pmxbot IRC Log Viewer

[16:00:02] <sumanah> Hi cooperlees!

[16:00:09] <cooperlees> Morning

[16:00:10] <sumanah> Welcome to the Warehouse/PyPI livechat, the first of four during this beta!

[16:00:19] <sumanah> Warehouse, the codebase behind the new PyPI, is available as a beta side https://pypi.org, and I'm one of the people working on it. (I'm Sumana Harihareswara, project manager.)

[16:00:46] <sumanah> first some context, then introductions! :)

[16:00:50] <sumanah> We're making steady progress on the developer roadmap https://wiki.python.org/psf/WarehouseRoadmap thanks to funding from Mozilla's Open Source Support Program https://pyfound.blogspot.com/2017/11/the-psf-awarded-moss-grant-pypi.html

[16:01:11] <sumanah> And it's on its way to fully replacing the legacy PyPI site pypi.python.org next month.

[16:01:28] <sumanah> We're currently seeking feedback from you about what does or doesn't work for you in the new interface. https://pyfound.blogspot.com/2018/03/warehouse-all-new-pypi-is-now-in-beta.html#workflows has a list of stuff we especially want people to test.

[16:01:49] <sumanah> And, as a reminder, participating in this livechat means you agree to abide by the PyPA code of conduct: https://www.pypa.io/en/latest/code-of-conduct/

[16:02:34] <sumanah> The main Warehouse developers we have in the channel today are me, EWDurbin, and dstufft I believe

[16:02:45] <sumanah> EWDurbin you wanna introduce yourself? and dstufft if you are around :)

[16:03:14] <EWDurbin> I'm Ernest! I wear a few hats in the Python community, but the one i've got on right now is as the lead on PyPI's infrastructure and a contributor to the pypa/warehouse codebase.

[16:03:34] <cooperlees> EWDurbin: Cheers for the bandersnatch fixes :D

[16:03:47] <sumanah> ok! if you're here for the livechat today, introduce yourself? cooperlees you can go first :)

[16:04:29] <EWDurbin> cooperlees: no problem! when do you think we'll see a release? i also had a question about rehashing every file on update... seems to be troublesome for projects like mxnet-* which have hundreds of large files.

[16:05:13] <cooperlees> Fb employee who based in California. Fell into maintaining bandersnatch as I Python 3'd it as it hurt me to run it internally under Python 2. Want to polish it up a bit with warehouse etc ... Currently working along side ambv in our core Python team here @ Facebook

[16:05:41] <cooperlees> EWDurbin: I can try release today. Just time. I've been travelling and trying to catch up @ work + sick etc.

[16:05:45] <EWDurbin> thanks for picking up bandersnatch cooperlees! we've been using it internally for PyPI for some time!

[16:05:56] <cooperlees> PyPI uses it internally ?

[16:06:05] <EWDurbin> yes! we run an internal mirror as fallback

[16:06:13] <cooperlees> haha :D

[16:06:15] <cooperlees> So meta

[16:06:16] <EWDurbin> if S3 is cranky, or soon if the backends are cranky!

[16:06:26] <EWDurbin> so the main reason i'm excited is having a fallback for /json

[16:06:29] <EWDurbin> thanks for that feature as well :)

[16:06:51] <cooperlees> haha :) I use it for my internal pip wrapper system of death to build wheels for Facebook main Python repo.

[16:07:06] <cooperlees> It's called PyFI

[16:07:09] <sumanah> hi edd_lc - if you're here for the PyPI livechat, welcome

[16:07:10] <sumanah> !logs

[16:07:10] <pmxbot> http://kafka.dcpython.org/channel/pypa-dev

[16:07:22] <sumanah> ^ you can see the logs of the start of the chat :)

[16:07:26] <edd_lc> sumanah: Cheers

[16:07:50] <EWDurbin> cooperlees: may i PM regarding the hashing thing? i don't want to muddy the water here and had a couple questions

