Remaining Keyboards for version 1.8

[GoogleCodeExporter]

#Summary:
  - Only a little bit of work needed to fix most keyboards (and create Zawgyi keyboard.)
  - Parabaik & Win Innwa should be developed but their hotkeys will prevent them from getting into the 1.8 release.
  - In other words, keyboards are almost done~~~

#Current keyboard status:

Ayar 
  - The '-' key does not function as "English". Need to test other keys; most should be ok.
  - Need to update the auto-test file.

Myanmar3
  - Appears complete; need to run a full test.
  - Need to create an auto-test file and manually check.

Parabaik
  - Uses Ctrl & Alt modifiers. 
  - Probably won't work for 1.8; disable.

Zawgyi-One
  - Can modify Ayar-Zawgyi fairly easily.
  - There are two styles: Ctrl+Alt+Shift and Shift+~+letter. Which one does the community prefer?
  - Need to create an auto-test file and manually check.

Win Innwa
  - There's two versions: one with Alt+XXXX for extra letters, the other with Ctrl+Alt+Shift.
  - Should ask the community which one they prefer.
  - Probably won't work for 1.8 due to either combination containing Alt.

Original issue reported on code.google.com by seth.h...@gmail.com on 1 Jul 2010 at 9:52

• Blocking: #89

[GoogleCodeExporter]

Ayar:
  - '-' works
  - Updated the auto-test file. Added some new cases for '-' as well.
  - Considering DONE; will check with testers once we have enough for a new nightly.

Original comment by seth.h...@gmail.com on 15 Sep 2010 at 12:21

[GoogleCodeExporter]

myWin:
   Added test keyboard, manually checked each key too. 
   8 tests fail, but they represent keys not in WZ 1.7 anyway.
   myWin is ready for release. 

Original comment by seth.h...@gmail.com on 3 Nov 2010 at 10:33

[GoogleCodeExporter]

There are ~110 errors. About 100 are duplicates (` and ~ being traced 
separately), so that's ~60 errors. Basic categories are:

 * Some ~+KEY return key, KEY, or ~+key instead of null. E.g. "~?", "B". These are hard-coded into Zawgyi.kbd, probably to avoid the default key matching (since KeyMagic can never NOT match on a non-switched variable). We can either keep these (comment out the tests) or replace with => NULL. 

 * Direct contradictions. E.g., "~N". In this case, we need to contact both dr. Carbon & the Zawgyi keyboard people, or get our hands on the Zawgyi Keyman keyboard. 

 * Key changes. The "}" key was moved to "CAPS LOCK". We might move it back, since we currently can't track Caps Lock anyway. But I want to talk to dr. Carbon first. 

 * Weird encodings. E.g., "~E". The encoding produced is correct, but the re-ordering rules mess it up. This is a big problem; it will take the longest time to fix. 

Original comment by seth.h...@gmail.com on 3 Nov 2010 at 1:54

[GoogleCodeExporter]

Ok, here we go. The inconsistency is between:
   http://www.planet.com.mm/fonts/zawgyi%20myanmar%20unicode%20keyboard%20L.gif
...and:
   http://burmatodaymedia.com/zawgyi_keyboard_layout.htm

The first one is more recent; I based my keyboard on that. The second one is 
more "forgiving" with multiple mappings per letter; it seems that dr. Carbon 
based his keyboard on that. Both are "Planet" keyboards.  

So... this may end up being a judgment call. 

Original comment by seth.h...@gmail.com on 3 Nov 2010 at 2:24

[GoogleCodeExporter]

Decided to run with the older keyboard, since more users are likely to be 
familiar with it. 

  * Removed a few rules (`g => g) that I thought were confusing.

Current problem:
   < VK_KEY_A > => $ZWS + $twh
...takes precedence over:
   ('zg_key') + 'a' => U1008 

Is this a bug in our sorting rules? 

Original comment by seth.h...@gmail.com on 5 Nov 2010 at 8:41

[GoogleCodeExporter]

Note that Ayar is malfunctioning for some tests ("--" and "u-F", notably). So 
there may be a regression.

Here's the trace of "u`a". It seems that rules match in order, but "VK_KEY_A" 
is not "consumed" by the first matching of a.

User typed:  u
   $row2K[*] => $row2U[$1]
      ==>u1000

User typed:  u1000`
   < VK_CFLEX > => ('zg_key')
      ==>u1000
STOP: Single-letter ascii

User typed:  u1000a
   ('zg_key') + 'a' => U1008 
      ==>u1000u1008

   < VK_KEY_A > => $ZWS + $twh
      ==>u1000u200bu1031

Original comment by seth.h...@gmail.com on 10 Nov 2010 at 9:31

[GoogleCodeExporter]

Fixed this by deleting the vkey after using it once. 
Fixes Ayar tests.
Zawgyi now has slightly fewer errors. :P

See how helpful it is to log? :D

Original comment by seth.h...@gmail.com on 10 Nov 2010 at 9:42

[GoogleCodeExporter]

Zawgyi tests run to completion; simple keys work as planned. 

Next up: checking for reordering problems. 

Original comment by seth.h...@gmail.com on 10 Nov 2010 at 10:47

[GoogleCodeExporter]

Reordering is mostly ok. 
I'm confident enough in the Zawgyi layout to release a nightly. 
Consider Zawgyi done; further glitches will have their own bug reports. 

Original comment by seth.h...@gmail.com on 15 Nov 2010 at 6:23

[GoogleCodeExporter]

(Just to be clear, that means that only Myanmar3 remains to be tested for this 
bug to be closed. I'll make a test file eventually...)

Original comment by seth.h...@gmail.com on 15 Nov 2010 at 6:24

[GoogleCodeExporter]

Added automatic tests for Yunghkio; all passed.
Added automatic tests for Myanmar3; some fail (but are ok).

One main Myanmar3 bug remains: typing "ya-yit" before the consonant won't 
re-order properly.

Original comment by seth.h...@gmail.com on 23 Nov 2010 at 10:03

[GoogleCodeExporter]

Myanmar 3 actually has you type ya-yit after. 
All right, all keyboards are ready to go!

Release Candidate is on its way.

Original comment by seth.h...@gmail.com on 24 Nov 2010 at 5:26

• Changed state: Fixed