Re: Latin ligatures and Unicode

From: Michael Everson (everson@egt.ie)
Date: Tue Dec 28 1999 - 08:19:40 EST


Ar 11:51 -0800 1999-12-27, scríobh John Jenkins:
>on 12/27/99 12:37 AM, Eberhard Pehlemann at e.pehlemann@gmx.de wrote:
>
>John Jenkins schrieb:
>3. There is a proposal to add a ligator character to specify the ligation
>of two Latin characters, but the proposal is controversial.
>
>In any event, your case wouldn't be appropriate for such a use -- you're
>dealing with a number of ligatures which are compulsory within a specific
>typeface.

Even in Fraktur, while ligation is compulsory, ligatures are not where
context-based. The rules are straightforward but complex. I am putting them
into my paper.

>Forcing the user to specify the formation of these ligatures
>whenever the characters are used -- and risking their showing up in other
>typefaces where they may not be appropriate -- isn't the best way to go.

You are mixing up inputting techniques and coding, John, aren't you? An
application may input ZWL automatically in certain cases (relieving the
user of the burden BUT ALSO GIVING HIM CONTROL), and ZWL is ignored if a
font doesn't happen to have the triplet it is associated with and so only
the base characters are displayed. Just like the SOFT HYPHEN, which can
pepper texts harmlessly unless needed.

>For situations where graphic alterations of the characters are required for
>proper display, the Unicode approach is to have the rendering software
>handle it.

That isn't true. Vide Devanagari and other Brahmic scripts, where conjuncts
are produced in coded text by the use of the virama. And please don't
retort that what the virama "really" does is kill an inherent vowel. The
naïve user sees the full syllable as a unit.

>I disagree! Blackletter ligatures cannot be handled just by the rendereing
>software!
>
>In the german language, written in a blackletter font, there are cases where
>only the meaning of a word decides if a ligature must or must not be used.
>(See below for an example.) But how shall the rendering software know about
>the meaning (the semantics) of my text?
>
>Fine. Then the user can manually override the default ligation behavior.
>Easy to do.

"Default" is defined simplistically in the AAT/OT model, and it doesn't
work for all European scripts. And when the user goes to all the trouble of
manual overriding, he should not have to lose all that work if he goes to
plain text.

ZWL solves the shortcomings in a way consistent with the Unicode model.

Michael Everson ** Everson Gunn Teoranta ** http://www.egt.ie
15 Port Chaeimhghein Íochtarach; Baile Átha Cliath 2; Éire/Ireland
Vox +353 1 478 2597 ** Fax +353 1 478 2597 ** Mob +353 86 807 9169
27 Páirc an Fhéithlinn; Baile an Bhóthair; Co. Átha Cliath; Éire



This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:57 EDT