Ayar Inconsistencies when typing Burglish

[GoogleCodeExporter]

Reported by Ma Nu: (Zawgyi encoding)
1) Problem typing pat-sint words.
2) Problem with typing "asat" (--္) by itself.
3) Problem typing inverted comma (`), hyphen (-), comma (,), full stop(။) 
---the only way to do it is by switching back to "English", which is 
non-optimal.
4) Request to add (--း) and (--့) as independent letters (like --ာ is 
now).
5) There's a question of typing dipthongs ("taung, kaung, pauk, out, pyaung" 
are given as examples)
6) Problems with ya-yit appearing (incorrectly) before tha way hto in some 
cases (notably: ျေခ)

Original Email, in Ayar Encoding:

ဘားဂလစ်အြေခခံြပီးရိုက်တာ 
အားလုံးနီးပါးရတယ်

မရေသးတာက

(၁)ပါတ်ဆင့်ေတွ ဥပမာ - မင်ဂလာပါ

(၂)အသတ်ေတွ ဥပမာ ဘားမိစ

(၃) (without changing into English version)  “-----”  (inverted coma 
)တုံးတို / အရှည် (hyphen/ dash) ပုတ်မ 
ပုတ်ထီး coma, full stop. 
ရိုက်လို့မရေသးဘူး

(၄)ြဖစ်နိုင်ရင် edit 
လုပ်လို့ရေအာင် ဝတ်စေပါက ( : ) , ( 
--့) , အသတ် (-တ်) 
သီးသန့်ထဲ့ေပးနိုင်မလား

(၅)Cannot type “diphthongs”  (e.g taung, kaung, pauk , out, pyaung)

(၆) order of ya-yit with tha-way-hto in some vowels   e.g-  ြေခ  

Original issue reported on code.google.com by seth.h...@gmail.com on 15 Jun 2010 at 3:22

[GoogleCodeExporter]

My comments:

1) You can type, e.g., ကမၻာ as "kam" "Bar". (Output: ကမၻာ). 
မငၤ can by typed "min" "NGa". There are two problems with this at the 
moment. First, the capitalized letters are not shown in caps (NGa is shown as 
nga, even though the correct output is produced). Second, the PI symbol (π) is 
shown in the temporary output window. This is unavoidable; most Unicode 
encodings have trouble displaying partial sequences. I will fix the first 
problem; for the second I'll try to convert to Zawgyi internally... after the 
first demo of Ayar4WaitZar.

2) One should be able to type, e.g., "ka" then "f" for asat. Currently, this 
produces "က--္". Oops; this is a bug. Will fix.

3) WaitZar usually allows symbols to filter through... I'm not exactly sure why 
Burglish4WaitZar doesn't. I'll look into this. On a related note, typing "," 
directly causes the system to crash! Definitely need to fix this. (Note: 
Burglish is messing up other input... time for some debugging.) Also, full stop 
should be part of Burglish; I'll check the spec. file one everything's fixed.

4) Will be addressed when I've stabilized the rest of the build. Changing the 
model isn't something I want to do to a system until all bugs are fixed.

5) "taung" and "paung" can be typed as "tg", "pg". ("pyg" too...) I'll have to 
ask Ko Soe Min about the full-length spelling.... I might have forgotten to 
expand it in my code. Not sure why "out", "pauk" are failing; I'll need to 
figure out why.

6) This is a problem with out Uni2Ayar filter. I'll fix it (should be easy) but 
the production WZ probably won't use Ayar for display, so...

Ok, 6 good bugs reported. Now time to fix them! :D

Original comment by seth.h...@gmail.com on 15 Jun 2010 at 3:42

• Changed state: InProgress

[GoogleCodeExporter]

Some bug fixes:

1) is fixed. (The display issue will have to wait for conversion.)
3)'s comma error is fixed; WaitZar wasn't appending the letter for some reason. 
There are two things left to fix: (a) the comma/period keys aren't passed 
through to Burglish; and (b) any other system keys aren't passed through. 
Originally, this was because Burglish couldn't tell if a key had had any effect 
on the model. I think we can re-enable system keys; will look at it later.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 7:13

[GoogleCodeExporter]

A note:

5) is somewhere in my code; "kaung" works fine on the Burglish web input.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 7:17

[GoogleCodeExporter]

A note:

5) occurs because "au" is missing as a suffix in WZ ---thus breaking the 
"chain" to "aun", "aung", etc.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 9:49

[GoogleCodeExporter]

5) Fixed. We were removing all those annoying u'...' strings in Python. So "au" 
and "ou" were affected. Did a manual search; these are the only ones. ("u" and 
"uu" were also affected, but I'd fixed those earlier.)

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 9:55

[GoogleCodeExporter]

2) Fixed. "-" is now stripped by Burglish, unless doing so would cause the 
string to be empty.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 10:09

[GoogleCodeExporter]

Note:

6) This occurs because Ayar insists that U+1031 appear (in encoding) before 
U+103C. Unicode enforces the other ordering. Should be easy to fix; I'd just 
assumed that U+1031 and U+103C only had to appear before the consonant, in any 
relative order.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 10:22

[GoogleCodeExporter]

6) is fixed. I also fixed the problem going from Ayar2Uni, just in case. 

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 10:35

[GoogleCodeExporter]

Note:

3) All "system" (non-A-z) keys are passed through to the basic key handler. So 
"." and "," would just go that way too. I won't change this until 1.9, since 
"," and "." already work. (However, I _should_ be able to fix ` and -).

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 1:59

[GoogleCodeExporter]

Note:

3) As a result of earlier changes (months ago), symbols are also broken in 
standard WZ. Whoops....

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 2:26

[GoogleCodeExporter]

3) Fixed. "," won't work, since it shares the key with "half stop". But all 
other keys work in Burglish & WaitZar. This code is a little shaky; it'd be 
well worth testing some more (and hopefully cleaning up in a future release.)

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 3:07

[GoogleCodeExporter]

Note:

4) "း" should be type-able with ";", but currently Burglish only recognizes 
alphanumeric characters. "--့" can be typed at any time with "h"; it works 
fine.

I need to scan for other non-alphabetic special letters. Will fix ";" after I 
do this.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 3:12

[GoogleCodeExporter]

";" is the only one, except for a few "," and "." characters.

Added a new bug for ","/".". (Issue 115) Otherwise, all possible fixes are 
implemented, so I'm closing this issue.

Original comment by seth.h...@gmail.com on 16 Jun 2010 at 3:21

• Changed state: Fixed