Yeah, I pushed it into <https://github.com/pantsbu...
# pexa
average-vr-56795
10/19/2018, 5:43 PMYeah, I pushed it into https://github.com/pantsbuild/pex/pull/605
e
enough-analyst-54434
10/19/2018, 5:47 PMSo, that's a bit of a weird pex. It's entrypoint is py.test which - without args, scans cwd for files to test (ie it won't look at the -D in the pex
enough-analyst-54434
10/19/2018, 5:47 PMIs that what you mean to be doing?
enough-analyst-54434
10/19/2018, 5:47 PMWouldn't remote exec not -D, but put sources on fs via snapshot?
enough-analyst-54434
10/19/2018, 5:49 PMIf you want to embed sources to test in the pex you'd need to mark it zip_safe False so it explodes on startup and then make sure PWD was the explode dir (may already be the case but I don't think so)
a
average-vr-56795
10/19/2018, 5:52 PMLet me step back a level
average-vr-56795
10/19/2018, 5:52 PMHow would you write a test that that running an entrypoint which exits non-zero leads to pex exiting non-zero?
e
enough-analyst-54434
10/19/2018, 5:53 PMSuggestion on the PR, but like you're doing - just don't try to make py.test do things it doesn't understand - like run on files inside a pex. Use the pex like a binary and pass it CLI args - the same ones that work with un-pexed pytest
enough-analyst-54434
10/19/2018, 5:55 PMIt's the -D that is left field reading the test - kill it
a
average-vr-56795
10/19/2018, 5:56 PMCool, thanks š
average-vr-56795
10/19/2018, 6:10 PMUgh, still failing š Would you mind taking a quick look? Iām sure itās something trivialā¦ Sorry, Iām about to disappear on holiday!
average-vr-56795
10/19/2018, 6:11 PMOh, only failing on OSX?
average-vr-56795
10/19/2018, 6:11 PMThatāsā¦ Confusingā¦
e
enough-analyst-54434
10/19/2018, 6:12 PMLooking, but OSX has only 2 interesting facts that bite that I know of - realpath (var/tmp) and HFS case wierdness
enough-analyst-54434
10/19/2018, 6:12 PMTime to test locally or print stdout/stderr from subproc call
a
average-vr-56795
10/19/2018, 6:13 PMI ran locally on OSX and it worked :S
average-vr-56795
10/19/2018, 6:13 PMNot with a temp_dir to be fair
e
enough-analyst-54434
10/19/2018, 6:14 PMHigher in log:
Copy code
============================= test session starts ==============================
platform darwin -- Python 2.7.15, pytest-3.9.1, py-1.7.0, pluggy-0.8.0
rootdir: /var/folders/nz/vv4_9tw56nv9k3tkvyszvwg80000gn/T/tmpLg6evn, inifile:
collecting 0 items
========================= no tests ran in 0.01 seconds =========================
ERROR: not found: /var/folders/nz/vv4_9tw56nv9k3tkvyszvwg80000gn/T/tmpLg6evn/tester.py
(no name '/var/folders/nz/vv4_9tw56nv9k3tkvyszvwg80000gn/T/tmpLg6evn/tester.py' in any of [])a
average-vr-56795
10/19/2018, 6:21 PMIs that obviously meaningful to you? :)
e
enough-analyst-54434
10/19/2018, 6:22 PMNope. I'd try combinations of os.path.realpath - and you might try hacking down .travis.yml to 1 shard and
tox -e<ENV> -- -vsk<failing test>a
average-vr-56795
10/19/2018, 6:31 PMrealpath did the job š Thanks for your help!
e
enough-analyst-54434
10/19/2018, 6:31 PMYou're welcome