Re: [counter-styles] i18n-ISSUE-282: Problems with Ethiopic-numeric algorithm

On 30/10/2013 01:24, Tab Atkins Jr. wrote:
>>      - In the Nov 2002 version of the Lists module, step 5 did not have the
>> >text "that did not have its digits removed in the third step". Addition of
>> >that text causes the algorithm to fail to produce the simple example of 100.
>> >Removal of that text seems appropriate.
> After I accept your correction of the third example, it looks like it
> should actually say "that does not have the value zero".  Those groups
> still don't generate group markers; it's just the ones with value 1
> that still do, and that the spec was incorrectly cutting out.
>
> This also reflects the instructions athttp://www.geez.org/Numerals/.

Yep.

I have fixed and updated my implementation at 
http://rishida.net/tools/numberconverter/ and tested successfully it 
against almost all the test cases at 
http://www.geez.org/Numerals/NumberSamples.html

One thing that's still problematic with the algorithm is that it doesn't 
produce a value for 1, because of "if the group is the most significant 
one and has the value 1" in step 3. I think we need to say "The 
following algorithm converts decimal digits higher than 1 to ethiopic 
numbers" or something like it (since testing for the value 1 at the 
start seems to be the most efficient approach).

Nit: in step 5 "exceptgroups with a value of zero" (note the typo) might 
be better as "except groups which originally had a value of zero", since 
the values have actually been removed by this point (in step 3).

RI

Received on Thursday, 31 October 2013 10:31:09 UTC