Re: Unicode patch

Moving this discussion to php-i18n list.

I guess a simple search and replace could do it except for the 
functions that do not use this macro and set the type directly instead, 
but there are few of them probably. I say we should apply the patch and 
then fix up the RETURN_STRING() macros and continue from there.

-Andrei

On Jan 10, 2006, at 9:43 PM, Dmitry Stogov wrote:

> the simplest way to make "old" function comatible with unicode - 
> substitute
> RETURN_STRING() by RETURN_RT_STRING().
>
> Thanks. Dmitry.
>
>> -----Original Message-----
>> From: Andrei Zmievski [mailto:[email protected]]
>> Sent: Wednesday, January 11, 2006 12:55 AM
>> To: Dmitry Stogov
>> Cc: Zeev Suraski; Rasmus Lerdorf; Andi Gutmans
>> Subject: Re: Unicode patch
>>
>>
>> Hi Dmitry,
>>
>> I will review it as soon as possible.
>>
>> The behavior you described is exactly what we talked about at the
>> meeting and it is also what my message to php-i18n list was about
>> (which no one replied to). I see solutions to this:
>>
>> 1) Don't do anything. The breakage should be a sufficiently high
>> motivator for people to start upgrading the functions.
>> 2) Implement a stop-gap measure, while functions are being upgraded.
>> Basically, make the engine auto-convert the return value of the
>> function to IS_UNICODE. We'd need a way to indicate that the return
>> value is a text value to avoid converting functions that are supposed
>> to return IS_BINARY.
>>
>> -Andrei
>>
>> On Jan 9, 2006, at 3:53 AM, Dmitry Stogov wrote:
>>
>>> Hi Andrei,
>>>
>>> I made a patch that removes IS_BINARY data type, but I
>> don't think, it
>>> will
>>> give a lot of advantages in maintainability
>>> Now IS_STRING behave as IS_BINARY in unicode mode.
>>> So IS_STRING and IS_UNICODE are incompatible and all old functions
>>> must be
>>> upgraded.
>>> For example the following code causes a fatal error in unicode mode.
>>>
>>> echo "hello".sprintf("%s","world");
>>>
>>> Note: I didn't remove old behavior from zend_hash.
>>> In unicode mode string keys are converted to unicode ones automatic.
>>> (This
>>> should be removed on next step).
>>>
>>> Please review the patch and make your notes about it.
>>>
>>> Thanks. Dmitry.
>>> <unicode-6.diff.gz>
>>
>>
>>
>