[Swan-dev] TESTLIST cleanout

Paul Wouters paul at nohats.ca
Thu Feb 4 10:54:40 UTC 2016

On Tue, 2 Feb 2016, Andrew Cagney wrote:

> I assume the UML (Userland Linux?) testing is dead, so I'd like to
> change all entries like:
>   umlplutotest    ikev2-07-biddown                        good
> to, pick 1:
>   skip    ikev2-07-biddown                        wip
>    kvmplutotest    ikev2-07-biddown                        wip
>    #kvmplutotest    ikev2-07-biddown                        wip
> it affects a lot of tests.

Why does it matter? The umltests are not converted yet but those are
valid tests which we will convert slowly. I'm fine turning them into
comments. Just not sure I understand the problem with them now :)

> For all the below, replace "bad" and "incomplete" with "wip":

So "bad" tests were not tests that were bad, but were tests that
were expected to mismatch their reference output. Image a test with
esp=aes on east and esp=3des on west. It would have the output of
the test with esp=aes on both sides. I really did not like this,
so I have changed all thoses tests to "good" tests, sometimes adding
a grep to confirm a failure.

> kvmplutotest    ipv6-v6-through-v6-klips-klips        incomplete
> kvmplutotest    ipv6-transport-mode-01-klips-klips    incomplete
> kvmplutotest    ipv6-transport-mode-02-netkey-netkey    incomplete
> kvmplutotest    ipv6-transport-mode-03-klips-netkey    incomplete
> kvmplutotest    ipv6-tunnel-mode-01-klips-klips        incomplete
> kvmplutotest    ipv6-tunnel-mode-03-klips-netkey    incomplete
> kvmplutotest    ipv6-transport-ts-mode-04-netkey-netkey    incomplete

I think those tests mostly worked on some systems but not on all?

> PS, I kind of think that TESTLIST is past its due date, it is
> duplicating knowlege in each test directory

Agreed, but not now please :)


More information about the Swan-dev mailing list