[16:07:58] <cooperlees> Please

[16:08:07] <sumanah> EWDurbin: as long as it's pretty quiet here I'm happy to have that happen here in channel

[16:08:15] <sumanah> I figure this first one will be the quietest

[16:08:19] <EWDurbin> let's start here then :-D

[16:08:32] <sumanah> and I would like to overhear stuff so it can feed into https://wiki.python.org/psf/PackagingSprints

[16:08:49] <EWDurbin> so cooperlees whenever a project updates, bandersnatch currently double checks _every_ file on disk to make sure that they are 1) there and 2) match the expected sha

[16:09:01] <EWDurbin> and also checks new files at download time to ensure they match expected sha

[16:09:04] <EWDurbin> this is _great_!

[16:09:10] <cooperlees> EWDurbin: note, all this design is BC (Before Cooper)

[16:09:32] <EWDurbin> no judgement, i love it :-p

[16:09:44] <cooperlees> Christine is very perfectionist :)

[16:09:48] <EWDurbin> but it's troublesome as package size limits have been slowly raised

[16:09:52] <cooperlees> *Christian even

[16:09:57] <edd_lc> \me wants to overhear, too.

[16:10:17] <EWDurbin> so when a project with hundreds of packages... often large does a release

[16:10:25] <sumanah> edd_lc: :) would you introduce yourself?

[16:10:31] <EWDurbin> https://usercontent.irccloud-cdn.com/file/veSDh8FW/Screen%20Shot%202018-03-27%20at%2012.10.21%20PM.png

[16:10:40] <EWDurbin> that's a minutely "status check" of the internal mirror

[16:11:00] <EWDurbin> the lag is caused by bandersnatch spending almost 10 minutes rehashing all existing _large_ files.

[16:11:15] <EWDurbin> now, PyPI doesn't allow for packages or their SHA to change... it did at one point.

[16:11:40] <edd_lc> Gladly. Name's Edward. I work for CyberBit as a backend developer and teach basic security 2 times a week at a local tech school.

[16:11:51] <EWDurbin> so i'm curious what we could do to help a mirror stay in sync _quickly_ without sacrificing statelessness.

[16:11:55] <cooperlees> Ok - I more don't trust Gluster internally where I store all the package files :P

[16:12:04] <cooperlees> So I personally don't mind the re-check

[16:12:14] <sumanah> Welcome :) thanks edd_lc for joining. And please do mention if there's anything you particularly find nice or any snags you're finding with pypi.org.

[16:12:18] <cooperlees> But, we could make a configuration option to only sha new files?

[16:12:30] <edd_lc> sumanah: Cheers

[16:13:05] <EWDurbin> seems reasonable, as clients _should_ be checking shas written to the simple pages

[16:13:11] <EWDurbin> okay, second and more fun one cooperlees!

[16:13:35] <EWDurbin> last night we shipped something in /json that might interest you

[16:13:52] <EWDurbin> the JSON api now reports last_serial

[16:13:54] <cooperlees> I also need to fix the deletion code

[16:14:00] <cooperlees> YES

[16:14:05] <cooperlees> I want to get rid of xml2rpc

[16:14:07] <EWDurbin> which means i _think_ a mirror can mirror a mirror now

[16:14:10] <cooperlees> *xmlrpc2

[16:14:24] <EWDurbin> but it's somewhat of an unsafe option.

[16:14:58] <EWDurbin> oh, so the changelog isn't exposed in JSON

[16:15:09] <EWDurbin> that's still an XMLRPC call :(

[16:15:17] <cooperlees> Ya, That's hopefully whaat I'lld do @ PyCON :)

[16:15:21] <cooperlees> *I'll

[16:16:26] <cooperlees> I install docker etc. to try and have a look at how it all works, but then saw twisted and killed it all with fire as I have nightmares still about twisted :P

[16:16:31] <cooperlees> *installed

[16:17:11] <dstufft> twisted is just a web server for us, so you can ignore the fact it exsits

[16:17:34] <sumanah> also cooperlees maybe you could rephrase that? there are probably folks in here who work on Twisted and maybe you could say that more sensitively?

