Re: PQconninfo function for libpq

On Thu, Nov 22, 2012 at 10:05 PM, Boszormenyi Zoltan <zb(at)cybertec(dot)at> wrote:
> 2012-11-22 12:44 keltezéssel, Magnus Hagander írta:
>>>>>>> Also, a question was buried in the other review which is - are we OK
>>>>>>> to remove the requiressl parameter. Both these patches do so, because
>>>>>>> the code becomes much simpler if we can do that. It has been
>>>>>>> deprecated since 7.2. Is it OK to remove it, or do we need to put
>>>>>>> back
>>>>>>> in the more complex code to deal with both?
>>>>>
>>>>> Just going to highlight that we're looking for at least one third
>>>>> party to comment on this :)
>>>>
>>>>
>>>> Yes, me too. A +1 for removing wouldn't count from me. ;-)

+1

> The second one is the product of what caught my attention while
> I was looking at pg_receivexlog. The current coding may write
> beyond the end of the allocated arrays and the password may
> overwrite a previously set keyword/value pair.

ISTM that such problem doesn't happen at all because argcount is
incremented as follows.

if (dbhost)
argcount++;
if (dbuser)
argcount++;
if (dbport)
argcount++;

Regards,

--
Fujii Masao