Getting rid of the ATF tools

[Garrett Cooper]

On Fri, Jul 19, 2013 at 2:06 PM, Julio Merino <julio at meroh.net> wrote:
> On Fri, Jul 19, 2013 at 5:02 PM, Garrett Cooper <yaneurabeya at gmail.com> wrote:
>> (Pardon my message mangling as I'm forwarding this to freebsd-testing@
>> for context)
>
> Hm, I wanted to rewrite things a little bit before sending this
> publicly but no big deal. Didn't know freebsd-testing existed!

It's up on the wiki, but it's not as clearly marked as it could be :).
It's a super duper low traffic list (I'm the only person that's posted
to it since it's been recreated, apart from you now).

>> gcooper> 2. Please move the OptionalObsoleteFiles.inc entries
>> [re]moved to ObsoleteFiles.inc as they need to be removed from systems
>> running CURRENT that have installed atf in the past (like my VM for
>> instance). There aren't many people using ATF I suspect, but for those
>> of us that are, it would be nice if the files were yanked completely.
>
> OK.  Note, however, that the entries in OptionalObsoleteFiles.inc are
> _outside_ of any MK* conditionals.  Is there any case where they'd be
> ignored?  (curiosity)

It's cleanliness and consistency more than anything else as the tools
are being yanked (and I assume aren't coming back ever :D..). Adding
in logic to check for commands seems unnecessary as make does a
reasonable job with this already.

On the plus side you're getting rid of bmake-isms which will make
moving away from bmake a possibility in the future. I am against the
fact that ATF/Kyua/etc can't easily be backported without the mess
that's involved with converting the src tree over to bmake. It's
already made my life more annoyingly painful than necessary.

>> gcooper> 3. Please put in the needed bits for generating Kyuafiles in
>> atf.test.mk.
>
> Yes, but that has to come separately.  (And it should probably be in
> bsd.test.mk, because Kyuafiles can represent both generic test
> programs and ATF-specific test programs -- a major selling point for
> using Kyua upfront in a system where non-ATF test programs already
> exist.)

Ok, but let's do that all in one fell swoop then. No reason to have an
awkward stage of flux :).

> Thanks for the comments!

np -- thanks for your care/concern :).