Re: please expand re bidi algorithm

From: Michael \(michka\) Kaplan (michka@trigeminal.com)
Date: Tue Oct 03 2000 - 09:40:37 EDT


From: "Roozbeh Pournader" <roozbeh@sina.sharif.ac.ir>
> On Mon, 2 Oct 2000, Michael (michka) Kaplan wrote:
>
> > The slash character "/" is in the Unicode Bidi algorithm classified as
> > a "European Separator" which means that text would
> > be expected to be reversed (i.e. 1/2 would be expected to be 2/1).
However,
> > the the Farsi language (which uses the Arabic script) the slash is also
a
> > decimal separator for currency, thus 123.45 would appear as 123/45.
>
> Slash is the date separator in Farsi, and not the decimal separator.
> Persian has it's own decimal separator encoded in Unicode: U+066B.
> So, 1/2 in Farsi may mean second day of the first month, or one half of a
> unit, but surely not 1.2.

It depends. I was ONLY referring to a currency decimal separator, which is
*not* U+066B. The monetary decimal separator is indeed U+002F. This can be
retrieved from the NLS database with the LOCALE_SMONDECIMALSEP flag in a
call to the Win32 API GetLocaleInfo.

For what its worth, you are looking at the Arabic decimal separator, not the
Farsi one. They are not the same here. This can be retrieved by using the
LOCALE_SDECIMAL flag. For Farsi, this is indeed U+002E.

> > The end result is indeed what a native Farsi speaker
> > expects. I would find it unfortunate if this "bug" were "fixed" since it
> > would be done at the expense of Farsi users.
>
> No. This is not what a Farsi speaker expects. There are two kinds of Farsi
> speakers, professional users, and novice users. Professional users need
> compatiblity in different environments, and so they need the behaviour to
> be consistent. Novice ones are also not familiar with many other things,
> so they will become accustomed to it. Of course, things like the lack of
> Farsi decimal separator in Microsoft Farsi keyboard really annoys, and
> that may be the reason behind possible non-conformance in some Microsoft
> products.

Well, we are talking about two different things here. If you have Farsi
users who prefer Arabic language conventions, then they probably would be
annoyed, no wouldn't they? :-)

> BTW, I've been on mozilla-i18n mailing list which also addresses bidi, and
> I have not heard anything about this mis-feature.

Well, I reported it once, to Jonathan Rosenne, and he forwarded it to some
Mozilla developers, who probably concluded that this was the way it was
supposed to be (I never heard back from the developers).

You can see the Mozilla behavior by going to the following URL:

http://64.38.165.18/FarsiVsArabic.asp

This takes two Arabic locales and one Farsi locale, and shows how things are
formatted. Note that Mozilla does the right thing with Farsi, at least with
the milestone and platform I am using (Milestone 16, Windows 2000).

MichKa

random junk of dubious value at the
multilingual http://www.trigeminal.com/ and
a new book on internationalization in VB at
http://www.i18nWithVB.com/



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:14 EDT