Re: Re: Missing rows with index scan when collation is not "C" (PostgreSQL 9.5)

On Mar 23, 2016 18:53, "Peter Geoghegan" <pg(at)heroku(dot)com> wrote:
>
> On Wed, Mar 23, 2016 at 10:46 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> > If I had to make a recommendation right now, I would go for your
> > option #4, ie shut 'em all down Scotty. We do not know the full extent
> > of the problem but it looks pretty bad, and I think our first priority
> > has to be to guarantee data integrity.
>
> +1, but only for glibc, and configurable. The glibc default might
> later be revisited in the stable 9.5 branch.
>

Are you talking about configurable at./configure time, or guc?

Making it a compile time option makes sense I think. But turning it into a
guc will expose users to a lot of failure scenarios if they *change* the
value, and that seems risky.

Putting it in autoconf and default to off in the upcoming minor seems like
a good idea. Then once we have more information, we can consider if we want
to turn it back on in backbranches our just in 9.6 (when/if properly
fixed).

/Magnus