Log file Viewer

[00:00:40] <sumanah> (also, parenthetically, right now I am watching a web-broadcast conversation about the making of "2001: A Space Odyssey" that includes how Clarke & Kubrick were consulting with IBM and Marvin Minsky about what computers would be capable of)

[00:04:06] <sumanah> (I believe their predictions did not include "helping strangers share with your their frank and unvarnished opinions about the software you are responsible for")

[00:15:39] <dstufft> sumanah: I think yanking should be fine? it's a pretty simple feature all said and done

[00:16:40] <dstufft> computers continue to surprise me with new and creative ways to fail though

[01:44:29] <sumanah> I predict the feature on PyPI isn't going to be so much the issue - I predict that clients not implementing yanking support will cause confusion that will lead to support tickets for us

[01:45:01] <sumanah> and so I figure a slower rollout will help us see those trickle in and then reach out to the developers of those tools, put out messaging in the right places, etc.

[02:15:34] <sumanah> techalchemy: hey - I am gonna head to bed soon. I recommend that you leave _some_ kind of update on the pipenv release tracking issue because being silent as the estimated delivery date passes is hard on credibiity.

[02:16:27] <techalchemy> ^ yeah super exhausted atm, will try

[02:16:54] <techalchemy> so far i can say that azure is terrible

[02:17:18] <sumanah> Thanks

[02:17:42] <techalchemy> + some other thing broke but is almost fixed, then should be ok

[03:48:42] <toad_polo> sumanah: And mirrors!

[13:28:26] <PSFSlack1> <deveshkusingh> Github down today again!

[16:18:59] <alanbato> Hello everyone :)

[18:01:43] <sumanah> toad_polo: I saw in the logs you said "And mirrors!" but I could not work out what you were replying to :-)

[18:02:11] <toad_polo> sumanah: You said that clients would be the problem with yanking support.

[18:02:15] <toad_polo> I was saying clients and mirrors.

[18:02:40] <sumanah> Ah! I did not mean to exclude other possible causes of problems and I am glad you have mentioned one toad_polo

[18:04:16] <toad_polo> Yeah, I didn't see it as excluding the possibility, just bringing up that those are the two big sources of support tickets we see when people start relying on a "new" feature.

[18:04:34] <toad_polo> People using incompatible clients (mostly old versions of `pip`) and people configured to hit a mirror like artifactory or piwheels that doesn't support the feature.

[18:05:20] <sumanah> toad_polo: ooooh Artifactory, now I'm both reminded and a bit steamed

[18:05:32] <sumanah> of how this issue plays out

[18:06:33] <toad_polo> Yeah, I wonder if we should proactively cultivate contacts with artifactory, devpi, piwheels and the other likely mirrors.

[18:06:52] <toad_polo> Maybe they're on it and maybe (especially in the case of Aritfactory, in my experience) they just don't care.

[18:07:26] <sumanah> I have contacted devpi already about this beta but haven't contacted the other few. Piwheels I believe heard about it via Twitter already. I need to ping Artifactory. toad_polo I welcome suggestions on additional platforms/groups to ping

[18:07:32] <sumanah> this is useful - thanks

[18:09:00] <toad_polo> Yeah, I CC-ed one of the piwheels maintainers on a twitter thread.

[18:09:13] <sumanah> ah that was you! yes! Thank you

[18:09:44] <sumanah> oh and I'm actually getting a bit mixed up at the moment and confusing the yanking feature with the pip beta - I want to publicize both but the latter is my job and the former is just a good idea

[18:09:59] <sumanah> ("confusing" in the sense that they are on my mind as things to tell the world about)

[18:54:08] <pradyunsg> FWIW, pip has had support for PEP 572 (yanked releases) since pip 19.2.

[18:57:03] <sumanah> right - yeah

[18:57:37] <pradyunsg> https://pip.pypa.io/en/stable/news/#id107

[18:58:13] <pradyunsg> Ah, it's PEP 592. :)

[19:02:09] <dhellmann> pradyunsg : https://github.com/pypa/pip/pull/8119 is a first stab at addressing that bug about overwriting files.

[19:04:46] <dalley> apologies if this has already been discussed, but are there any known issues with caching on PyPI at the moment?

