https://pantsbuild.org/ logo
Powered by
#general
  • e

    enough-analyst-54434

    07/29/2015, 6:08 PM
    OK - good, yw
  • e

    enough-analyst-54434

    07/29/2015, 6:20 PM
    @mammoth-autumn-2551: not sure if you still wanted 0.9.1 for python, but the script to build is here: https://github.com/pantsbuild/binaries/blob/gh-pages/build-thrift-0.9.1.sh
  • e

    enough-analyst-54434

    07/29/2015, 6:20 PM
    The linux binaries are both built and synced up to bintray
  • m

    mammoth-autumn-2551

    07/29/2015, 6:21 PM
    Thanks, I’ll build that just in case
  • e

    enough-analyst-54434

    07/29/2015, 6:35 PM
    I'm thinking of stripping pants static binaries - its reducing size by ~90%. Are there reasons folks know of not to do this? I'm assuming the incidence of a pants user a.) finding the binary pants uses in the pants cache and b.) running gdb against it are small
  • h

    happy-kitchen-89482

    07/29/2015, 6:38 PM
    which binaries?
  • e

    enough-analyst-54434

    07/29/2015, 6:42 PM
    The static ones we host on bintray: thrift, protoc, ragel
  • m

    mammoth-autumn-2551

    07/29/2015, 6:44 PM
    @enough-analyst-54434: what is the advantage given that they’re already hosted externally?
  • e

    enough-analyst-54434

    07/29/2015, 6:44 PM
    The download of 30MB vs 3MB
  • e

    enough-analyst-54434

    07/29/2015, 6:44 PM
    is ... for Montanans - a significant difference
  • m

    mammoth-autumn-2551

    07/29/2015, 6:44 PM
    so download time if its not in your local cache?
  • m

    mammoth-autumn-2551

    07/29/2015, 6:44 PM
    haha
  • h

    happy-kitchen-89482

    07/29/2015, 6:47 PM
    sgtm
  • e

    enough-analyst-54434

    07/29/2015, 6:55 PM
    Thanks for the build help @mammoth-autumn-2551 , 0.9.1 is now available for all supported architectures on bintray
  • w

    witty-crayon-22786

    07/29/2015, 7:00 PM
    ...this is going to sound crazy, but: the chroot created during: 
    [python-setup]
    should not be affected by whether the target calls 
    import random
    ... right? i think i'm imagining import sideeffects
  • r

    rough-minister-58256

    07/29/2015, 7:01 PM
    importing shouldn’t have any bearing on whats created in the chroot
  • e

    enough-analyst-54434

    07/29/2015, 7:09 PM
    however, it is known that ... one of the bits of the python toolchain does leak PYTHONPATH into its chroots iirc
  • e

    enough-analyst-54434

    07/29/2015, 7:09 PM
    looking...
  • e

    enough-analyst-54434

    07/29/2015, 7:11 PM
  • e

    enough-analyst-54434

    07/29/2015, 7:11 PM
    Thats a bug
  • e

    enough-analyst-54434

    07/29/2015, 7:11 PM
    But it should only be chroots under test that have this leak
  • e

    enough-analyst-54434

    07/29/2015, 7:11 PM
    You should be able to run repl against the same test target and not import from outside its chroot
  • r

    rough-minister-58256

    07/29/2015, 7:11 PM
    that would only happen if you have an existing PYTHONPATH env variable defined?
  • e

    enough-analyst-54434

    07/29/2015, 7:12 PM
    correct
  • e

    enough-analyst-54434

    07/29/2015, 7:12 PM
    wrapper scripts to run pants often do
  • e

    enough-analyst-54434

    07/29/2015, 7:12 PM
    even if they didn't mean to
  • r

    rough-minister-58256

    07/29/2015, 7:12 PM
    i see
  • w

    witty-crayon-22786

    07/29/2015, 7:12 PM
    it looks like changing anything about this file is sufficient to avoid the bug I'm seeing. so it's possible we're getting cache results from elsewhere
  • e

    enough-analyst-54434

    07/29/2015, 7:12 PM
    I'm thinking pants_internal
  • e

    enough-analyst-54434

    07/29/2015, 7:12 PM
    if that repo still exists
1...899091...2420Latest