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

    astonishing-jelly-60479

    07/23/2015, 7:42 PM
    problem was also present in 0.0.33
  • r

    rough-minister-58256

    07/23/2015, 7:42 PM
    you can test this theory by building a pex from the problematic python_library (which probably will throw ImportError at runtime), unzipping that and then moving the ./gen directory out of the way - then running 
    python2.7 ./unzipped_pex_dir
  • e

    enough-analyst-54434

    07/23/2015, 7:43 PM
    OK - well I fixed path issues in the ThriftBuilder as part of another fix just yesterday
  • e

    enough-analyst-54434

    07/23/2015, 7:43 PM
    and added a test
  • e

    enough-analyst-54434

    07/23/2015, 7:43 PM
    So, issue may be solved by that
  • e

    enough-analyst-54434

    07/23/2015, 7:43 PM
  • r

    rough-minister-58256

    07/23/2015, 7:43 PM
    oh cool, will ask Stu to cherrypick that in our in-flight release and see if it fixes the problem
  • e

    enough-analyst-54434

    07/23/2015, 7:44 PM
    That would be great
  • a

    astonishing-jelly-60479

    07/23/2015, 7:46 PM
    Copy code
    18:09:35 00:00 [main]
               (To run a reporting server: ./pants server)
18:09:35 00:00   [bootstrap]
18:09:35 00:00   [setup]
18:09:35 00:00     [parse]
               Executing tasks in goals: bootstrap -> imports -> unpack-jars -> deferred-sources -> gen -> resolve -> compile -> resources -> binary
18:09:36 00:01   [bootstrap]
18:09:36 00:01     [bootstrap-jvm-tools]
18:09:36 00:01   [imports]
18:09:36 00:01     [ivy-imports]
18:09:36 00:01   [unpack-jars]
18:09:36 00:01     [unpack-jars]
18:09:36 00:01   [deferred-sources]
18:09:36 00:01     [deferred-sources]
18:09:36 00:01   [gen]
18:09:36 00:01     [thrift]
18:09:36 00:01     [protoc]
18:09:36 00:01     [antlr]
18:09:36 00:01     [ragel]
18:09:36 00:01     [jaxb]
18:09:36 00:01     [wire]
18:09:36 00:01     [scrooge]
18:09:36 00:01     [spindle]
18:09:36 00:01   [resolve]
18:09:36 00:01     [ivy]
18:09:36 00:01   [compile]
18:09:36 00:01     [compile]
18:09:36 00:01     [jvm]
18:09:37 00:02       [jvm-compilers]
18:09:37 00:02     [python-eval]
                   Invalidated 1 target.
18:09:37 00:02       [eval-targets].
                         Failed to eval XXXXX/XXXXX:XXXXX
                         Failed to eval 'entry_point XXXXX.run':
                           File "/tmp/tmp5C_LS1/XXXXX/run.py", line 11, in <module>
                             from XXXXX.create_app import create_app
                           File "/tmp/tmp5C_LS1/XXXXX/create_app/__init__.py", line 1, in <module>
                             try:
                           File "/tmp/tmp5C_LS1/XXXXX/create_app/normal.py", line 8, in <module>
                             from XXXXX.blueprints import register_blueprints
                           File "/tmp/tmp5C_LS1/XXXXX/blueprints.py", line 1, in <module>
                             from XXXXX.oscar.YYYYYY.YYYYYY_blueprint import YYYYYY_blueprint
                           File "/tmp/tmp5C_LS1/XXXXX/oscar/WWWWW/WWWWW_blueprint.py", line 7, in <module>
                             from gen.thrift.ZZZ.service import ThriftService
                           File "/tmp/tmp5C_LS1/gen/thrift/ZZZ/service/ThriftService.py", line 10, in <module>
                             from ttypes import *
                         No module named ttypes
  • a

    astonishing-jelly-60479

    07/23/2015, 7:46 PM
    that's the specific point it's failing for us, before the pex even gets built
  • e

    enough-analyst-54434

    07/23/2015, 7:46 PM
    yup - ok, well good!
  • e

    enough-analyst-54434

    07/23/2015, 7:46 PM
    without the "compile" step the pex would have silently built ok
  • a

    astonishing-jelly-60479

    07/23/2015, 7:46 PM
    yeah
  • e

    enough-analyst-54434

    07/23/2015, 7:47 PM
    @rough-minister-58256: I assume you have zip_safe=False useage for apps at Twitter - if so you'll blow up on your upgrade if it includes pex 1.0.x - https://rbcommons.com/s/twitter/r/2517/
  • r

    rough-minister-58256

    07/23/2015, 7:48 PM
    ah good to know
  • e

    enough-analyst-54434

    07/23/2015, 7:48 PM
    So you may need to build a custom pex
  • e

    enough-analyst-54434

    07/23/2015, 7:48 PM
    (pex lib)
  • e

    enough-analyst-54434

    07/23/2015, 8:02 PM
    Speak up if you can't get a repro for public consumption by EOD and I can try to figure this out staring at code. A quick attempt at a repro worked for me both using 0.0.38 and pants master
  • a

    astonishing-jelly-60479

    07/23/2015, 8:07 PM
    k, working on repro atm
  • a

    ancient-stone-50795

    07/23/2015, 8:30 PM
    Please let me know if there is something I should wait for before a 0.0.39 release tomorrow. I was planning to knock that out first thing in the morning
  • e

    enough-analyst-54434

    07/23/2015, 8:33 PM
    SGTM. I'd love to get the fix for https://github.com/pantsbuild/pex/issues/135 in but it won't happen. Wickman will need to release pex on top of the fix hitting master.
  • a

    astonishing-jelly-60479

    07/23/2015, 8:37 PM
    ooookay, reproducer incoming
  • a

    astonishing-jelly-60479

    07/23/2015, 8:38 PM
    lemme just commit it all
  • e

    enough-analyst-54434

    07/23/2015, 8:38 PM
    Awesome - thank you
  • a

    astonishing-jelly-60479

    07/23/2015, 8:42 PM
    ok, clone https://github.com/wyattanderson/pantstest and run 
    ./pants binary src/python:test
  • a

    astonishing-jelly-60479

    07/23/2015, 8:42 PM
    appreciate all the help so far
  • a

    astonishing-jelly-60479

    07/23/2015, 8:42 PM
    that repo also might have a bunch of garbage (pycs, etc.) committed
  • e

    enough-analyst-54434

    07/23/2015, 8:45 PM
    K - taking a look
  • a

    astonishing-jelly-60479

    07/23/2015, 8:45 PM
    without python-eval loaded it'll build a pex that won't run, with it loaded it'll blow up in that phase (which is nice, really)
  • e

    enough-analyst-54434

    07/23/2015, 9:01 PM
    Yeah - the issue is the thrift python codegen winding up in the root of the pex, should only exist in the embedded wheel in .deps. IE: if you unzip the pex, then rm -rf geocode, then 
    python .
    , it works
1...767778...2420Latest