[16:19:52] <cooperlees> Twisted is good, don't get me wrong. I just can't think that way. I mean no harm in my comment.

[16:20:10] <sumanah> thanks cooperlees

[16:20:16] <sumanah> so intros:

[16:20:32] <sumanah> dstufft is Donald Stufft, who has been one of the key maintainers of Warehouse and who started the project

[16:21:10] <cooperlees> sumanah: It's like you knew dstufft wouldn't intro or something :P

[16:22:02] <sumanah> bandersnatch question for cooperlees

[16:22:10] <sumanah> bitbucket - you like staying there?

[16:22:20] <sumanah> I want to check, that's all

[16:23:06] <cooperlees> sumanah: I have no ties there. Christian (owner) hates how everything is on GitHub and how they basically own Open Source and could kill the ecosystem tomorrow

[16:23:22] <sumanah> cooperlees: *ah*, yeah. I can understand that. I keep a lot of my personal repositories on gitlab

[16:23:29] <EWDurbin> i like working on bandersnatch, keeps my mercurial skills sharp :-D

[16:23:35] <cooperlees> He did basically say in an email tho I can do what ever is best for Bandersnatch tho - So we could move it to github if everything thinks it's worth while

[16:23:37] <sumanah> I ought to learn Mercurial

[16:23:43] <cooperlees> EWDurbin: haha - I like that too

[16:23:51] <cooperlees> FB is majority hg internally

[16:23:55] <EWDurbin> TIL

[16:24:02] <sumanah> cooperlees: thanks for the context!

[16:24:22] <cooperlees> sumanah: no worries

[16:24:24] <sumanah> cooperlees: I think I may have missed: where is the bandersnatch public roadmap?

[16:25:44] <dstufft> Facebook has made a bunch of improvements to Hg I think for large scale use cases, it's pretty cool

[16:26:09] <dstufft> (Also hi, I'm only half way here)

[16:27:42] <sumanah> cooperlees: EWDurbin: go ahead and talk about anything else you need to sort out regarding bandersnatch, I didn't mean to interrupt

[16:28:05] <cooperlees> sumanah: I don't have a roadmap :\

[16:28:17] <cooperlees> Just issues / incidents or what ever bit bucket calls em

[16:28:24] <EWDurbin> i got no blockers, but i would like to let cooperlees know that bandersnatch at it's current default/Head is fully pypi.org compat

[16:28:30] <cooperlees> (This is my first time owning a open source project)

[16:28:33] <EWDurbin> i know this cause our new internal mirror is pointing at pypi.org :)

[16:28:57] <sumanah> cooperlees: hey I hear ya. I helped Warehouse develop its roadmap & update it

[16:29:01] <cooperlees> Sweet - With sha256 turned on? I need to pull the latest version internally and hit pypi.org too

[16:29:55] <cooperlees> sumanah: basically I want to fully asyncio and fix the deletion code for 3.0

[16:30:08] <cooperlees> remove xmlrpc2 if we can

[16:30:13] <cooperlees> Or I'll async that

[16:30:20] <dstufft> *pours one out for xmlrpc2*

[16:30:26] <sumanah> oh I need to read https://bitbucket.org/pypa/bandersnatch/issues/94/package-files-are-never-purged which has a bunch of history in it

[16:30:42] <cooperlees> dstufft: Translate that for a dumb Aussie please :\

[16:31:08] <cooperlees> sumanah: Summary - bandersnatch still expects a super old PyPI file layout ... so today, we never delete anything

[16:31:38] <dstufft> cooperlees: https://www.urbandictionary.com/define.php?term=pour%20one%20out

[16:32:02] <cooperlees> dstufft: Cheers. Yeah, it's served it's time.

[16:32:07] <cooperlees> gg xmlrpc2

[16:32:26] <cooperlees> I've had a bitch of a time trying to get it's travis jobs to pass for 3.6 and 3.7 ... I can't repro locally etc.

