libexec: atf is needed by kyua, which is wired by MK_TESTS_SUPPORT

[markj]

I don't see how kyua depends on libexec/atf, and a WITHOUT_TESTS= buildworld completes for me without errors. What exactly are you building, and how does it fail?

I think the patch is probably right in any case, but the justification doesn't make sense to me.

There is another bug in that WITHOUT_TESTS= implies WITHOUT_TESTS_SUPPORT=, but the converse is not true. That's backwards.

[franco_opnsense.org]

I have WITHOUT_TESTS=yes and WITH_TESTS_SUPPORT=yes which adds kyua and libatf, but not atf tools in libexec. On a build without WITHOUT_TESTS=yes the atf tools are placed in the base system as a side effect.

[franco_opnsense.org]

(kyua depends on e.g. libexec/atf-sh for script execution via shebang in /usr/src/tests/sys/netpfil/pf/*.sh)

[franco_opnsense.org]

Meh, sorry for the misinformation. All of this is very difficult to trace.

Simply put: If I delete /usr/libexec/atf* then kyua says:

1. kyua debug -k /usr/tests/sys/netpfil/pf/Kyuafile killstate:match
2. kyua E: Unknown test case 'killstate:match'.

If the atf tools are there it works as expected so I think this was meant to be clustered together.

[markj]

In D47660#1086689, @franco_opnsense.org wrote:

Meh, sorry for the misinformation. All of this is very difficult to trace.

Simply put: If I delete /usr/libexec/atf* then kyua says:

1. kyua debug -k /usr/tests/sys/netpfil/pf/Kyuafile killstate:match
2. kyua E: Unknown test case 'killstate:match'.

If the atf tools are there it works as expected so I think this was meant to be clustered together.

I see now. I thought you meant that kyua has a build-time dependency on atf.

I'd commit this with the following log:

atf: Guard libexec/atf with WITH_TESTS_SUPPORT

Other atf components are guarded by WITH_TESTS_SUPPORT already.

Is that ok with you? Or, feel free to mail me a patch with the commit log that you want.

[franco_opnsense.org]

Sounds good to me, thanks :)