Travis Vitek wrote:
>
>
> Travis Vitek wrote:
>> Martin Sebor wrote:
>>
>>> Travis Vitek wrote:
>>>
>>>> + || 22.LOCALE.CONS.MT.CPP || *1,+ ||
>>>> + || 22.LOCALE.CTYPE.CPP || *2 ||
>>>> + || 22.LOCALE.CTYPE.IS.CPP || *2 ||
>>>> + || 22.LOCALE.CTYPE.MT.CPP || *1,+ ||
>>>> + || 22.LOCALE.CTYPE.NARROW.CPP || *2 ||
>>>> + || 22.LOCALE.CTYPE.SCAN.CPP || *2 ||
>>>> + || 22.LOCALE.CTYPE.TOLOWER.CPP || *2 ||
>>>> + || 22.LOCALE.CTYPE.TOUPPER.CPP || *2 ||
>>> I thought the ctype tests were being run in all installed locales,
>>> just like the numpunct one? Which is what we want to move away from.
>>> IMO, exercising a small set (less than a dozen) of known locales and
>>> encodings should be plenty.
>>>
>> Yes, the non-mt ctype tests iterate over each locale for which
>> the function call `setlocale (LC_CTYPE, name)' succeeds. The mt
>> ctype tests all limit the number of tested locales to 32.
>>
>
> Any suggestions on which languages/countries/codesets that we should
> be testing against for the ctype tests?
I think we should cover a few Western locales and a few Asian ones.
For the first group, here are some candidates: one of each of en_US,
de_*, fr_*, es_*, in a mix of ISO-8859 and UTF-8. For the second
group, I'd consider one of each of ja_JP, ru_*, zh_* in EUC-JP,
Shift_JIS, KOI*, GB*, and UTF-8.
>
> Reducing the number of selected locales to 32 is pretty easy. Selecting
> which locales is a little more difficult.
You're telling me! :)
> Another issue is that the
> mechanism I have defined doesn't support selecting only one locale for
> each match.
So there's no way to ask for just one of locale out of the three
here: ja_JP.{EUC-JP,Shift_JIS,UTF-8} That may not be too much of
a problem unless each of the expansions matches multiple aliases
of the same locale. Will see how it goes as we come up with query
strings for each test.
Martin
|