On 01/29/2014 10:59 AM, Ian Lawrence Barwick wrote:
> 2014/1/29 Ian Lawrence Barwick <barwick@gmail.com>:
>> 2014-01-29 Andrew Dunstan <andrew@dunslane.net>:
>>> On 01/28/2014 05:55 AM, Ian Lawrence Barwick wrote:
>>>>
>>>> Hi Payal
>>>>
>>>> Many thanks for the review, and my apologies for not getting back to
>>>> you earlier.
>>>>
>>>> Updated version of the patch attached with suggested corrections.
>>> On a very quick glance, I see that you have still not made adjustments to
>>> contrib/file_fdw to accommodate this new option. I don't see why this COPY
>>> option should be different in that respect.
>> Hmm, that idea seems to have escaped me completely. I'll get onto it forthwith.
> Striking while the keyboard is hot... version with contrib/file_fdw
> modifications
> attached.
>
>
I have reviewed this. Generally it's good, but the author has made a
significant error - the idea is not to force a quoted empty string to
null, but to force a quoted null string to null, whatever the null
string might be. The default case has these the same, but if you specify
a non-empty null string they aren't.
That difference actually made the file_fdw regression results plain
wrong, in my view, in that they expected a quoted empty string to be
turned to null even when the null string was something else.
I've adjusted this and the docs and propose to apply the attached patch
in the next day or two unless there are any objections.
cheers
andrew
