On 04.02.2013 17:32, Alvaro Herrera wrote:
> Phil Sorber wrote:
>> On Mon, Feb 4, 2013 at 10:16 AM, Alvaro Herrera
>> <alvherre@2ndquadrant.com> wrote:
>>> Phil Sorber wrote:
>>>> On Mon, Feb 4, 2013 at 9:13 AM, Alvaro Herrera<alvherre@2ndquadrant.com> wrote:
>>>
>>>>> Uh, no existing code can use this new functionality? That seems
>>>>> disappointing.
>>>>
>>>> I wrote this because I wanted to use it in pg_isready. I also wrote
>>>> something for pg_isready to get around not having this. I think adding
>>>> these two functions to libpq would be the better option, but wanted
>>>> something that could fix an existing issue without having to patch
>>>> libpq so late in the 9.3 development process. Actually, I think it was
>>>> you who suggested that approach.
>>>
>>> Yes, I realize that (and yes, I did). But is no code other than
>>> pg_isready doing this? Not even the libpq URI test program?
>>
>> I think it probably would be able to benefit from this. Are you
>> suggesting I patch that too? I thought it was usually frowned upon to
>> touch random bits of working code like that. I'd be more than happy to
>> do it if it helps build the case for getting this added.
>
> Well, this qualifies as refactoring, so yeah, it helps the case.
I think this patch would simplift the patch to pass a connection string
to pg_basebackup and pg_receivexlog:
http://www.postgresql.org/message-id/005501cdfa45$6b0eec80$412cc580$@kommi@huawei.com.
I note that pg_dumpall also has a similar issue as pg_basebackup and
pg_receivexlog; there's no way to pass a connection string to it either.
If you could come up with a unified patch that takes care of all of
those, that would be great.
- Heikki
