Yasuo Ohgaki wrote:
> Hartmut Holzgraefe wrote:
> 
>> Yasuo Ohgaki wrote:
>>
>>> Hi,
>>>
>>> I've updated function name that has been changed for PHP 4.2.0.
>>> I need to know how to set appropriate version info to ref?
>>>
>>> I need to show
>>>
>>> pg_affected_rows
>>> (PHP3, PHP 4.0.0)
>>>
>>> instead of
>>>
>>> pg_affected_rows
>>> (PHP 4 CVS only)
>>
>>
>>
>>
>> the availability information is created by checking out *all*
>> previous releases from CVS and parsing the source code
>>
>> the function isn't known under its current name in previous
>> versions, so the output is absolutely correct, and no, you
>> *don't* want to change it
>>
>> we might talk about adding stuff like
>>
>>  pg_affected_rows
>>  (PHP 4 CVS only, known as pg_cmdtuples in PHP3, PHP 4.0.0 - 4.1.2)
>>
>> when i finally rewrite the function table generator that
>> also generates the version files for the manual, but printing
>>
>>  pg_affected_rows
>>  (PHP3, PHP 4.0.0)
>>
>> would be a ly, so no way this will ever happen
> MySQL functions have also renamed recently. They
> have fake version as you mentioned, I suppose.

which functions are you refering to?

> So there should be is inconsistency. I don't mind
> this inconsistency much, though.



> We need to some poilcy, since I'm going to rename
> remaining functions that do not confirm current
> naming standard sometime before 4.3.0.

the policy is easy: whatever the function table code
writes into version.dsl or version.xml is *fact*

> I hope I don't have to care about version, but
> 
> php_version
> (PHP 4 CVS only)
> 
> does not look good to me...

has there been a function or alias named php_version before?
definetly not

>  >  pg_affected_rows
>  >  (PHP 4 CVS only, known as pg_cmdtuples in PHP3, PHP 4.0.0 - 4.1.2)
> 
> If this is possible, it would be great.

the parser has to keep trak of functions and aliases over release
versions and keep track of old functions that become aliases to
newly introduced functions in later releases, and the output code
has to take care of this, so this is not a quick hack ...

-- 
Hartmut Holzgraefe  [EMAIL PROTECTED]  http://www.six.de/  +49-711-99091-77

Reply via email to