before triisap or muusikatoan, the shift should keep it as it should be, not changed it to vowel u. please refer to attachment. 2.
 
 
 

  អ៊‍ឹកធឹក
  qa,triisap,zwj,_y,ka,tho,_y,ka

 
I can't find any proof how to handle register shift when combine with zwj, maybe it should handled similar as zwnj. I have no idea about this. Maybe we can refer to uniscribe.  Br,Dean 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/harfbuzz/attachments/20130507/f814539d/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: KhmerOS.ttf
Type: application/octet-stream
Size: 265552 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/harfbuzz/attachments/20130507/f814539d/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: km_test.png
Type: image/png
Size: 15967 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/harfbuzz/attachments/20130507/f814539d/attachment.png>