filter tests:: --filter-ancestor=-blacklist:blacklist ain't working. try to fix

Review Request #925 - Created Aug. 20, 2014 and submitted

Larry Hosken
patricklaw, stuhood
filter tests:: --filter-ancestor=-blacklist:blacklist ain't working. try to fix

add a test for this case

the problem: at the time the ancestor filter does its thing, the blacklist:blacklist target doesn't "realize" it has deps. My hazy understanding: we added blacklist:blacklist to the build graph, but didn't tell it to, uhm, recurse its deps. I think this change tells it to recurse its deps. And I guess that's what we want here. But I don't necessarily know about better approaches, silly repercussions, etc.
Stu Hood
Eric Ayers

maybe assert that it is there before you apply the argument?
Larry Hosken
Review request changed

Status: Closed (submitted)

Larry Hosken

probably a great suggestion, but unsurprisingly I don't understand it: assert what is where?
Eric Ayers

After you add in your unwanted target:

      targets=self.targets('common/::') # blacklist is in the list of targets

Also, I am working to enable --exclude-target-regexp so that it will work with filter and list goals and filter out targets before we try to inject them in the build graph.
  1. oh, good idea. coming soon in an RB near you
Patrick Lawson

I'm mildly surprised that `<spec_path>/::` works as expected.  Do we explicitly support the optional trailing slash?
  1. on the command line, yeah. It's nice to type ./pants goal something sr<Tab complete>py<Tab complete>::