[Python-Dev] Odd lines in unicodedata_db.h

Guido van Rossum guido at python.org
Mon Apr 5 01:03:20 CEST 2010


On Sun, Apr 4, 2010 at 10:43 AM, MRAB <python at mrabarnett.plus.com> wrote:
> Amaury Forgeot d'Arc wrote:
>>
>> 2010/4/4 MRAB <python at mrabarnett.plus.com>:
>>>
>>> I've just downloaded the daily snapshot at
>>> http://svn.python.org/snapshots/python.tar.bz2
>>>
>>> In the header file /python/Modules/unicodedata_db.h, there are the
>>> following lines in the change_records_3_2_0 struct:
>>>
>>>       { 255, 255, 255, 255, 1.0 },
>>>       { 255, 255, 255, 255, 2.0 },
>>>       { 255, 255, 255, 255, 3.0 },
>>>       { 255, 255, 255, 255, 4.0 },
>>>       ...
>>>       { 255, 255, 255, 255, 1e+16 },
>>>       { 255, 255, 255, 255, 1e+20 },
>>>
>>> Looks like a bug to me.
>>
>> I don't think so. Unicode 3.2 did contain two entries with large numeric
>> values.
>> The file Unihan-3.2.0.txt contains these two lines:
>>
>> U+4EAC  kPrimaryNumeric 10,000,000,000,000,000 ten quadrillion (American)
>> U+5793  kPrimaryNumeric 100,000,000,000,000,000,000 hundred quintillion
>> (American)
>>
>> For some reason newer versions of the unicode standard removed these
>> values.
>>
> It causes a type warning:
>
> warning C4244: 'initializing' : conversion from 'double' to 'const int',
> possible loss of data

The numeric_changed field isn't used anywhere in the code AFAIK, so I
don't see why we need to preserve them.

-- 
--Guido van Rossum (python.org/~guido)


More information about the Python-Dev mailing list