[HarfBuzz] Clustering and Hit Detection

[Richard Wordingham]

On Sat, 6 Apr 2013 16:53:19 +0100
Richard Wordingham <richard.wordingham at ntlworld.com> wrote:

I prepared this this morning before Behdad replied, but I'll send it
regardless as a record of the situation.  I originally thought Behdad
had replied to this post.  I have some thoughts on how to fix the
issue, but I'll have to do some more work to learn that it isn't as
simple as it might seem.

> I don't know whether Thai justification should occur with Tai
> Tham...

It occurs in Tai Khuen, at least as printed by Wat Tha Kradas in
Chiangmai, Thailand.  The text string <HIGH CA, E, MAI SAT, TONE-2, AA>,
which should (and does when the script is specified as Tai Tham) reorder
as E, HIGH CA, MAI SAT, TONE-2, AA is justified in print as three
clusters:

1) E
2) HIGH CA, A, TONE-2
3) AA 

HarfBuzz reports it as a single cluster.

Indeed it looks as though all spacing marks are having their
separations increased, even those composed as SAKOT + consonant.  That
might make thing difficult for justification to work perfectly with all
fonts, for:

a) in some environments, the dependent vowels U and UU have
significant advance widths in Tai Khuen, as in Burmese; and

b) even within Northern Thai styles, there is variation in whether
some subjoined consonants have an ascender.

However, the most significant effects are those relating to the
preposed vowels, U+1A6E to U+1A72 and the trailing vowels U+1A61,
U+1A63 and U+1A64.

Richard.