Front page | perl.perl5.porters |
Postings from February 2017
On 02/21/2017 03:15 PM, Leon Timmermans wrote:
> On Tue, Feb 21, 2017 at 6:51 PM, <pali@cpan.org> wrote:
>>
>> On Tuesday 21 February 2017 12:50:16 Sawyer X wrote:
>>> On 02/21/2017 01:07 AM, Zefram wrote:
>>>> Kent Fredric wrote:
>>>>> Though that said, I think we can find a clearer name than "is_upgraded"
>>>> I think that's the best possible name. The "upgraded"/"downgraded"
>>>> terminology for describing the flag state is the only one we have that
>>>> doesn't give a misleading impression of its significance. Hence my use
>>>> of that terminology in Scalar::String.
>>>
>>> What do you think about Leon's suggestion of is_multibyte?
>>
>> It is bad :-(
>>
>> Why? Because... first ask: What is utf8::encode() doing? It is
>> converting string to (UTF-8) octets, so into single byte string. So
>> people start blindly use construct (because it could make sense):
>>
>> if (utf8::is_multibyte($str)) { utf8::encode($str); }
>>
>> And we have there same again problem.
>>
>> I proposed name is_upgraded() because we already have functions
>> upgrade() and downgrade(). And it is fully correct to use construction:
>>
>> unless (utf8::is_upgraded($str)) { utf8::upgrade($str); }
>>
>> Similarly you could write:
>>
>> if (utf8::is_upgraded($str)) { utf8::downgrade($str); }
>>
>> In both cases it just change internal representation of $str which is
>> invisible in pure perl code. And such construction can be used without
>> any problem as they are correct. Second one just can die, so eval
>> handling needs to be used.
>>
>> If people start using those constructions there will be less broken
>> code.
>>
>> So for me usage of is_upgraded() fits better into already used names
>> upgrade() and downgrade().
>
> That is a good point, even if neither example is really sensible
> (upgrading or downgrading unconditionally achieves exactly the same;
> they're sensible like that).
>
> Part of me feels like this should have been very dehuffmanized (e.g.
> internal_representation_upgraded), but given there's already a short
> alias that may not be useful anymore.
>
> Leon
>
I would be most comfortable with is_utf8_upgraded() to disambiguate from
other types of upgraded. I realize this is somewhat redundant given the
package name.
Thread Previous
|
Thread Next