https://bz.apache.org/ooo/show_bug.cgi?id=4638

--- Comment #30 from JC Ahangama <[email protected]> ---
There are three objections to my suggestion that Write should follow the OT
Standard when treating Standard ligatures. 

The first objection is that the request is too trivial to bother about. The
next two objections are specifically directed against my solution for Indic
scripts. My solution eliminates complexities of Unicode Indic leveling the
languages with Western European languages tremendously helping to reduce
illiteracy in India. My solution does not violate any technical or linguistic
standard but upholds them.

I have demonstrated that Indic can be romanized without loss or distortion.
This fact is proven by simply displaying romanized Singhala in the native
script by using  an orthographic smart font. You test by phonetically typing
the text. 

The objections are as below.

Objection 1 -- Trivial issue

It is indeed trivial for those who see no difference between implementation of
Standard ligatures and no ligatures at all. This is the case with users of the
Latin script except some who think there are language level ligatures. What
they do not realize is that Write actually interferes with the display of
ligatures prematurely or unnecessarily. There is no requirement in the Standard
to prevent display of Standard ligatures provided by a font. They should show
by default. In the case of Desktop Publishing software, you provide options to
the author for selecting non-standard ligature types.


Objection 2 -- The language of the text cannot be discerned from codes

Although it sounds technical, no engineer would say it because they know
immediately that there is no definitive way to identify languages just from the
underlying code set because many languages share the same script, notoriously,
Latin.

The way to identify the language in the case of web pages is simple. The
language tag will tell. (e.g. lang="en-US"). In other cases, you compare
frequencies of character codes to known code frequency charts of languages and
increase the confidence level by identifying frequently occurring words.
(Google won't confuse Singhala and Indonesian / Icelandic if they do this).


Objection 3 -- Rendering large number of ligatures is math intensive and could
cause the computer to crash

Again, this is not a statement by an engineer. A font responds to a keystroke
by supplying the drawing coordinates and glyph drawing instructions. If the
glyph returned is a base character, add the Advance Width of the already
painted glyph to its origin of coordinates to get the origin for the new glyph.
If the new glyph is a ligature, the existing origin is the origin of the new
glyph as well. You might wonder if a font full of ligatures would paint the
text faster everything else being equal. (Mixed Singhala text with my font is
already 98% ligatures and shows no difference in performance. Mozilla in 2008
reported that the difference is 2% without saying which way).

=====================================
My plea is simply to get rid of code that handles OT features that I suspect
relies on the OS, and to reserve them for an advanced version of Write, may be
WriteAndMore made for desktop publishing. Trusting Uniscribe in Windows adds a
level of unknown which, in this case, is faulty.

-- 
You are receiving this mail because:
You are on the CC list for the issue.

Reply via email to