Re: pgsql: Give a better error for duplicate entries in VACUUM/ANALYZE colu

On Mon, Sep 25, 2017 at 1:50 PM, Peter Eisentraut <
peter(dot)eisentraut(at)2ndquadrant(dot)com> wrote:

> On 9/25/17 15:09, Tom Lane wrote:
> > Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com> writes:
> >> On 9/21/17 18:13, Tom Lane wrote:
> >>> Give a better error for duplicate entries in VACUUM/ANALYZE column
> list.
> >
> >> In the error message, we should write "specified more than once" instead
> >> of "specified twice", because that could otherwise look a bit silly:
> >> VACUUM ANALYZE vaccluster(i,i,i);
> >> ERROR: column "i" of relation "vaccluster" is specified twice
> >
> > OK.
> >
> >> (Also, the "is" doesn't seem to fit there.)
> >
> > Hm, reads fine to me, and I'd still rather include "is" in the
> > revised wording. Anybody else agree with Peter's wording?
>
> Note a big deal. I'm just working off existing error messages:
>

​About half of those, especially the "appears" ones, seem unhelpful for
deciding whether to add "is" here; "is appears" just doesn't work.

​I think the added length due to the "of relation %", makes dropping the
'is' sound more odd than those like "column % specified"

The middle ground would be writing: column "i" of relation "vaccluster"
appears more than once; I'm good with using appears instead of deciding
between [is] specified.

I'm not seeing that we have a formal distinction between "specified" and
"appears"...

David J.