[XeTeX] Kerning in spite of a ZWNJ(was Kerning in spite of a NWNJ)

[Adam Baker]

Yes, I certainly agree that that would look bad. I wasn't thinking of
adding anything, but rather of pulling two glyphs closer together. For
instance, suppose I have a word پند (peh-noon-dal), which I want to
follow with the isolated form of a glyph, which is part of the same
word. I want to be able to control the space between the dal and the
isolated glyph. If the following glyph can be گ (gaf) or ر (re), it's
going to be a very different kerning situation.

Adam

On Tue, Sep 15, 2009 at 7:46 AM,  <xetex-request at tug.org> wrote:
> Message: 3
> Date: Tue, 15 Sep 2009 00:37:51 +0430
> From: Adam Baker <adamb924 at gmail.com>
> To: xetex at tug.org
> Subject: [XeTeX] Kerning in spite of a NWNJ
> Message-ID:
>        <c2c3a0990909141307l2a2d74b2vd2fd0429e337f78e at mail.gmail.com>
> Content-Type: text/plain; charset=ISO-8859-1
>
> Although I hope that XeTeX can help me address this question, it might
> be more of a Unicode question. I was using XeTeX when I came up with
> it. :-)
>
> I just found that when a zero-width non-joiner is placed between two
> Arabic letters, kerning rules are not applied. This strikes me as a
> bit odd, because in some languages (like Persian) a ZWNJ is used to
> force a final form, i.e., to indicate a morpheme break. It's
> conceivable that a following glyph would be an isolated form. In that
> case, you might need kerning to make the [final]-[isolated] sequence
> look better. Is there a way to produce this? (And alternative to ZWNJ
> perhaps?)
>
> Thanks,
> Adam
>
>
> ------------------------------
>
> Message: 4
> Date: Tue, 15 Sep 2009 09:13:40 +1000
> From: Vafa Khalighi <vafa at users.berlios.de>
> To: xetex at tug.org
> Subject: [XeTeX] Kerning in spite of a ZWNJ(was Kerning in spite of a
>        NWNJ)
> Message-ID:
>        <605202f20909141613k5c252362g3fcb7d1c011528ad at mail.gmail.com>
> Content-Type: text/plain; charset="utf-8"
>
>>
>> I just found that when a zero-width non-joiner is placed between two
>>
> Arabic letters, kerning rules are not applied. This strikes me as a
>>
> bit odd, because in some languages (like Persian) a ZWNJ is used to
>>
> force a final form, i.e., to indicate a morpheme break. It's
>>
> conceivable that a following glyph would be an isolated form. In that
>>
> case, you might need kerning to make the [final]-[isolated] sequence
>>
> look better. Is there a way to produce this? (And alternative to ZWNJ
>>
> perhaps?)
>>
>
>
>
> No, I am Persian myself and that is not true. If the rule were to be
> visible between ZWNJ, then it would look absolutely ugly. Imagine the
> usual word ?????? would look like ??|???. Actually if XeTeX does that,
> then it will become a XeTeX bug,
> that rule should be invisible (inserted but invisible). If I recall
> correctly it was visible in the past which clearly was a bug and
> Jonathan fixed it. That discussion can be fount at
> http://tug.org/mailman/htdig/xetex/2008-April/009520.html
>
> --
> Vafa