On Fri, Aug 8, 2014 at 1:19 PM, Amit Kapila <amit.kapila16@gmail.com> wrote:
> On Thu, Aug 7, 2014 at 12:36 PM, Fujii Masao <masao.fujii@gmail.com> wrote:
>> On Thu, Aug 7, 2014 at 12:36 PM, Amit Kapila <amit.kapila16@gmail.com>
>> wrote:
>> > On Wed, Aug 6, 2014 at 1:32 PM, Fujii Masao <masao.fujii@gmail.com>
>> > wrote:
>
>> >> What about picking up only data_directory at the first pass?
>> >
>> > I think that will workout and for that I think we might need to add
>> > few checks during ProcessConfigFile. Do you want to address it
>> > during parsing of config file or during setting of values?
>>
>> I prefer "during paring". The attached patch does that. In this patch,
>> the first call of ProcessConfigFile() picks up only data_directory. One
>> concern of this patch is that the logic is a bit ugly.
>
> I think handling 'data_directory' in ParseConfigFp() looks bit out of
> place as this API is used to parse other config files as well like
> recovery.conf. I agree that for all other paths DataDir might be
> already set due to which this new path will never be executed, still
> from code maintenance point of view it would have been better if we
> can find a way to handle it in a place where we are processing
> the server's main config file (postgresql.conf).
Yep, right. ParseConfigFp() is not good place to pick up data_directory.
What about the attached patch which makes ProcessConfigFile() instead of
ParseConfigFp() pick up data_directory just after the configuration file is
parsed?
Regards,
--
Fujii Masao
