[HarfBuzz] Regression test Bengali (0.9.5 VS 0.9.14)

[datao zhang]

Hi :
Sorry to reply so late because I need check the font license. Unfortunately, I can't send the font to all because license issue. I also try to find some similar free font, but I can't find (the font used to test based on new tag bng2). 
I start to dig the issue through code now, the first case I list here is clear. For 0.9.5 version, the code look like below (hb-ot-shape-complex-indic.cc, function:
initial_reordering_consonant_syllable()):
...    /* Reph */
    for (unsigned int i = start; i < end && info[i].indic_position() == POS_RA_TO_BECOME_REPH; i++)
      info[i].mask |= indic_plan->mask_array[RPHF];
    /* Pre-base */
    mask = indic_plan->mask_array[HALF];
    for (unsigned int i = start; i < base; i++) {
        if (info[i].indic_category() == OT_Ra &&
            info[i + 1].indic_category() == OT_H && i > start) {
          info[i].mask      &= ~mask;
          info[i + 1].mask  &= ~mask;
          info[i].mask      |= indic_plan->mask_array[BLWF];
          info[i + 1].mask  |= indic_plan->mask_array[BLWF];
        }
      else
          info[i].mask  |= mask;
    }.... 0.9.14 version:    /* Pre-base */
    mask = indic_plan->mask_array[HALF];
    for (unsigned int i = start; i < base; i++)
      info[i].mask  |= mask;
    /* Base */
    mask = 0;
    if (base < end)
      info[base].mask |= mask;
    /* Post-base */
    mask = indic_plan->mask_array[BLWF] | indic_plan->mask_array[ABVF] | indic_plan->mask_array[PSTF];
    for (unsigned int i = base + 1; i < end; i++)
      info[i].mask  |= mask; From the code, we can see blwf feature never applied those consonant before base character in new version. According to MS spec: http://www.microsoft.com/typography/OpenTypeDev/bengali/intro.htm, blwf feature should be applied before half feature. so that is the reason.  I will continue to dig more for other cases. Br,Dean----------------------------------------
> Date: Mon, 6 May 2013 10:14:42 +0100
> From: jfkthame at googlemail.com
> To: dataozhang at hotmail.com
> CC: harfbuzz at lists.freedesktop.org
> Subject: Re: [HarfBuzz] Regression test Bengali (0.9.5 VS 0.9.14)
>
> On 6/5/13 09:58, datao zhang wrote:> Hi all:
> >
> > After compare harfbuzz output (0.9.5 and 0.9.14), I find 5 main
> > differences. It seems that some change are not expected. Please argue
> > if you have different opinion.
>
> Please state exactly what font you're using, so that others can test and
> reproduce your results. Thanks.
>
> >
> > 1. স্ব্য়ং sa,virama,ba,virama,ya,nukta,anusvara
> >
> > In 0.9.14 version, (sa,virama,ba) not form ligature. please refer
> > to attachement 1.png
> >
> > 2. রেকর্ডিং‌এর ra,_e,ka,ra,virama,dda,_i,anusvara,zwnj,e,ra
> >
> > In 0.9.14 version, the top of (_i) is too long, it should align
> > with vertical stem of dda . please refer to attachement 2.png
> >
> > 3. আশ্র্য় aa,sha,virama,ra,virama,ya,nukta
> >
> > In 0.9.14 version, (sha,virama,ra) not form ligature, please refer
> > to attachment 3.png
> > 4. ক্রমান্ব্য়ে ka,virama,ra,ma,_aa,na,virama,ba,virama,ya,nukta,_e
> >
> > In 0.9.14 version, (na,virama,ba) not form ligature and the matra
> > (_e) should be before (ya,nukta), after (virama). please refer to
> > attachment 4.png
> >
> > 5. প্র্য়োজন। pa,virama,ra,virama,ya,nukta,_o,ja,na,danda
> >
> > In 0.9.14 version, (pa, virama,ra) not form ligature. please refer
> > to attachment 5.png
> >
> >
> >
> >
> >
> > Br,
> > Dean
> >
> >
> > _______________________________________________
> > HarfBuzz mailing list
> > HarfBuzz at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/harfbuzz
> >
> 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/harfbuzz/attachments/20130508/bb74d6b5/attachment.html>