On Wed, Aug 24, 2011 at 12:35 PM, Richard Wordingham <[email protected]> wrote:
> Expanding on Mark's answer, the basic difference is whether a character > of Bidi class ET (percentage-type and currency symbols) when stored > before or after European or Persian etc. digits goes to their left or > right. For Arabic, these symbols are not part of the number; for > Hebrew, they are part of the number. I'm not I understand the effect on > ES (variants of plus and minus signs) - may be it says that 'e-less' > exponents in Fortran output go on the left in Arabic but on the right > in Hebrew and LTR scripts. Aha! Now what about exponents in Algol using ⏨ - with its "Symbol; other" property and directionality "undefined" (despite my suggestions to make it Symbol;math and European number separator)? For "Hebrew text 31.415926⏨-1 Hebrew text" I get עברית טקסט 31.415926⏨-1 עברית טקסט (rendered as 31.415926 Hebrew text⏨-1 Hebrew text by Mozilla Thunderbird 3.1.12) and, similarly, النص باللغة العربية 31.415926⏨-1 النص باللغة العربية for "Arabic text 31.415926⏨-1 Arabic text". Not good. Leo
