W3C home > Mailing lists > Public > www-math@w3.org > February 2013

Re: Mirroring Unicode symbols in Arabian

From: Khaled Hosny <khaledhosny@eglug.org>
Date: Wed, 20 Feb 2013 19:33:34 +0200
To: Murray Sargent <murrays@exchange.microsoft.com>
Cc: Neil Soiffer <NeilS@dessci.com>, Kent Karlsson <kent.karlsson14@telia.com>, David Carlisle <davidc@nag.co.uk>, Daniel Marques <dani@wiris.com>, "www-math@w3.org" <www-math@w3.org>
Message-ID: <20130220173334.GA17365@khaled-laptop>
Right, that is is more or less the what is OpenType specification is
saying, and indeed XITS Math should be fixed to follow the specification
(it was done that way to simplify testing since the system I was using
wasn't applying the Bidi algorithm).
http://www.microsoft.com/typography/otspec/TTOCHAP1.htm#ltrrtl

On Wed, Feb 20, 2013 at 04:35:54PM +0000, Murray Sargent wrote:
> I think that mirrored characters than don't have a mirrored character
> counterpart should be accessed by the 'rtlm' OpenType feature. But
> those that do have a mirrored character counterpart should not use the
> 'rtlm' feature since else they'll get remirrored when shaped. This is
> the approach I used in my RTL math zone prototype for MathML 3. I have
> a special copy of Cambria Math with 'rtlm' features for various
> operators such as integrals and summations. The approach works well.
> The font isn't complete enough though for more than just prototyping.
> 
> Murray
> 
> Khaled Ghetas noted:
> 
> XITS Math fonts has mirrored math characters, but they are accessible
> only with 'rtlm' OpenType feature (but it is possible to build a
> special version of it with the RTL glyphs as the default).
> https://github.com/khaledhosny/xits-math
Received on Wednesday, 20 February 2013 17:34:12 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 20 February 2013 17:34:13 GMT