Re: BUG #5304: psql using conninfo fails in connecting to the server

On Thu, Feb 4, 2010 at 1:26 PM, Joe Conway <mail(at)joeconway(dot)com> wrote:
> On 02/02/2010 10:23 PM, Tom Lane wrote:
>> Joe Conway <mail(at)joeconway(dot)com> writes:
>>> Should I also be looking to replace all (or most) other instances of
>>> PQsetdbLogin()?
>>
>> I think we at least wanted to fix pg_dump(all)/pg_restore.  Not sure if
>> the others are worth troubling over.
>
> OK, this one includes pg_dump(all)/pg_restore and common.c from
> bin/scripts (createdb, vacuumdb, etc). I still need to adjust the docs,
> but other than that any remaining complaints?

Thanks a lot!

Here is the small comments:

* expand_dbname is defined as a "bool" value in PQconnectdbParams()
and PQconnectStartParams(). But we should hide such a "bool" from
an user-visible API, and use an "int" instead?

* conninfo_array_parse() calls PQconninfoFree(str_options) as soon
as one "dbname" keyword is found. So if more than one "dbname"
keywords are unexpectedly specified in PQconnectdbParams(), the
str_options would be free()-ed doubly.

Regards,

--
Fujii Masao
NIPPON TELEGRAPH AND TELEPHONE CORPORATION
NTT Open Source Software Center