Rick said:
> We also learn from the "bird stamps" web site cited later that the
> government of Ghana is extremely inconsistent about their images and usage
> of their own currency sign. I.e., they apparently don't have a standard for
> it.
>
> So, I don't know... is this "cedi sign" a unique beast, or is it just some
> set of variations on the ordinary "cent sign"? Since the government of
> Ghana can't agree on a single representation (or even two representations
> in their collection of stamps, I would tend to think that that what we have
> is just a set of variations on the ordinary "cent sign", and any number of
> variant glyphs can be used.
>
> But if someone can point to a Ghana government standard, decree, design
> recommendation, or similar official document, then we could talk about
> encoding something new.
I draw a somewhat different conclusion.
I think the evidence of the bird stamps shows that the cedi sign has a significant typographical history in Ghana, showing similar variation to the storied history of the dollar sign. It is also clear that the cent sign and just a capital 'C' are common fallbacks for it, even in official documents. But that would make sense, because the cent sign and capital 'C' are easily available, both in fonts and in character sets.
The evidence presented in the stamps is at least as good as the evidence we used to encode the guarani sign and the austral sign, and I don't see a good case here for unification with U+20A1 COLON SIGN.
The cedi sign is the kind of decorated Latin letter symbol that we fairly regularly assent to for encoding as a currency sign, and it makes sense to me to go ahead with encoding it, if somebody feels strongly enough about it to put the relevant proposal together. Documentation for it should indicate the relevant information about the range of glyphic variation, as shown in the bird stamps.
Something that people have not mentioned so far is that one of the reasons that currency signs are separately encoded as symbols is that they have a distinct property: gc=Sc. Graphic constructions involving ordinary letters with combining overlays might appear acceptable, but would end up with the wrong properties.
In case of the Cedi sign / Cent sign duality, another issue arises. The line breaking classes for these would be different PR vs PO.
The alternative would be to change Cent sign to some more ambiguous linebreak class such as IS, but that's not wholly satisfactory.
It might be worth noting this in the proposal.
(For definitions see http://www.unicode.org/reports/tr14#PR etc.)
The potential drawback, as already pointed out, for encoding a cedi sign is that the already established fallback practice of simply using a cent sign (or a 'C') for the cedi will have a lot of inertia. Adding a cedi sign now, which might be available in earliest implementation a couple years from now, and then take a considerable amount of time before such implementations could be widely available in Ghana, would mean a very long period of transition where both the fallbacks and the new cedi sign would be in use. This might *cause* more problems than it would resolve.
--Ken
