Re: Testing 9.2 in ~production environment

Peter Eisentraut <peter_e(at)gmx(dot)net> writes:
> On mn, 2012-06-18 at 17:57 -0400, James Cloos wrote:
>> I presume that lc_ctype is the significant difference?

> It certainly makes some difference, but it's a bit shocking that makes
> things that much slower.

If James is testing text-comparison-heavy operations, it doesn't seem
shocking in the least. strcoll() in most non-C locales is a pig.

>> LC_CTYPE *is* specified as 'C' in the dump from which I created the 9.2
>> cluster, so it must have been overridden by pg_restore. I see that my
>> dist's /etc rc script now sets LC_CTYPE. Would that explain why lc_ctype
>> changed between the two clusters?

> It's possible, depending on how exactly the start up script maze is set
> up on your particular OS.

pg_dumpall should generate a script that correctly restores database
locales. However, pg_dump+pg_restore is dependent on user creation
of the specific database, which is likely to be environment sensitive.
We really oughta do something about that ...

regards, tom lane