From: Roozbeh Pournader (roozbeh@htpassport.com)
Date: Tue Aug 18 2009 - 13:36:56 CDT
On Tue, 2009-08-18 at 10:51 +0100, Julian Bradfield wrote:
> Is it really the case that Arabic and Hindi speakers *can't read*
> unshaped script?
Yes. Most of them can't. Disjoint Arabic script is very hard to read by
the average person (if we go beyond two or three words, of course). It's
*not* comparable to, say, reading Latin in ALL CAPS.
> What did Hindi and Arabic look like on 1970s
> computers with 9x15 bitmaps on green screens?
I can't tell you about Hindi, but I've used (and coded for) Arabic
script on terminals, and I think some of them were even 8x8 bitmaps. I
clearly remember the green terminals, but I think we used orange
terminals too. We still used them in the university I went to at least
until 1997.
First iteration of these had different characters for different
contextual shapes of the Arabic letters. Look, for example, at the
compatibility characters Arabic Presentation Forms-B block, U+FE70..U
+FEFF. But there was rarely four different characters per letter. They
usually unified some shapes, so you usually had only one or two shapes
for most of the letters. For example, you had U+FE8F and U+FE90 unified
with a glyph that could work for both (same with U+FE91 and U+FE92, or U
+FEA9 and U+FEAA). For some letters, like Heh and (Farsi) Yeh, you had
three shapes in some character sets, and for others, like Ain and Ghain,
some had four. Users typed in the explicit contextual shape (the first
Persian keyboard I uses had four Ain's on it). Sorting wasn't really
done properly (most people just sorted binary), but if you wanted proper
sorting, you wrote your own comparison routine.
Second iteration had an input method. So while the character set
remained glyph-based, users now only had one key per letter. Languages
like Persian also added a ZWNJ (U+200C) key combination. The input
methods were mostly application based, so they differed slightly when
you went to a different application.
Similar technology was used both on centralized computers (I used
VAX/VMS and Ultrix machines), and MS-DOS-running PCs.
One character per letter was of course played with, but at least in
Iran, it was considered something fancy and academic until the age of
Windows. In Iran, some people are still creating and selling software
that internally uses a glyph-based character set. They are so used to
the old mindset.
Roozbeh
This archive was generated by hypermail 2.1.5 : Tue Aug 18 2009 - 13:40:15 CDT