[19:05:33] <PSFSlack1> <di> dalley: what are you experiencing?

[19:06:09] <dalley> PSFSlack1, for our package, according to the main page & the releases page, the latest version is 3.2.0 https://pypi.org/project/pulp-rpm/

[19:07:03] <dalley> we pushed 3.3.0 a few hours ago https://pypi.org/project/pulp-rpm/3.3.0/, and you can download it fine from PyPI, and the direct link to the release works, but otherwise you could never tell it exists

[19:08:05] <pradyunsg> dalley: I see 3.3.0 on my end.

[19:09:31] <dalley> pradyunsg, I just took these screenshots after refreshing the pages, seconds ago https://imgur.com/a/IIPCvxo

[19:09:42] <dalley> others on my team see the same thing

[19:10:01] <dalley> keep in mind, 3.3.0b1 isn't what I'm referring to

[19:11:13] <dalley> do you see the actual 3.3.0 release even on the primary landing page (first link)?

[19:11:20] <PSFSlack1> <deveshkusingh> I can see 3.3.0 as the first entry in https://pypi.org/project/pulp-rpm/#history

[19:12:21] <PSFSlack1> <deveshkusingh> And the landing page `https://pypi.org/project/pulp-rpm/` also lists `pulp-rpm 3.3.0`

[19:13:57] <dalley> I don't really know what to say then, I even just tried multiple browsers. Does PyPI have load balancers or static caches or something where one might possibly be serving old content?

[19:15:21] <pradyunsg> dalley: PyPI has a Fastly CDN in front of it -- and seems like one of their POP locations have degraded performance. https://status.fastly.com/

[19:16:38] <pradyunsg> dalley: Would you happen to know if you're hitting that POP?

[19:16:53] <dalley> I'm in NC, so it is plausible that a bad CDN in Virginia would be the cause

[19:17:31] <dalley> pradyunsg, I'm not sure how I would check though

[19:18:06] <pradyunsg> dalley: me neither. :)

[19:18:11] <pradyunsg> others here are probably do though.

[19:21:10] <dalley> I asked my coworker to VPN into another office's network and see if they still get the bad pages

[19:23:24] <PSFSlack1> <di> Running `curl -sI pypi.org/ | grep X-Served-By` should tell you what POP you're using

[19:29:51] <pradyunsg> dalley: ^

[19:30:27] <dalley> X-Served-By: cache-wdc5524-WDC

[19:31:27] <dalley> pradyunsg, although the page seems to have started working in the past 10 minutes xD it was broken for 3 hours though

[19:32:23] <pradyunsg> dalley: well, caching in distributed system is... tricky. :)

[19:32:28] <PSFSlack1> <di> That's probably because I manually purged it. It's possible purging isn't working right now. We'll keep an eye out for more reports

[19:33:06] <pradyunsg> @di: curl -X PURGE?

[19:33:17] <PSFSlack1> <di> pradyunsg: yeah

[19:33:26] <pradyunsg> :)

[19:33:50] <pradyunsg> I was gonna do that, but didn't wanna mess w/ someone if they were debugging. :)

[19:38:55] <pradyunsg> thanks dalley for the reaching out and letting us know! :)

[19:39:05] <dalley> np

[19:39:08] <dalley> thank you!

[21:49:31] <sumanah> techalchemy: heading off for the evening. Wishing you well

[21:49:47] <techalchemy> thanks sumanah

[21:49:55] <sumanah> thanks for the issue update techalchemy

[21:49:57] <techalchemy> pretty close i think

[21:50:06] <sumanah> :-)

[21:50:15] <techalchemy> also should be able to swap to github tests

[21:50:16] <techalchemy> :D

[21:50:23] <techalchemy> thanks again for everything

[21:50:34] <sumanah> :) :)

[22:53:09] <travis-ci> pypa/pip#15850 (master - daff811 : Pradyun Gedam): The build was fixed.

[22:53:09] <travis-ci> Change view : https://github.com/pypa/pip/compare/c2443143185204f916e0af919fb743cdfde68d33...daff81124ab76138da323a5173b120dd88190f56

[22:53:09] <travis-ci> Build details : https://travis-ci.org/pypa/pip/builds/678807084