[EXTERNAL] Subscript Manual WA (was: Zawgyi Tonemarks in Latin Script)

[Richard Wordingham]

On Thu, 18 Feb 2021 19:36:37 +0000
Andrew Glass via Unicode <unicode at unicode.org> wrote:

The lack isn't where I thought it was - it turns out that the shaper
specification already supports the non-medial subscript WA!  I tweaked
the OpenType lookup in Padauk to to generate the ɡlyph for <U+1039,
U+101D> to check where the problem lay, but didn't realise that the
HarfBuzz test program hb-view would by default use the Graphite
shaping!  When I selected the OpenType renderinɡ, I got the correct
rendering from the tweaked font.

The problem is that *fonts* seem not to be including the subscript
WA, because it isn't required for *Modern Burmese*.  It so happens
that the major fonts' rendering of MEDIAL WA is suitable for
<VIRAMA, WA> - the pain of overlapping glyph ranges!
 
> Great question Richard, can you provide some examples? Do we have an
> agreed encoding mechanism for this?

I'll give a detailed answer, though the renderers already have the
solution.

The need for a distinction was put forward by Michael Everson at al. in
at least the following:

L2/06-029
L2/06-077 p2 (a.k.a. WG2 N3043)
L2/06-213


L2/06-077 p3 states, "Note that kwa with MEDIAL WA may take a teardrop
or triangular WA shape, which is never the case with true subjoined WA
(which is rare, though it occurs in Sanskrit)."

Martin Hosken put forward other arguments, but I'm not sure that they
were found convincing.

As to examples, just look at the absolutives in
https://www.alamy.com/stock-photo-burmese-writing-pali-canon-buddhist-canon-tripitaka-library-of-stone-21244784.html .
I'd been goinɡ to say look for -itvā for both Pali and Sanskrit, but
this forms seems commoner in word lists than actual text. 

TUS 13.0 Section 16.3 p647 says, "In Pali and Sanskrit texts written in
the Myanmar script, as well as in older orthographies of Burmese, the
consonants ya, ra, wa, and ha are sometimes rendered in subjoined form.
In those cases, U+1039 ္ myanmar sign virama and the regular form of
the consonant are used."

Thus, examples abound, and the encoding is defined.  The codechart
currently shows a teardrop shape for U+103D MYANMAR CONSONANT SIGN
MEDIAL WA - that would not be suitable for <VIRANA, WA>.

Richard.