On Tue, Mar 06, 2018 at 02:54:16PM -0500, Peter Eisentraut wrote:
> This contains a slight problem: The tests contain two different
> branches, depending on whether tls-server-end-point is supported. But
> these branches run a different number of tests, so the test count of the
> top of the test
On 2/24/18 10:12, Peter Eisentraut wrote:
> On 2/24/18 07:37, Michael Paquier wrote:
>> On Fri, Feb 23, 2018 at 01:57:44PM -0500, Peter Eisentraut wrote:
>>> Oh. I actually had that file as 0600 in my checked-out tree, probably
>>> from earlier experiments. Fixed that. And I also changed it to m
On 2/24/18 07:37, Michael Paquier wrote:
> On Fri, Feb 23, 2018 at 01:57:44PM -0500, Peter Eisentraut wrote:
>> Oh. I actually had that file as 0600 in my checked-out tree, probably
>> from earlier experiments. Fixed that. And I also changed it to make
>> another temp file that is explicitly 064
On Fri, Feb 23, 2018 at 01:57:44PM -0500, Peter Eisentraut wrote:
> Oh. I actually had that file as 0600 in my checked-out tree, probably
> from earlier experiments. Fixed that. And I also changed it to make
> another temp file that is explicitly 0644, because we can't rely on that
> being the d
On 2/22/18 23:58, Michael Paquier wrote:
> One of the tests is failing:
> t/001_ssltests.pl .. 1/62
> # Failed test 'certificate authorization fails with revoked client cert:
> matches'
> # at /home//git/postgres/src/test/ssl/../../../src/test/perl/TestLib.pm
> line 354.
> #
On Thu, Feb 22, 2018 at 08:34:30AM -0500, Peter Eisentraut wrote:
> I noticed that a couple of test cases in the SSL tests fail to connect
> not for the reason that the tests think they should. Here is a patch to
> augment the test setup so that a test for connection rejection also
> checks that w
