Fixup pants dev mode

Review Request #121 — Created March 19, 2014 and submitted

benjyw, lahosken, travis
commit 8315fb62f9492117d576b791cc37b3fc1eb25315
Author: John Sirois <>
Date:   Wed Mar 19 10:04:07 2014 -0600

    Fixup pants dev mode
    This is a crutch fix only - use the full requirement set from the
    pants target to bootstrap with.  This works for bootstrapping a
    pants.pex since its additive and the pex creation should be hermetic
    anyhow and it gives full functionality when running under PANTS_DEV=1
    against the live sources.
    Also align 3rdparty/python:setuptools version with common/python's default.

 3rdparty/python/BUILD         |  2 +-
 build-support/python/ | 25 ++++++++++++++++++-------
 pants.bootstrap               |  8 +-------
 3 files changed, 20 insertions(+), 15 deletions(-)
$ ./build-support/python/ 
$ ./pants.bootstrap
$ PANTS_DEV=1 ./pants.bootstrap goal fred
... ansicolors! ...
  • 0
  • 0
  • 1
  • 0
  • 1
Description From Last Updated
  2. 3rdparty/python/BUILD (Diff revision 1)
    Wickman and I paired and got the prototype fix, which I'm cleaning up to post today. Mind holding off on this?
    1. Not at all.
    2. Any objections to proceeding with this now?  I rebased and it all works just as before.
    3. Since Travis isn't in the office right now, I was gonna guess "oh, he's still sick, don't expect progress on the Travis-Wickman pair-fix loveliness today." But now I remember it's work-from-home Thursday. tl;dr: Unsurprisingly, I have no useful info to to aid this decision.
    4. Hold the phone: he just walked in. *waves* Hi, Travis!
  1. Yay! Does this mean we can get rid of various places where we surround imports with try-except?
    1. Exactly.
      I am going to hold off though for Travis' work before sending this home.
    2. Status update on the full fix:
      I have the fix ready that resolves the setuptools issue (Wickman actually wrote a substantial portion while we paired and I'm finishing it up). However, now there's an issue with synthetic thrift targets that are added as dependencies ( I'm working through that part now and hope to get the review out today.
    3. Tests just passed, but I'd like to clean a few things up before posting for review. Given this is taking longer than I hoped, LMK if you'd like to submit this change, submit knowing it needs some cleanup, or wait for that to be cleaned up.
      Something I noticed in the test run ( is the atexit handler issues. Brian and I talked about that and its because objects referenced in the atexit are being cleaned up before they're used for the final time. That should be pretty straightforward now that we know it needs to happen.
    4. Absolutely no rush on this.  As the commit message says - "This is a crutch fix only...".  I think we want the functionality in this RB for sure, but this implementation requires manually updating a list of requirements to match pants real requirements, so that + it only affects us pants devs and only in a minor way t that says - charge on sir.
  1. Ship It!
  1. Thanks - merged
Review request changed

Status: Closed (submitted)