On 17 May 2017 at 14:30, Robert Haas <robertmhaas@gmail.com> wrote:
> On Tue, May 16, 2017 at 9:20 PM, Andrew Dunstan
> <andrew.dunstan@2ndquadrant.com> wrote:
>> Inheriting variables from the environment is a part of make by design.
>> We have PG_PROVE_FLAGS for our own forced settings.
>
> I don't buy this argument. We've had previous cases where we've gone
> through and added -X to psql invocations in the regression tests
> because, if you don't, some people's .psqlrc files may cause tests to
> fail, which nobody wants. The more general principle is that the
> regression tests should ideally pass regardless of the local machine
> configuration. It's proven impractical to make that universally true,
> but that doesn't make it a bad goal. Now, for all I know it may be
> that setting PROVE_FLAGS can never cause a regression failure, but I
> still tend to agree with Peter that the regression tests framework
> should insulate us from the surrounding environment rather than
> absorbing settings from it.
>
In that case we're going to need to invent a way to do this similarly
in vcregress.pl. I'm not simply going to revert to the situation where
it and the makefiles are completely out of sync on this. The previous
patch was made more or less by ignoring the existence of vcregress.pl.
I will look at this again probably after pgcon. I don't think it's
terribly urgent.
cheers
andrew
--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services