Log file Viewer
#pypa-dev logs for Monday the 16th of April, 2018
[15:19:41] <sumanah> We know about the redirect issue and are fixing it -- thanks for the reports, all
[15:22:29] <sumanah> redshadowhero: no problem, we're the ones who made the typo on the status page :)
[15:23:12] <redshadowhero> But, am I crazy, or is anyone else having issues with pip doing too many redirects?
[15:23:56] <redshadowhero> There's always that moment when I'm super paranoid that something is really screwed up in my environment/network/setup/whatever.
[15:24:09] <sumanah> http://status.python.org/incidents/1y1f44q6srh2 -- we updated the statuspage so more people know we're working on it :)
[15:25:06] <sumanah> lghampton: can you watch the warehouse, pip, and virtualenv repositories and reply to new issues about the redirect and point people to http://status.python.org/incidents/1y1f44q6srh2 you can subscribe to this incident to get an update when it's fixed ?
[15:25:31] <ardzoht> anyone with this exception message? 'pip._vendor.requests.exceptions.TooManyRedirects: Exceeded 30 redirects.'
[15:25:38] <sumanah> ardzoht: yup! http://status.python.org/incidents/1y1f44q6srh2 you can subscribe to this incident to get an update when it's fixed
[15:27:54] <sumanah> I'm grateful we have so many users who actually report issues instead of thinking it's someone else's problem. :)
[15:28:26] <pmac> I was going to comment on https://github.com/pypa/warehouse/issues/3702 but it's now locked (which makes sense). But I did fix the issue by upgrading pip to 10.0.0
[15:28:59] <sumanah> HUH. INTERESTING. dstufft EWDurbin ^ for at least one user, upgrading to pip 10.0.0 fixes the redirect problem
[15:29:23] <pmac> haven't tried that in my failing CI docker build, but it did fix it for me locally
[15:29:44] <sumanah> pmac: IMO waiting ~10 min would probably be a reasonable step, I know Ernest is working on fixing the prob right now
[15:30:18] <pmac> great. I'm confident it'll be fine. just wanted to add that tidbit incase it's useful. best of luck.
[15:31:45] <sumanah> lghampton: https://github.com/issues?q=user%3Apypa+is%3Aopen is a way to look at the queue of new issues in all pypa projects
[15:32:18] <sumanah> lghampton: it's in newest-first order. If you keep refreshing that you'll see any new stuff that comes in (I think only some of the repos' issues show up via bot in this channel)
[15:33:13] <cdunklau> sumanah: got a different exception report from #python: 503 Server Error: Backend is unhealthy
[15:36:30] <redshadowhero> I just managed to `pip install --upgrade pip` and then my install worked. I have another box that I'll test an arbitrary package in with an older version of pip just for another sanity check.
[15:41:04] <redshadowhero> Well, it installed something last night, and I was able to get speedtest-cli installed. It's probably just what the debian repos had for jessie.
