On 10/17/2011 10:08 AM, Eli Zaretskii wrote:
Date: Mon, 17 Oct 2011 08:42:02 +0200
From: Simon Montagu<[email protected]>
List-software: Ecartis version 1.0.0
On 10/15/2011 05:19 PM, Andreas Prilop wrote:
I return to
http://www.unicode.org/mail-arch/unicode-ml/y2011-m10/att-0059/1999-12-31.html
Microsoft programs (Internet Explorer, MS Word), display this as
31/12/1999
Other programs (Firefox, Opera, OpenOffice) display this as
1999/12/31
Given that the bidi category of U+002F is CS, 1999/12/31 seems to be the
correct rendering following rule W4 of the Bidi Algorithm: "A single
common separator between two numbers of the same type changes to that type".
Note that this changed in version 4.0.1 of Unicode -- before that the
category of U+002F was ES, which would make the correct rendering
31/12/1999 as in Microsoft programs.
Actually, it's the other way around: U+002F being a CS, the correct
display is 31/12/1999, while if it where an ES, the correct display
would have been 1999/12/31. I verified this with the Reference
Implementation (where, confusingly, `/' is ES, so you need to use
something else, like a comma).
Btw, according to my testing, the current Firefox displays this as
31/12/1999. I don't have Opera or OO to check there, but it's
possible that the OP was using old versions of these that were still
using the old Unicode data base.
Sorry, I think I confused you here. I quoted the expected rendering in
ASCII digits, but the testcase in fact uses Arabic-Indic digits.