[16:32:35] <cooperlees> O, sorry if I can't say that

[16:32:43] <cooperlees> pretend that says bad

[16:32:45] <cooperlees> :O

[16:33:01] <sumanah> cooperlees: thanks for the apology

[16:33:39] <cooperlees> dstufft: Thoughts on https://github.com/dstufft/xmlrpc2/pull/3

[16:33:51] <sumanah> cooperlees: throwing what you just said about asyncio, etc. (plus from your emails that I saw) into a ROADMAP.txt file and putting it in the root of the bandersnatch directory would be worthwhile IMO

[16:34:11] <cooperlees> Rightio - PRs welcome :P

[16:34:13] <dstufft> cooperlees: want to own xmlrpc2? :)

[16:34:19] <cooperlees> Nope

[16:34:22] <cooperlees> lol

[16:34:27] <dstufft> I have not thought about it in like

[16:34:28] <dstufft> years

[16:34:39] <cooperlees> I can help with it sure

[16:34:47] <dstufft> last commit was uh

[16:34:48] <dstufft> Dec 15, 2012

[16:34:54] <dstufft> that was probably the last time I thought about it

[16:34:57] <cooperlees> This is going to lead to a big night in Clevland :P

[16:35:24] <cooperlees> dstufft: You confirmed you're coming yet?

[16:35:36] <EWDurbin> dstufft: BOOK YER TRAVEL KTHNXBYE

[16:35:48] <cooperlees> EWDurbin++

[16:36:10] <dstufft> cooperlees: I possess a ticket in my name via work, I don't have travel or like a hotel booked yet

[16:36:12] <dstufft> need to do that

[16:36:31] <dstufft> else I'm going to sneak into EWDurbin's house and sleep on his couch :P

[16:36:37] <sumanah> dstufft: I believe that couch is already booked

[16:36:43] <cooperlees> LoL - If aws is a little tight I'll change my room to a two room suite

[16:36:49] <cooperlees> I know cash flow is a problem there

[16:36:54] <dstufft> nah they'll pay for it np

[16:36:57] <sumanah> cooperlees: I think that is a burn! and a pretty funny one.

[16:37:06] <cooperlees> dstufft: I know :P

[16:37:28] <sumanah> (do Australians use "burn" to talk about an insult that hits its mark? or is it an Americanism?)

[16:37:32] <dstufft> I am just lazy :(

[16:37:39] <dstufft> or bad at prioritizing

[16:37:40] <dstufft> or something

[16:38:05] <cooperlees> Carl Meyer says you're a ghost

[16:38:22] <dstufft> there are a handful of people who have managed to successfully meet me

[16:38:31] <sumanah> AnyTimeTraveler: welcome. You here for the IRC livechat about PyPI beta?

[16:38:33] <dstufft> or at least the paid actor who plays me

[16:39:00] <sumanah> cooperlees: Facebook-related heads-up: since the Mozilla grant is going to run out probably in late April or early May, we'd love money to keep the team going.

[16:39:35] <cooperlees> haha - Do you have calculations on how much to finish etc. ?

[16:39:40] <cooperlees> I can ask people

[16:39:59] <cooperlees> We send pretty solid checks each year to PSF

[16:40:01] <sumanah> cooperlees: If we had about $150K we could do a really quite large chunk of security work + internationalization + accessibility work

[16:40:05] <EWDurbin> cooperlees: if you want to procrastinate some more, i'd love to get a PR in for "dangerously_skip_hashing_the_world_ or similar

[16:40:11] <EWDurbin> before next bandersnatch release

[16:40:33] <cooperlees> EWDurbin: I can suck at releasing for you :D

[16:40:39] <sumanah> cooperlees: thanks! I saw Facebook is a PyCon sponsor right now (but did not see it on the PSF sponsors page -- if I'm working on obsolete info, sorry)

[16:40:49] <EWDurbin> sweet. look for a PR around like 7am eastern tomorrow :-D

[16:41:08] <sumanah> cooperlees: and we're almost certainly applying for https://research.fb.com/programs/research-awards/proposals/secure-the-internet-grants/ in case you can put in a word for us

[16:41:22] <cooperlees> sumanah: Last two year's we've hosted and sponsored getting the top 20 core devs together for a week on campus too :)

[16:41:38] <sumanah> cooperlees: Much thanks!

