Re: Overlength socket paths (was Re: [COMMITTERS] pgsql: Refactor flex and bison make rules)

Andrew Dunstan <andrew@dunslane.net> writes:
> On 11/29/2012 07:16 PM, Tom Lane wrote:
>> Um ... that's *another* place that needs to change then, because the
>> test script fires up postmasters on its own, outside of pg_upgrade.

> True, but it doesn't do anything about setting the socket dir, so those 
> just get the compiled-in defaults. What exactly do you want to change 
> about the test script?

Well, I was thinking about also solving the problem that the compiled-in
default might be no good in a build environment.
        regards, tom lane