t/81_hosts.t fails when OpenDNS is configured as DNS provider

[Marc Lehmann]

On Wed, Oct 02, 2013 at 05:55:17PM +0200, Christian Walde <walde.christian at gmail.com> wrote:
> I never said that the entire cpantesters infrastructure has broken DNS, i
> said that i tried to explain what the cpantesters infrastructure expects
> a FAIL to mean in the ideal case and what the average user expects testing
> results to mean. As long as AnyEvent supplies a fail in this case, users
> will suspect it and not know that their system is broken.

What's your evidence that the average user expects this, as opposed to you
making up some statistic on the spot?

> Also do note that this is not the case of a single user. You have a number
> of reports on this in your cpantesters matrix:

I just went through all the fails on the latest perl release, and none are
due to that(or even to that test).

What are these failures you talk about, and what's your evidence that
these are indeed due to opendns (or a similar service), and not due to
some other bug?

> >Can you explain how this actually detects a broken dns service? It seems
> >you are simply skipping the test on failure, which is the same as not
> >having it in the first place.
> 
> It forcibly disables all hosts overrides by ensuring that ENV does not
> declare any host overrides and that any overrides in /etc/hosts are
> ignored.

/etc/hosts cannot override anything in AnyEvent. If DNS actually works and
/etc/hosts supplies bogus data, then this falls squarely into the realm of
"system misconfigured for the purpose of breaking stuff".

> Then it tries to resolve a broken hostname and expects no results
> on a healthy system. If there are no results it tests the override as
> intended, and if the system is broken it notes that and continues.

You miss the purpose of the test - the test is not, as you assume, to test
the host system, the test exists to check whether AnyEvent can properly
resolve on the host system. The test simply assumes that the host system
is working properly.

All that your patch does is to effectively disable the whole test, which
means it cannot even fulfill the purpose that, according to you, the
average user expects it to have.

-- 
                The choice of a       Deliantra, the free code+content MORPG
      -----==-     _GNU_              http://www.deliantra.net
      ----==-- _       generation
      ---==---(_)__  __ ____  __      Marc Lehmann
      --==---/ / _ \/ // /\ \/ /      schmorp at schmorp.de
      -=====/_/_//_/\_,_/ /_/\_\