2013/1/2 Heikki Linnakangas <hlinnakangas(at)vmware(dot)com>

> On 02.01.2013 17:27, Marko Kreen wrote:
>
>> On Wed, Jan 2, 2013 at 5:11 PM, Boszormenyi Zoltan<zb(at)cybertec(dot)at>
>>  wrote:
>>
>>> 2012-12-11 16:09 keltezéssel, Simon Riggs írta:
>>>
>>>  On 11 December 2012 12:18, Boszormenyi Zoltan<zb(at)cybertec(dot)at>  wrote:
>>>>
>>>>  Such mechanism already exist - you just need to set
>>>>>>> your PGresult pointer to NULL after each PQclear().
>>>>>>>
>>>>>>
>>>>>> So why doesn't PQclear() do that?
>>>>>>
>>>>>
>>>>>
>>>>> Because then PQclear() would need a ** not a *. Do you want its
>>>>> interface changed for 9.3 and break compatibility with previous
>>>>> versions?
>>>>>
>>>>
>>>> No, but we should introduce a new public API call that is safer,
>>>> otherwise we get people continually re-inventing new private APIs that
>>>> Do the Right Thing, as the two other respondents have shown.
>>>>
>>>>
>>> How about these macros?
>>>
>>
>> * Use do { } while (0) around the macros to get proper statement
>> behaviour.
>> * The if() is not needed, both PQclear and PQfinish do it internally.
>> * Docs
>>
>> Should the names show somehow that they are macros?
>> Or is it enough that it's mentioned in documentation?
>>
>
> IMHO this doesn't belong into libpq, the interface is fine as it is. It's
> the caller's responsibility to set the pointer to NULL after PQclear(),
> same as it's the caller's responsibility to set a pointer to NULL after
> calling free(), or to set the fd variable to -1 after calling close(fd).
> There's plenty of precedence for this pattern, and it shouldn't surprise
> any programmer.

True. +1

-- 
// Dmitriy.

In response to

• Re: allowing multiple PQclear() calls at 2013-01-02 15:52:36 from Heikki Linnakangas

pgsql-hackers by date

Next:	From: Kohei KaiGai	Date: 2013-01-02 16:35:13
	Subject: Re: Review of Row Level Security
Previous:	From: Alvaro Herrera	Date: 2013-01-02 16:28:19
	Subject: Re: fix bgworkers in EXEC_BACKEND