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

    astonishing-jelly-60479

    08/17/2015, 4:51 PM
    would need to somehow remotely cache the mapping between third party python packages and imports
  • a

    astonishing-jelly-60479

    08/17/2015, 4:51 PM
    for performance, rather than requiring everyone to have all packages installed locally
  • a

    astonishing-jelly-60479

    08/17/2015, 4:52 PM
    unless you can get manifests from pypi or something without downloading/installing 100+ packages
  • h

    happy-kitchen-89482

    08/17/2015, 4:57 PM
    So you're talking about doing away with BUILD files, essentially?
  • a

    astonishing-jelly-60479

    08/17/2015, 4:58 PM
    in my case i'm talking about doing away with specifying most python dependencies for python targets
  • h

    happy-kitchen-89482

    08/17/2015, 4:58 PM
    We're getting closer to that
  • a

    astonishing-jelly-60479

    08/17/2015, 4:59 PM
    at which point any entry in the build file for python things could probably be inferred somehow unless you've got special needs for a given target
  • h

    happy-kitchen-89482

    08/17/2015, 4:59 PM
    We still need occasional manually-specified deps because some pathological deps aren't inferrable
  • s

    salmon-helmet-84594

    08/17/2015, 5:00 PM
    Other question @enough-analyst-54434 : if I do zip_safe=False, and I include a number of resources, is there a way to get at those RAW files, or is it only through pkg_resources?
  • h

    happy-kitchen-89482

    08/17/2015, 5:00 PM
    E.g., classes loaded via class.forName
  • h

    happy-kitchen-89482

    08/17/2015, 5:01 PM
    But it still sounds like the dep extraction part overlaps with buildgen?
  • e

    enough-analyst-54434

    08/17/2015, 5:02 PM
    @salmon-helmet-84594 - if you do 
    zip_safe=False
    you can use os.path and file to find the resources
  • e

    enough-analyst-54434

    08/17/2015, 5:02 PM
    which is how'd you'd normally do it not running from a zip I presume
  • w

    witty-crayon-22786

    08/17/2015, 5:10 PM
    @happy-kitchen-89482: yes, it does. when i said "we have", i meant "the community has, in various forms". but in particular, twitter does have another scala compiler plugin that runs up to the typer and outputs this information
  • w

    witty-crayon-22786

    08/17/2015, 5:11 PM
    and there are two copies of the java dependency parsing around
  • f

    fancy-queen-20734

    08/17/2015, 5:40 PM
    Has anyone tried running pants with Jython?
  • h

    happy-kitchen-89482

    08/17/2015, 5:53 PM
    at some point it will probably make sense to standardize on a single way of extracting deps.
  • w

    witty-crayon-22786

    08/17/2015, 5:58 PM
    yea
  • f

    fancy-queen-20734

    08/17/2015, 7:14 PM
    Is anyone currently working towards the jar representation refactor? https://docs.google.com/document/d/1sEMXUmj7v-YCBZ_wHLpCFjkHOeWjsc1NR1hRIJ9uCZ8/edit#
  • w

    witty-crayon-22786

    08/17/2015, 7:28 PM
    any other scala reviewers interested in looking at https://rbcommons.com/s/twitter/r/2629/ ?
  • b

    billowy-alarm-80641

    08/17/2015, 10:22 PM
    I'm thinking about making this throw an exception instead of just doing a print: https://github.com/pantsbuild/pants/blob/master/src/python/pants/backend/core/register.py#L61
  • b

    billowy-alarm-80641

    08/17/2015, 10:22 PM
    what do we think, is it time?
  • w

    witty-crayon-22786

    08/17/2015, 10:25 PM
    #shipit
  • b

    billowy-alarm-80641

    08/17/2015, 10:26 PM
    cool, thanks
  • m

    mammoth-pharmacist-54215

    08/17/2015, 11:02 PM
    Trying to troubleshoot a test in style checker and I’m hoping someone has some idea of why this is happening.
  • m

    mammoth-pharmacist-54215

    08/17/2015, 11:03 PM
    the import_order plugin uses module.file as part of its test to see if a module is part of stdlib and compares it against what it find sin sysconfig.get_python_lib(standard_lib=1)
  • m

    mammoth-pharmacist-54215

    08/17/2015, 11:03 PM
    but I’m seeing paths like: /Users/molsen/workspace/pants/build-support/pants_dev_deps.venv/lib/python2.7/lib-dynload/math.so
  • m

    mammoth-pharmacist-54215

    08/17/2015, 11:04 PM
    any idea why we are pulling stdlib stuff into venv? Is there a sane way for me to grab the build support path if that is where we expect stdlib files to exist?
  • j

    jolly-chef-92794

    08/17/2015, 11:17 PM
    I find it unintuitive that 
    Distribution.locate(minimum_version=‘1.7’, maximum_version=‘1.7’)
    will almost invariably fail to find anything, because it’s counting things like 
    1.7.0.80
    as 
    > 1.7
    .
  • a

    ancient-stone-50795

    08/17/2015, 11:18 PM
    I always use 1.7.9999 or something like that as nax
1...182183184...2420Latest