Request: Option to load a model entirely from mywords.txt

[GoogleCodeExporter]

#Is this a feature request? If so, describe your feature here:
To allow for full customization (and to facilitate extending WaitZar to
minority scripts) there should be five options added to config.txt:

ignoremodel ==> only use mywords.txt to add words
allownonmmwords ==> allow any character in a myanmar word, even if it's
outside the 0x1000 to 0x109F range.
dictionarysize ==> set the dictionary explicitly to this value (0 == what
we do now)
nexussize ==> set the nexus size explicitly to this value
prefixsize ==> set the prefix size explicitly to this value

Test these with regards to performance: load the WZ model using only
mywords.txt.

Original issue reported on code.google.com by seth.h...@gmail.com on 12 Nov 2008 at 2:38

[GoogleCodeExporter]

Note: "ignoremodel" should disable the F1 help key.

Original comment by seth.h...@gmail.com on 9 Feb 2009 at 6:19

[GoogleCodeExporter]

ignoremodel, dictionarysize, nexussize, and prefixsize all work. (Still need to
implement allownonmmwords). More testing is needed, along with another option:

fontFile = default|x.png
  ...load a font from a file (default=myfont.png)

>Note: "ignoremodel" should disable the F1 help key.
Nope, actually, "allownonmmwords" should disable it. We can function perfectly 
fine
with a mywords-only model & the Help keyboard.

Original comment by seth.h...@gmail.com on 27 Mar 2009 at 4:03

• Changed state: InProgress
• Added labels: Priority-High
• Removed labels: Priority-Medium

[GoogleCodeExporter]

Done. Tested a Chinese mywords & font files, and it took forever to load & used
150MB. Reduced the size to 250 words in the very beginning of the character 
range,
and we got usable results.

charaset = any disables the help key, and doesn't perform a conversion from 
Zawgyi to
Unicode. (This probably won't cause an error, but let's be safe).

Font files will load cautiously, and fall back on the embedded file if they 
can't
find the proper data. Re-loading waitzar's fonts externally requires an 
additional
100K of memory.

Re-loading waitzar's model file from mywords.txt exclusively does not incur a
performance hit (but I'd like to try it on an older computer). It uses about 
800KB of
additional RAM. This leads me to suspect a memory leak.... but I've no idea 
where it
could be.

Marking fixed. 

Original comment by seth.h...@gmail.com on 31 Mar 2009 at 8:26

• Changed state: Fixed

[GoogleCodeExporter]

NOTE: This is most likely NOT a memory leak, but rather a result of 
over-estimating
the dictionary/nexus/prefix size when "0" is specified.

Original comment by seth.h...@gmail.com on 31 Mar 2009 at 8:31