On Thu, Feb 11, 2016 at 1:41 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
>> On Thu, Feb 11, 2016 at 1:19 PM, Andres Freund <andres@anarazel.de> wrote:
>>> Well, I can't do anything about that right now. I won't have the time to
>>> whip up the new/more complex API we discussed upthread in the next few
>>> days. So either we go with a simpler API (e.g. pretty much a cleaned up
>>> version of my earlier patch), revert the postmaster deatch check, or
>>> somebody else has to take lead in renovating, or we wait...
>
>> Well, I thought we could just revert the patch until you had time to
>> deal with it, and then put it back in. That seemed like a simple and
>> practical option from here, and I don't think I quite understand why
>> you and Tom don't like it.
>
> Don't particularly want the git history churn, if we expect that the
> patch will ship as-committed in 9.6. If it becomes clear that the
> performance fix is unlikely to happen, we can revert then.
>
> If the performance change were an issue for a lot of testing, I'd agree
> with a temporary revert, but I concur with Andres that it's not blocking
> much. Anybody who does have an issue there can revert locally, no?
True. Maybe we'll just have to start doing that for EnterpriseDB
benchmarking as standard practice. Not sure everybody who is
benchmarking will realize the issue though.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company