Re: Question regarding bidirectional algorithm

From: Markus Scherer (
Date: Thu Jul 27 2000 - 13:56:26 EDT

> David Tooke wrote:
> The bidirectional algorithm mentions mirrored glyphs. The reference code handles them by replacing these characters with their mirror image. Is this the preferred method of doing this? If so, is there any where in the Unicode database that correlates the two characters?

i have proposed such a database file for consideration by the next utc meeting in august. you will find a preliminary version in icu at
there is an api in icu that provides this data: u_charMirror(). about icu, see

> Otherwise, do I render just render the glyph in reverse, flipping the glyph along the horizontal axis? Wouldn't this screw up italicised characters?

sometimes you can graphically mirror glyphs, but the best way to do this is to select a variant glyph. italics are an example of where graphical mirroring does not work (well).


This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:21:06 EDT