#general
- w
witty-crayon-22786
08/31/2015, 5:58 PMso, the compile_classpath already contains symlinks - w
witty-crayon-22786
08/31/2015, 6:03 PM@bored-art-40741: ...which i think means you can just kill the
usage. cc @fancy-queen-20734ivy_resolve_symlink_map - f
fancy-queen-20734
08/31/2015, 6:07 PMthough, compile_classpath get_for_target has different semantics from what the earlier code was doing - f
fancy-queen-20734
08/31/2015, 6:07 PMeg, if JarLib A has pants dependencies in its declaration, those jars will be in the transitive list - w
witty-crayon-22786
08/31/2015, 6:08 PMmm, yes. - f
fancy-queen-20734
08/31/2015, 6:09 PMTo have the same meaning, we’d need a way to pull the paths out just for the target in question without its transitive closure - w
witty-crayon-22786
08/31/2015, 6:14 PMprolly worth it, yea. - b
bored-art-40741
08/31/2015, 7:00 PMWas the conclusion here that we can't be 100% sure with the current state of things? - b
bored-art-40741
08/31/2015, 7:02 PMIn this particular situation our custom resolver is the only thing that populatescompile_classpath - b
bored-art-40741
08/31/2015, 7:02 PMAnd
for that matterivy_symlink_map - b
bored-art-40741
08/31/2015, 7:02 PMBut I'm trying to be conservative to make sure I don't break this for others - e
enough-analyst-54434
08/31/2015, 7:04 PMNo one else uses a custom resolver, so appropriate tests in pants should exercise the ivy resolver and learn us about breaks - e
enough-analyst-54434
08/31/2015, 7:04 PMNot sure if those appropriate tests exist or not - e
enough-analyst-54434
08/31/2015, 7:05 PMie: we shouldn't need to be conservative wrt the ivy resolver, it should be robustly tested since we have our hands on it - w
witty-crayon-22786
08/31/2015, 7:14 PM@bored-art-40741: er, no: you can be 100% sure. nick's comment was about an independent issue - b
bored-art-40741
08/31/2015, 7:41 PMOkay, great - b
bored-art-40741
08/31/2015, 7:41 PMThat removes another product dependency! - b
bored-art-40741
08/31/2015, 7:41 PMThat significantly simplified that step of the task, sweet - a
ancient-stone-50795
08/31/2015, 7:41 PMThanks Patrick LGTM - b
bored-art-40741
08/31/2015, 7:41 PMQuestion for the crowd, especially twitter: do we intend to kill
eventually?ScalaLibrary.java_sources - a
ancient-stone-50795
08/31/2015, 7:42 PMI think you mean on ScalaLibrary targets? - b
bored-art-40741
08/31/2015, 7:42 PMYeah - b
bored-art-40741
08/31/2015, 7:42 PMClarified - a
ancient-stone-50795
08/31/2015, 7:42 PMhttps://github.com/pantsbuild/pants/issues/65 says we are, but it is old - e
enough-analyst-54434
08/31/2015, 7:44 PMJohn Townsend filed a related issue last week - e
enough-analyst-54434
08/31/2015, 7:45 PMThey both depend on addressable SourceSets which no one has built yet - e
enough-analyst-54434
08/31/2015, 7:45 PMJohn's was re JvmApp bundles - folks want to share bundles not in their tree - e
enough-analyst-54434
08/31/2015, 7:46 PM - w
witty-crayon-22786
08/31/2015, 7:48 PMwe desire to kill it, but there are a few legitimate usecases - w
witty-crayon-22786
08/31/2015, 7:49 PM(we're down to 22 internally) (EDIT: 50 ->22)