Fix java missing dep detection

Review Request #1759 — Created Feb. 12, 2015 and discarded

patricklaw
pants
pants-reviews
benjyw, ity, jsirois, zundel
Fix java missing dep detection

This is a "WIP" even though we have it patched in internally, so I'm quite sure it works. I want to backstop this with tests though. The way these products are passed around also needs to be rethought in the new world of round engine.

JS
  1. Thanks for posting this Patrick.
  2. It definitely requires 'symlink_map' and 'classes_by_target' too.  I can take this up this weekend if no-one else has and add a test.
    1. Note though that it's always instantiated as part of a JvmCompile subclass execute, so in a couple of cases (classes_by_target) it already implicitly has the product, and more importantly, it would create a cycle if it explicitly requested it.

    2. K - I'll take that in account by commenting well.
  3. 
      
PG
  1. This doesn't seem to work for me, although I've only tried patching it on top of vanilly 0.0.28.

    1. Since your issue was with source -> source missing deps, that's not super surprising (unfortunately). This is specifically dealing with third party deps brought in by ivy. If I had to hazard a guess based on your problem description, I'd say maybe the annotation processing task needs to be populating a product for jvm missing dep detection to consider? Benjy would be a good person to ask about this.

  2. 
      
ZU
  1. FYI, I believe commit 34ebbe1 incorporates all of these changes up to this point.

    See https://rbcommons.com/s/twitter/r/1753

  2. 
      
PA
Review request changed

Status: Discarded

Change Summary:

This change is subsumed by https://rbcommons.com/s/twitter/r/1753. Thanks Eric! Soon I will create a new RB that adds specific integration tests to guard against future regression.

Loading...