PMXBOT Log file Viewer

Help | Karma | Search:

#pypa logs for Saturday the 5th of November, 2016

(Back to #pypa overview) (Back to channel listing) (Animate logs)
[10:48:35] <AlecTaylor> hi
[12:04:06] <apollo13> dstufft: around?
[12:04:34] <apollo13> dstufft: https://gist.github.com/apollo13/ac53e268d23b4e31b6406817ba96e3eb do you have any idea what happened here? "Successfully installed pip-6.1.1" -- wait what :D
[12:27:58] <AlecTaylor> How do I force virtualenv pip to be used? - http://stackoverflow.com/q/40438089
[13:52:07] <agronholm> why does twine still upload to the old url? doing so doesn't update Warehouse's index
[15:58:07] <dstufft> apollo13: probably it didn't actually install 6.1.1, sometimes pip gets confused in that message
[15:58:16] <dstufft> agronholm: latest twine uses warehouse not legacy
[15:58:25] <dstufft> unless you have the legacy url in your ~/.pypirc
[18:33:17] <agronholm> dstufft: let me check the logs for the twine version
[18:34:59] <agronholm> dstufft: so twine 1.8.1 should upload to pypi.io?
[18:36:51] <agronholm> or rather upload.pypi.org
[18:37:21] <agronholm> because travis seems to install twine 1.8.1 and it's uploading to the legacy url: https://travis-ci.org/asphalt-framework/asphalt-mongodb/builds/173479583
[18:37:41] <agronholm> I guess now I just have to check if travis sets up a .pypirc
[18:37:46] <agronholm> with an explicit url that is
[21:30:44] <ronny> agronholm: the travis pypi deploy is simply just bad ^^
[21:31:05] <agronholm> ronny: can you elaborate on that?
[21:31:33] <agronholm> it lets you choose the url to upload to, I'm just wondering where it gets its default value
[21:31:43] <agronholm> other than that problem, it's been working fine for me
[21:32:55] <ronny> agronholm: it reports no errors, and unless you limit it, it simply runs the deploy on each job, its quite a pain
[21:34:12] <agronholm> why is that a problem?
[21:34:22] <agronholm> running deploy on each job that is
[21:37:20] <agronholm> ronny: would you prefer for the build to fail if deploy fails on one of the jobs?
[21:42:06] <ronny> agronholm: well, it hides actual releasing errors
[21:42:24] <ronny> so you run a build, and get no release and no failure for example
[21:42:33] <ronny> which is pretty much shitty
[21:42:36] <agronholm> well, the failures are visible in the logs
[21:42:40] <ronny> or missing release files
[21:42:42] <agronholm> but would you like the build to fail?
[21:42:47] <agronholm> if deploy fails
[21:42:52] <ronny> thats not the point, it *breaks* and reports *im fine*
[21:43:11] <agronholm> so...an email to your account about it?
[21:44:10] <ronny> agronholm: well, it would report that, if it reported actual failures as failures
[21:44:47] <agronholm> a few straight answers would be great
[21:45:06] <agronholm> how exactly do you want it to report deploy failures (assuming the build itself worked)?
[21:46:53] <agronholm> should it mark the particular build as failed where deploy failed?
[21:47:13] <agronholm> or should there just be a different marker in the travis web ui?
[21:51:26] <agronholm> I do admit that travis's deploy could be improved