On Mon, 16 Jan 2017 at 10:32:37 +0100, Martin Pitt wrote:
> This was already discussed several times, last time in
> <https://bugs.debian.org/801895>.
Since this comes up repeatedly, would you be OK with having some text in
the specification that explicitly says "it is not a bug for a test to
make itself visible on the network"? That way it's at least unambiguous.
Or a feature, Features: no-network or something. Even if nobody actually
added it to their tests in practice, its existence would act as
documentation that this is not the default assumption.
> > Subject: [PATCH 2/2] doc: Add restrictions for using or reconfiguring apt
>
> This seems unrelated.
I had viewed it as a special case of networking (communicating with one
particular server), as the submitter of #833503 did.
I recognise that you think the submitter of #833503 was wrong about that,
and that's why I wanted to document it one way or the other.
> Such a test should add
> "Restrictions: breaks-testbed" if it's doing something dangerous
I would have expected that we would want to force the testbed to be
reverted after messing with apt configuration (as this patch does,
although I didn't make that clear enough in the commit message), even if
the testbed is not actually "broken". But I can keep using breaks-testbed
for that in nss-mdns if you prefer.
S
