https://pantsbuild.org/ logo
Powered by
#announce
  • l

    limited-country-20626

    09/13/2017, 12:03 AM
    Yeah, I just threw our hosted ones in the trash
  • r

    rough-minister-58256

    09/13/2017, 12:03 AM
    using that tho in lieu of a travis cache may be the sanest option afaict
  • l

    limited-country-20626

    09/13/2017, 12:03 AM
    I put up the updated release notes - will land + release at the end of the night
  • l

    limited-country-20626

    09/13/2017, 12:04 AM
    talk to you soon sir
  • r

    rough-minister-58256

    09/13/2017, 12:05 AM
    cya
  • l

    limited-country-20626

    09/13/2017, 12:17 AM
    hmmm…I see those SSL failures when pulling binaries. Gah. It is not clear to me that we should release, but I will this channel considers it the best option
  • l

    limited-country-20626

    09/13/2017, 12:18 AM
    <removed deadend conjecture>
  • r

    rough-minister-58256

    09/13/2017, 12:22 AM
    it’s failing locally?
  • r

    rough-minister-58256

    09/13/2017, 12:22 AM
    or in CI you mean
  • l

    limited-country-20626

    09/13/2017, 12:22 AM
    no, just confirming what you said above
  • r

    rough-minister-58256

    09/13/2017, 12:22 AM
    ah
  • l

    limited-country-20626

    09/13/2017, 12:23 AM
    the release is likely fine
  • r

    rough-minister-58256

    09/13/2017, 12:23 AM
    I suspect that issue is some travis environment-specific thing - I never had problems pulling binaries locally
  • r

    rough-minister-58256

    09/13/2017, 12:23 AM
    so maybe fine to just note it, release and move on
  • l

    limited-country-20626

    09/13/2017, 12:23 AM
    I am going to try and downgrade openssl
  • l

    limited-country-20626

    09/13/2017, 12:24 AM
    deprecation landed in 1.1.0
  • l

    limited-country-20626

    09/13/2017, 12:24 AM
    not sure how feasible this will be
  • r

    rough-minister-58256

    09/13/2017, 12:24 AM
    should be possible via transitive pin
  • l

    limited-country-20626

    09/13/2017, 12:25 AM
    hmm…I was thinking in the travis.yaml
  • r

    rough-minister-58256

    09/13/2017, 12:25 AM
    ah
  • l

    limited-country-20626

    09/13/2017, 12:26 AM
    I think that all historic versions of pyopenssl are going to trigger because they conflict with the openssl in the container
  • l

    limited-country-20626

    09/13/2017, 12:26 AM
    I am just now really looking, I was hopeful that pyopenssl would bail me out before
  • r

    rough-minister-58256

    09/13/2017, 12:28 AM
    fwiw, I do not get a deprecation warning w/ this diff applied:
    Copy code
    diff --git a/3rdparty/python/requirements.txt b/3rdparty/python/requirements.txt
index 3371aaf..4af4dfa 100644
--- a/3rdparty/python/requirements.txt
+++ b/3rdparty/python/requirements.txt
@@ -16,11 +16,12 @@ pex==1.2.11
 psutil==4.3.0
 pyflakes==1.1.0
 Pygments==1.4
+pyopenssl==17.1.0
 pystache==0.5.3
 pytest-cov>=2.4,<2.5
 pytest>=3.0.7,<4.0
 pywatchman==1.3.0
-requests==2.5.0,<2.19
+requests[security]==2.5.0,<2.19
 scandir==1.2
 setproctitle==1.1.10
 setuptools==30.0.0
  • r

    rough-minister-58256

    09/13/2017, 12:29 AM
    17.1.0 vs 17.2.0 (latest)
  • r

    rough-minister-58256

    09/13/2017, 12:29 AM
    17.2.0 was where the deprecation warning was added
  • l

    limited-country-20626

    09/13/2017, 12:29 AM
    huh, maybe the deprecation is actually raised from the python-openssl. Okay, will try that real fast omw out
  • r

    rough-minister-58256

    09/13/2017, 12:30 AM
    it definitely is raised in pyopenssl
  • r

    rough-minister-58256

    09/13/2017, 12:31 AM
    but triggered elsewhere, via import w/ deprecation in a base 
    __init__.py
  • l

    limited-country-20626

    09/13/2017, 12:33 AM
    Okay. I was under the impression that their use was the problem. The PR they landed just removed the OpenSSL.rand from their inits
  • l

    limited-country-20626

    09/13/2017, 12:34 AM
    okay, anyway, I don’t doubt you are right
1...555657...206Latest