Thread: Disabling inferred "group by" columns in 9.1?
Is there any way to disable the guessing of missing columns in a group-by (the feature added in 9.1)? I haven't found any option but I might not be searching for the right thing. http://wiki.postgresql.org/wiki/What%27s_new_in_PostgreSQL_9.1#SQL_and_PL.2FPgSQL_features
"Evan D. Hoffman" <evandhoffman@gmail.com> writes:
> Is there any way to disable the guessing of missing columns in a
> group-by (the feature added in 9.1)?
Why would you want to do that? There's no "guessing" involved: the
behavior is always correct, and it is required by the SQL standard.
regards, tom lane
Because we have both 9.0 and 9.1 running, and a query that succeeds under 9.1 was failing under 9.0. Can I assume the answer then is "no"? On Thu, Oct 11, 2012 at 2:38 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > "Evan D. Hoffman" <evandhoffman@gmail.com> writes: >> Is there any way to disable the guessing of missing columns in a >> group-by (the feature added in 9.1)? > > Why would you want to do that? There's no "guessing" involved: the > behavior is always correct, and it is required by the SQL standard. > > regards, tom lane
Evan D. Hoffman wrote on 11.10.2012 20:54: >> Is there any way to disable the guessing of missing columns in a >> group-by (the feature added in 9.1)? > > Because we have both 9.0 and 9.1 running, and a query that succeeds > under 9.1 was failing under 9.0. Can I assume the answer then is > "no"? > If you cannot upgrade the 9.0 installation, the only way how you can solve that problem is to develop the statements using9.0
"Evan D. Hoffman" <evandhoffman@gmail.com> writes:
> Because we have both 9.0 and 9.1 running, and a query that succeeds
> under 9.1 was failing under 9.0. Can I assume the answer then is
> "no"?
There are any number of reasons a query might work in version N and not
in N-1. If you've got requirements like that, you should be developing
on the oldest version you're worried about supporting ...
regards, tom lane
Ok. Thanks for your help. On Oct 11, 2012, at 3:31 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > "Evan D. Hoffman" <evandhoffman@gmail.com> writes: >> Because we have both 9.0 and 9.1 running, and a query that succeeds >> under 9.1 was failing under 9.0. Can I assume the answer then is >> "no"? > > There are any number of reasons a query might work in version N and not > in N-1. If you've got requirements like that, you should be developing > on the oldest version you're worried about supporting ... > > regards, tom lane