bug#20140: 24.4; M17n shaper output rejected

[Richard Wordingham <richard.wordingham@ntlworld.com>]

[-- Attachment #1: Type: text/plain, Size: 1546 bytes --]

On Wed, 25 Mar 2015 23:25:54 +0900
handa@gnu.org (K. Handa) wrote:

> Could you try these patches and test the usability of
> forward-char-intrusive and backward-char-intrusive?

The results weren't quite what I'd hoped for, but the results are
usable.  Thank you.

The text I principally tried the commands out on was Tai Tham text <LOW
PA, SIGN II, TONE-1, SAKOT, NA, OA BELOW, TONE-2>.  I used corrections
for the bugs that had been affecting its rendering.

It seems that the commands prevent shaping across the cursor, but do
not inhibit shaping within the former cluster.  I was only doing shaping
on complete orthographic syllables, so entering a cluster chiefly had
the effect of losing all positioning of marks and making the text
unreadable.  However, the behaviour may make a good teaching aid!

I then tried the command on Thai, and there the commands worked well.
I therefore added to LANA-OTF.flt, for marks not in complete syllables,
the command:

    ("(M)" [ (1 = ) ] ) ; For stepping through.

I attach the results of not stepping through (labelled 0), and
stepping through by 1 to 7 characters (labelled 1 to 7).  The result
is not so good at 3 steps - I think because the extra rendering
command does not handle SAKOT.  Aggressive use of dotted circles might
improve the display.  I don't know why, at the end, there is a delay in
TONE-2 rising to its proper height.

C-S-f is not a good key sequence for me.  C-S is one of my X-keyboard
switching combinations - I chose it for compatibility with the Xming X
server.

Richard.

[-- Attachment #2: emacs_entry.png --]
[-- Type: image/png, Size: 5776 bytes --]