When Khmer consonants and Independent vowels are followed by another consonant
or independent vowel, the second one changes shape and is placed under
(sometimes under and before, sometimes under and after) the first consonant or
independent vowels. 

The change to subscript form is not automatic. To indicate that a consonant or
independent vowel must take a subscript for, Unicode uses the Character u+17D2
(preceding the consonants code-point).

u+1780 u+17d2 u+1785 indicates that consonant u+1785 is a subscript to consonant
u+180

The present code in Pango works correctly when a consonant is followed by
another consonant or independent vowel, but not when an independent vowel is
followed by a consonant or independent vowel, as should be the case of
u+17b2 u+17d2 u+1799, in which the shape of u+17d2 u+1799 should be completelly
different from the shape of u+1799 if the character u+17d2 was not present.

To solve this problem requires a little change in the Khmer module code.

In the present code we have:

 if (state != Khmer_Cons) {
    if (state != Khmer_IndV && state != Khmer_Other)
        *invalid = 1;
    goto finish;
   }

It needs to be changed to:

 if (state != Khmer_Cons && state != Khmer_IndV) {
    if (state != Khmer_Other)
        *invalid = 1;
    goto finish;
   }