Обсуждение: Floating point timestamps

Hi,

Is the plan to remove support for floating point timestamps at some
stage?  If so, what is that waiting on, and would it provide
sufficient warning if (say) 9.6 were documented as the last major
release to support that build option?

Thanks!

-- 
Thomas Munro
http://www.enterprisedb.com

Thomas Munro <thomas.munro@enterprisedb.com> writes:
> Is the plan to remove support for floating point timestamps at some
> stage?  If so, what is that waiting on, and would it provide
> sufficient warning if (say) 9.6 were documented as the last major
> release to support that build option?

AFAIK there is no particular plan to do that.  It's not like leaving
that code in place is costing us huge amounts of maintenance effort.
        regards, tom lane

* Tom Lane (tgl@sss.pgh.pa.us) wrote:
> Thomas Munro <thomas.munro@enterprisedb.com> writes:
> > Is the plan to remove support for floating point timestamps at some
> > stage?  If so, what is that waiting on, and would it provide
> > sufficient warning if (say) 9.6 were documented as the last major
> > release to support that build option?
>
> AFAIK there is no particular plan to do that.  It's not like leaving
> that code in place is costing us huge amounts of maintenance effort.

Agreed, and I have little doubt that it's still used in the field given
how long it was the default for some distributions.

Thanks!

Stephen

Stephen Frost wrote:
> * Tom Lane (tgl@sss.pgh.pa.us) wrote:
> > Thomas Munro <thomas.munro@enterprisedb.com> writes:
> > > Is the plan to remove support for floating point timestamps at some
> > > stage?  If so, what is that waiting on, and would it provide
> > > sufficient warning if (say) 9.6 were documented as the last major
> > > release to support that build option?
> > 
> > AFAIK there is no particular plan to do that.  It's not like leaving
> > that code in place is costing us huge amounts of maintenance effort.
> 
> Agreed, and I have little doubt that it's still used in the field given
> how long it was the default for some distributions.

... and the fact that you can pg_upgrade from old versions that had such
defaults, but only if the new install uses the same datetime
representation.  IOW it's quite likely that there are versions in the
field still working with FP datetimes.

-- 
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services