L2/16-115

Comments on L2/16-084 "Proposal to encode single dot sign at 1133B"
Shriramana Sharma, jamadagni-at-gmail-dot-com, India
2016-May-02

First, Ganesan's document L2/16-084 is not a properly formed proposal in so much
as it neither provides formal character properties nor a proposal summary form.

Ignoring that for the moment, one notes that Ganesan proposes a single-dot nukta
to be encoded in the Grantha block at U+1133B with the name GRANTHA SIGN VINDU.

The accepted rendering of the Sanskrit word for drop is BINDU and not VINDU.
Also, the word BINDU in the context of Sanskrit-supporting scripts in India is 
commonly recognized as an alternate name for the anusvara, which is of the shape
of a dot or small circle. In the scripts Ganesan refers to, though they are
certainly descended from Grantha, they are *outside* India and the BINDU-related
names apply to *punctuation* characters there. Thus such a name in a script used
within India would be misleading.

In any case, I argue that this character is not a valid candidate for encoding
in the Grantha block. With no offence meant to anyone I must respectfully submit
that the Grantha block (or any particular script's block for that matter) is a
not a dumping ground for characters which are for whatever reason outcast from
other blocks. Characters which belong in the Grantha block are those which are
attested in the Grantha script or which the Grantha user community newly desires
to use with other Grantha characters (and likewise for other scripts too).

It is on this latter basis that we native users of the Grantha script had
proposed the two-dot nukta for Grantha. We have no use for a contrastive
single-dot Grantha. Indeed, we had not proposed a single-dot nukta for Grantha
as it was our subjective judgment that a two-dot nukta better fits in with South
Indian scripts rather than a single-dot nukta. As such, the single-dot nukta
which is neither attested in Grantha nor is desired by the Grantha user
community is not a valid candidate for encoding in the Grantha block.

In fact, the character is attested in use with the Tamil script. The Tamil block
would be the ideal place for its encoding. One is hindered from doing so for
certain non-technical reasons. Thus one resorts to the more modern technology of
using Script Extensions. Insofar as other characters from the Devanagari block
have already been recommended for use with the Tamil script (among other Indic
scripts, see my earlier doc on this topic L2/16-058), there is no technical
problem with the single-dot nukta being "borrowed" from Devanagari as well.

The fact that Grantha is the closer related script to Devanagari would be
applicable as a deciding factor for usage recommendation if there already
existed a single-dot nukta in the Grantha block as it does in Devanagari. It
however cannot be treated as a sufficient ground for *encoding* a non-existent
single-dot Grantha nukta when there is already a similar Devanagari nukta for
the taking.

Ganesan's document quotes Ken's e-mail chapter and verse as if it were Gospel
not to be questioned. However, I seem to detect a certain amount of
tongue-in-cheek-ness in Ken's mail, but perhaps that's just me. Anyhow, despite
what Ken said, it does not seem valid to me to encode a character in the Grantha
block just because that character is not welcome (to some) in the Tamil block,
especially when the same requirement would be satisfied by a character already
existing in the Devanagari block.

Also note that using the existing Devanagari character would enable *with
just a few adjustments to fonts and the logic of text shaping engines* the
minority users to far earlier start using the single-dot nukta they prefer,
compared to having to wait for a future version of a standard for a character
with very little original content and *then* start bugging font/software vendors
to support their usage. Too much effort for a minority requirement, isn't it?

Thus I reiterate my earlier recommendation from L2/16-058 that 093C DEVANAGARI
SIGN NUKTA should be used for the minority Tamil script orthography requirement
and for this, Tamil should be added to the Script Extensions property of this
character.

-o-