[15:41:43] <sumanah> Hi! We are working on it. http://status.python.org/incidents/1y1f44q6srh2 you can subscribe to this incident to get an update when it's fixed
[15:43:11] <cooperlees> There is always `curl https://bootstrap.pypa.io/get-pip.py -o get-pip.py && python3 get-pip.py`
[15:44:29] <ronny> sumanah: thanks for that link, i didnt correctly identify it on the status page and missed
[15:46:23] <ronny> sumanah: im still seeing tonns of 404's on the files front tho (is there a expectation when things are de-escalating
[15:47:12] <cooperlees> FYI - I saw no errors from my bandersnatch instance from the cut over - Last ran @ ~07:35am pacific
[15:47:28] <sumanah> ronny: if you look at the update on status.python.org right now you see we've fixed one problem and are working on another
[15:48:22] <ardzoht> sumanah: thanks for the updates, we're still getting 404's but is something we should expect for the packages refreshing
[15:48:54] <ronny> sumanah: oh well - we had it coming - too much on pypi is too volunteery (but its a critical piece of infra for many of us)
[15:49:20] <sumanah> ronny: donate.pypi.org is probably live :) thanks for the kind words about the volunteer work
[15:57:23] <sumanah> Today we launch the new PyPI and it's a little bumpy -- this is the first major upgrade to PyPI in more than a decade. You may be interested in this article I wrote about the new PyPI https://lwn.net/SubscriberLink/751458/81b2759e7025d6b9/
[16:00:55] <thea> https://travis-ci.org/pypa/python-packaging-user-guide/builds/367230976?utm_source=github_status&utm_medium=notification
[16:07:29] <ronny> tbh, im glad im not part of that vlunteer team doing the deploy, i cant even imagine the stress
[16:09:35] <sumanah> ronny: it's mostly volunteers most of the time -- in this case we got a limited award from Mozilla which is the reason this is possible https://pyfound.blogspot.com/2017/11/the-psf-awarded-moss-grant-pypi.html
[16:19:16] <thea> newnenw: are you getting download errors or something else? See https://status.python.org
[16:20:23] <newnenw> The last error in the stack is `pip._vendor.requests.packages.urllib3.exceptions.MaxRetryError: HTTPSConnectionPool(host='files.pythonhosted.org', port=443): Max retries exceeded with url: /packages/20/d7/04a0b689d3035143e2ff288f4b9ee4bf6ed80585cc121c90bfd85a1a8c2e/setuptools-39.0.1-py2.py3-none-any.whl (Caused by ResponseError('too many 503 error responses',))`
[16:20:50] <sumanah> newnenw: I think that's still because of the cutover -- we are working on it. thanks for the report
[16:21:22] <sumanah> https://status.python.org/incidents/1y1f44q6srh2 this is the incident you can subscribe to, to get notifications when it's fixed
[16:21:51] <sumanah> newnenw: if you want, you can search your favorite social media site for "pypi" and help us by replying to a few folks with this info :)
[16:22:30] <zzzzx> seeing repeated HTTP 503 for https://files.pythonhosted.org/packages/22/66/9dcf61c1756a0de1ac86547a0fdb8c8641e345049cbf6f18c2ad6dab69d0/botocore-1.10.4-py2.py3-none-any.whl
[16:22:48] <sumanah> thanks for the report zzzzx -- https://status.python.org/incidents/1y1f44q6srh2 this is the incident you can subscribe to, to get notifications when it's fixed
[16:23:35] <JosefAssad> We're actually going to look at deploying our own warehouse instances (perhaps)
[16:23:35] <apollo13> haha, just clicked the subscribe button -- subscribe via text message; wth would I wanna do that :D
[16:24:04] <Yhg1s> JosefAssad: you probably dont' want all the Warehouse bells, whistles and weight.
[16:26:05] <JosefAssad> got it :) It's still pretty far away in our roadmap, but the tips are appreciated
[16:45:18] <moldy> since today, pipenv is complaining about hash mismatches. i guess this is related to the pypi change? any hints on how to best resolve this?
[16:50:42] <newnenw> Thanks for your hard work in fixing the funky 503 errors. I can confirm I'm able to `setup.py install` again :D
[16:55:25] <di_codes> moldy: looks like that project added more distributions recently: <https://pypi.org/project/pluggy/#files>
[16:57:00] <moldy> here's the faulty Pipfile.lock just in case it helps anyone: https://dpaste.de/RbxQ
[16:59:35] <techalchemy> will have a bugfix release of pipenv out today, not sure what all is broken from the downtime
[17:14:31] <sumanah> lghampton: please take the text of https://blog.python.org/2018/04/new-pypi-launched-legacy-pypi-shutting.html and send both the text of the post, and a link to the post, to the python-announce email list https://mail.python.org/mailman/listinfo/python-announce-list
[17:15:05] <sumanah> lghampton: AND, please also take the text of https://blog.python.org/2018/04/new-pypi-launched-legacy-pypi-shutting.html and send both the text of the post, and a link to the post, to the pypa-dev list and the distutils-sig list
[17:17:57] <lghampton> sumanah: for this annoucement: https://blog.python.org/2018/04/new-pypi-launched-legacy-pypi-shutting.html
[17:22:55] <raydeo> should https://status.python.org/ update pypi.python.org to legacy.python.org now or is it going to be alive for only a couple weeks and not worth updating everywhere?
