Re: [sinhala-technical] collation: order of composed and decomposed dependent vowels

[Ruvan W. <ar...@uc...>]

great work harshula!


Harshula wrote:
> On Sun, 2008-10-12 at 17:10 +0530, Gihan Dias wrote: 
>   
>> Harshula wrote:
>>     
>>> Hi Gihan & Ruvan,
>>>
>>> How should the composed and decomposed forms of dependent vowels be
>>> sorted relative to each other? They can be considered equivalent, which
>>> seems to be the standard practice of Indic scripts.
>>>
>>>   
>>>       
>> Harshula,
>>
>> As far as SLS1134 is concerned, the decomposed forms are not considered 
>> to be valid, and therefore, should be either marked as incorrect, or 
>> converted to the canonically composed form.
>>
>> Therefore, they would not be considered in collation.
>>
>> My recommendation is:
>>
>> 1. Never generate such decomposed forms.
>>
>> 2. Mark any such forms as errors.
>>
>> 3. If you are having an application such as a spelling checker which 
>> corrects errors in text, then convert any such sequences to the composed 
>> form.
>>
>> I do *not* recommend any application treating decomposed forms as valid 
>> representations of Sinhala text.
>>     
>
> Gihan,
>
> Thanks for the info, what I have done is to *not* add decomposed forms
> in the collation sequence.
>
> We now have Sinhala collation in MySQL 6
> (http://dev.mysql.com/downloads/mysql/6.0.html), still alpha s/w, and
> will have it in GNU libc 2.8 when it is released.
>
> cya,
> #
>
>