On Wed, 4 Apr 2001, Marco Cimarosti wrote:
> But all other characters in general category "Sc" (Currency Symbol) have
> bidi category "ET" (European Number Terminator), so I would suggest to
> avoid this exception.
>
> This makes sense, because currency symbols are normally associated with
> numbers (they normally occur before or after an amount), and bidi category
> "ET" ensures the correct behavior with numbers.
I noted that, but the point is that it's not a unique sign for
international use. It's only a compatiblity character, and it will be used
only in "legacy" Persian texts that are converted from ISIRI 3342 (which
has its own bidi, with only three categories: Left-to-right,
Right-to-left, and Persian Number).
It is a seqence of Arabic letters regarding bidi: If it is used after a
sequence of "EN"s (Persian numbers are considered to be in that category)
in an RTL context, it should appear to their left, and not right. Making
it "ET" will make it appear to the right. (Note that the example I'm
giving is almost always the case. Almost always, the user will press the
key only after a sequence of Persian digits in an RTL context.)
If this exception comes to be problematic, we prefer it to be in general
categort "Lo" (Letter, Other) rather than changing the bidi category to
"ET".
> 1) "ARABIC LIGATURE RIAL will normally be used in a RTL context".
> Also U+20AA (NEW SHEQEL SIGN, the Israeli currency symbol) is used in a RTL
> context, but it has bidi category "ET", not "R" (RTL Letter).
I noted that about U+20AA. The difference is "will normally be used" vs
"will always be used". New Sheqel as a currency sign in unambiguous, but
this symbol as a currency sign in international context will be both
ambigiuous and unreadable.
> 2) "ARABIC LIGATURE RIAL is decomposed by four letters of bidi
> category 'AL'". Also U+20A8 (RUPEE SIGN is decomposed as the two letters
> U+0052 U+0073 (i.e. "Rs"), but it has bidi category "ET", not "L" (LTR
> Letter).
The same points work here. (BTW, the Rupee sign seems to be the only
currency sign having a decomposition. It would be great to know about
the history of the character...)
> My other point has already been mentioned by others: the Currency Symbols
> block (U+20A0...) sounds like a more appropriate area.
Same points. And we want to hide it somewhere that people won't come to it
and implement it. :) And we want to ask everyone to convert it to "ZWNJ
Reh Farsi-Yeh Alef Lam ZWNJ" as soon as they could. Taking X keybaord, we
will ask the application to do the conversion.
> On the other hand, the main common feature of characters in the Arabic
> Presentation Forms blocks A and B is their being "presentation forms" (i.e.
> glyphs) for Arabic contextual forms and ligatures.
We believe this to be something like a ligature. A currency sign for Rial
never existed. It was only the word so common that people assigned a key
on typewriters for it. It may also be considered a presentation form,
because it's written narrower than the word Rial itself.
And then it made its way into the national standard character set. And now
we need it for round-trip. And we do want to discourage its use in all
ways (unless unavoidable), and we want to mark it as deprecated in all
ways.
(...Your message went really serious, thanks. I'm really out of energy
now... :)
I have updated the proposal with points others made. Does anyone believe
that we should write about the reasons for chosing the categories in the
proposal?
--roozbeh
This archive was generated by hypermail 2.1.2 : Fri Jul 06 2001 - 00:17:15 EDT