[17:24:19] <EWDurbin> raydeo: pypi.python.org will live on as a hostname for quite some time and may be independently broken.
[17:28:18] <techalchemy> EWDurbin: hwo urgently do I need to make sure pipenv points everyone to the new hostname?
[17:28:49] <EWDurbin> techalchemy... like medium priority? we won't be a be able to kill that hostname for some time
[17:29:12] <techalchemy> ok, hopefully we can get pip10 integrated in a shorter timeline than 'years' lol
[17:29:50] <techalchemy> EWDurbin: regardin the json api on the new warehouse, does it still have the same caveats around dependency listings as the old one? what about for specific version endpoints
[17:30:21] <techalchemy> I think you mentioned the old json api only listed the dependencies of the most recently uploaded version
[17:46:23] <sumanah> glghampton: ok. now please Submit https://blog.python.org/2018/04/new-pypi-launched-legacy-pypi-shutting.html news to The Changelog https://changelog.com/news/submit
[17:48:06] <sumanah> thanks for the mailing list posts, the PyPUG PR, and the GitHub replies lghampton
[18:11:14] <thea> davisagli: it's not listed on status.python.org, do you mind filing an issue over at https://github.com/pypa/warehouse/issues ?
[18:14:44] <sumanah> thea: do you think that we ought to do some SEO or something to make packaging.python.org show up higher in search results?
[18:15:20] <thea> possibly. The biggest thing we can do it ensure that other highly ranked pages link to it
[18:15:50] <sumanah> thea: so, same for https://mail.python.org/pipermail/python-announce-list/ and https://mail.python.org/mm3/archives/list/pypi-announce@python.org/ ?
[18:17:19] <thea> sumanah: actually there's several things we can do to pypug to improve how search engines see it. Can you file a general SEO bug on it? I'll try to take a look on Friday.
[18:23:03] <sumanah> dstufft: hey, if we have a Warehouse phone call tomorrow at 12:30pm or 1pm ET, could you join that?
[18:27:18] <dstufft> sumanah: uhhh, I have a drs appointment thing at 11:30 tomorrow, probably can't make those sorry
[20:08:02] <techalchemy> hmmm pip no longer supports installing VCS dependencies without passing -e?
[21:30:06] <techalchemy> pf_moore: I assume you're busy but any idea how to get appveyor to move something if it gets stuck in 'queued' mode w/ nothing else pending?
[21:30:52] <pf_moore> techalchemy: No, not really, I generally just wait. I think there's some sort of quota involved...
[21:48:44] <techalchemy> pf_moore: btw I think someone mentioned you might be interested in this: https://github.com/techalchemy/pythonfinder
[21:50:00] <pf_moore> techalchemy: wow, yeah, I've been meaning to do something like that for ages. I'll definitely take a look at it :-)
[21:50:20] <techalchemy> we needed it for pipenv (we aren't actually using it yet but that's why I wrote it)
[21:51:45] <pf_moore> techalchemy: Yeah, I could see it being used in tox, virtualenv, nox, ... Having a uniform way of referring to Python versions would be awesome.
[21:52:47] <techalchemy> pf_moore: for sure, our code was just getting really messy/disjointed with regard to finding python in general, cleaning up the API just made a lot of sense because this is how we use it currently
[21:53:19] <techalchemy> I just assume that if you say '3' you mean 'the highest released version of python 3 installed on the system' so it would return 3.6.5 but not 3.7.0b3
[23:27:51] <techalchemy> pf_moore: can you cancel our pending appveyor builds or something? idk what's going on with them but they're just sitting around