Ergonomic, hermetic, user-friendly software build system for Python, Go, Java, Kotlin, Protobuf, Scala, Shell. Pants lets you fearlessly scale up your codebase!

Pants

```$ l /Users/pl/code/foursquare.web/.pants.d/compile/jvm/zinc/isolated-classes/.pants.d.gen.spindle.scala_record.src.thrift.com.foursquare.tip.tip-scala/
total 0
drwxr-xr-x   2 pl staff  68 Sep  8 13:10 .
drwxr-xr-x 363 pl staff 13K Sep  8 13:10 ..```

Not sure what I can do to convince you then :stuck_out_tongue_winking_eye:

I did kill the compile not too long after that finished

But I would be surprised if zinc would return before it had written out the classfiles

I'm going to try this again without the debugger attached

```13:34:31 01:07             [zinc]
                           [info] Compiling 7 Scala sources and 7 Java sources to /Users/pl/code/foursquare.web/.pants.d/compile/jvm/zinc/isolated-classes/.pants.d.gen.spindle.scala_record.src.thrift.com.foursquare.tip.tip-scala...
                           [info] Compile success at Sep 8, 2015 5:34:32 PM [43.976s]```

Either the debugger was the cause, or it takes a while to get to the repro

our best theory is still deep dependency graphs / long classpaths

you'll see lots of system time, because zinc is doing lots of stat'ing

Probably the structure of the codegen matters a lot too

Since zinc generates a classfile for every character in the source

But this is mitigated significantly by the jar-output patch you're working, right?

as soon as i'm able to do large scale testing i'll post some numbers

So, isolated only without a debugger attached, I compiled a reasonable graph in 12 minutes

remaining blockers for that testing are: 1) zinc support for jar'ing up javac outputs, 2) enough pants product changes to support compilation with jars