From: Asmus Freytag (email@example.com)
Date: Tue Nov 04 2008 - 11:54:07 CST
On 11/4/2008 9:21 AM, John Hudson wrote:
> Kent Karlsson wrote:
>> I'm not sure what you mean by "character level mirroring" vs. "glyph
>> mirroring" here.
> By 'character level mirroring' I mean application of
> BidiMirroring.txt. This, as you and Asmus note, is intended as a
> fallback mechanism in the absence of glyph mirroring, but I don't
> believe it is reliably anticipated as such. There are already
> applications that perform this kind of 'character level mirroring',
> and they have done so in the absence of any standardised mechanism for
> glyph level -- i.e. font-controlled -- mirroring. This means that what
> was intended as a fallback mechanism may be active regardless of the
> presence or absence of glyph mirroring in newer fonts, so the latter
> must be aware of this and not reverse mirroring that has already been
> applied at the character level.
Right - if applications do stupid things, fonts must not make things
>> When the bidi algorithm says not to mirror, the font is *not* to do
>> mirroring (based on any bidi results). Otherwise there is something
>> wrong. That is why I was talking about what the bidi algorithm says
>> to mirror, and not to mirror, as well as the (ugly) loophole for LTR
>> overridden to RTL (but, currently, no such loophole for RTL
>> overridden as LTR).
> This presumes that Unicode is the last word on glyph directionality,
> but I don't think it is. As Chris Fynn notes, at the glyph level the
> decision to flip or not flip the form of a letter may be a stylistic
> variation, even a user preference as catered for by the individual font.
As written, the standard *does* claim to be the last word ;-) at least
as far as conformant implementations are concerned. Perhaps the kind of
fancy glyph variation, as in Toys 'R' Us would need to be added to HL6
in UAX#9 to make clear what's intended.
>> This would be something handled by the font handling system (mirror
>> the given glyph),
> That presumes that simply flipping the outline produces visually
> acceptable results. But type designers mean by mirroring involves
> design of directionally reversed glyphs, which is seldom achieved by
> simply flipping an outline. Consider this image
> contrasting a flipped letter e with one designed as a reversed form
> (for phonetic transcription in this case, but the same holds true for
> direction mirroring). The flipped form looks distorted because the
> ductus is reversed.
While we can all agree that the geometric mirroring of an outline is not
always applicable, your example is misleading as the "mirrored" glyph
here is for the *same* reading direction as the original glyph. That's
precisely not the case in bidi mirroring.
> Such designed mirrored forms need to be accessed, which implies a GSUB
> feature in OT or similar mechanism in AAT or Graphite.
No argument here - I think that Kent's use of "font handling system" was
meant more generically than you construed it.
> John Hudson
This archive was generated by hypermail 2.1.5 : Tue Nov 04 2008 - 11:55:42 CST