Hi,

Background
==========
Harsha Senanayake's Pango patch for adding Sinhala support was
submitted to Pango by Anuradha Ratnaweera in Bug 153517. The
aforementioned patch was mostly a simple copy-modify-paste of the
Malayalam/Tamil script support code, including worst case expansion
and script flags. Sinhala orthography is subtly different from both
the North Indian and South Indian orthographies.

1) North Indian Vs Sinhala
--------------------------
In North Indian orthographies a consonant cluster (Cons-Virama-Cons)
results in an implicit conjunct glyph therefore the ICU/Pango code denotes
this sequence of three codepoints as a 'matra'. This is not the case with
Sinhala where conjuncts must be *explicitly* created. In Sinhala a
sequence of Cons-Virama-Cons shall result in two 'matras', Cons-Virama
and Cons.

2) South Indian Vs Sinhala
--------------------------
Sinhala orthography has three ways of writing Cons-Virama-Cons - 2
separate glyphs, a conjunct, 2 touching letters. The last two ways
require the explicit use of ZWJ. My understanding is that South Indian
orthographies do not behave this way.


Issues resulting from (1) were fixed in Bug 161981 by introducing a
CC_AL_LAKUNA character class to Pango. However, CC_AL_LAKUNA was only
recently added to ICU (http://bugs.icu-project.org/trac/ticket/6232).
During the process, I realised SF_MPRE_FIXUP was inappropriate for
Sinhala because of (2).

In Pango, SF_MPRE_FIXUP in Sinhala script flags does not appear to have
a negative affect on layout, much like the incorrect worst case expansion
(Bug 385321) did not. However, technically it should never have been
included in the first place.

I'll attach a patch to remove SF_MPRE_FIXUP from the Sinhala script flag.

Regards,
Harshula