[16:41:43] <cooperlees> That's all Lukasz Langa's work tho

[16:41:44] <cooperlees> Not me

[16:42:06] <cooperlees> Then we just get to listen to Guido say no for a week

[16:42:25] <cooperlees> (that was a joke everyong)

[16:42:30] <cooperlees> *everyone

[16:42:38] <cooperlees> He's great 1:1

[16:42:40] <sumanah> hi waseem18 & thanks for your work on the trove classifier https://github.com/pypa/warehouse/pull/3273 -- lots of iterative improvement :)

[16:42:51] <waseem18> Your welcome sumanah :)

[16:42:55] <sumanah> cooperlees: yeah! I've experienced that too

[16:43:00] <sumanah> !logs

[16:43:00] <pmxbot> http://kafka.dcpython.org/channel/pypa-dev

[16:43:41] <sumanah> waseem18: ^ in case you want to catch up on the bandersnatch conversation -- cooperlees is interested in getting PRs to improve bandersnatch, a PyPI mirroring system that Facebook and other big orgs use, and that PyPI uses as its fallback in case of certain outages

[16:44:01] <EWDurbin> anyone that's here for the office hours experienced TLS issues? this is one of our largest blockers to making the switch.

[16:44:29] <EWDurbin> we're currently performing rolling brownouts of TLSv1.0 and TLSv1.1 https://status.python.org/incidents/hdx7w97m5hr8

[16:44:42] <waseem18> Good to hear sumanah

[16:45:23] <waseem18> I've heard of Bandersnatch but never actually looked into it

[16:47:24] <cooperlees> sumanah: I'm writing email to my VP to kick off sponsorship opportunites

[16:47:59] <EWDurbin> tell them we don't _need_ it, but one of those huge novelty checks would be really cool :-p

[16:48:00] <sumanah> cooperlees: in case you want advice (I know you didn't actually ask) about running your first open source project: https://www.harihareswara.net/sumana/2016/08/04/1 is my short list of advice types you might need, and https://www.harihareswara.net/sumana/2015/08/09/0 is on how to improve bus factor in the project

[16:48:02] <cooperlees> EWDurbin: Thanks for telling me bandersnatch is used instally. Made my day :)

[16:48:09] <EWDurbin> <3

[16:48:17] <sumanah> cooperlees: thank you!

[16:48:27] <dstufft> ooo

[16:48:31] <dstufft> I want a big novelty check

[16:48:32] <sumanah> cooperlees: and https://changeset.nyc/resources.html has more resources in case you want any lunchtime reading sometime soon

[16:48:43] <dstufft> I wonder if I can convince AWS to send me one

[16:49:10] <cooperlees> dstufft: Send you one of what?

[16:49:18] <cooperlees> O

[16:49:18] <cooperlees> lol

[16:49:21] <dstufft> cooperlees: a big novelty check!

[16:49:23] <cooperlees> Do great things then

[16:49:36] <cooperlees> dstufft: Want to co sponsor?

[16:49:40] <cooperlees> lol

[16:49:48] <sumanah> look I think that if it is mission critical for morale that every member of the MOSS team get a big novelty check then I can start putting that in each milestone

[16:49:54] <cooperlees> Philly is cheap anyways :P

[16:49:58] <sumanah> I'm sure Mozilla will understand

[16:50:12] <dstufft> cooperlees: lol

[16:50:38] <cooperlees> dstufft: Any you haz super bowl now

[16:50:42] <cooperlees> And a great Aussie basketballer

[16:50:44] <cooperlees> You have it all

[16:50:55] <dstufft> I don't watch the eagles, I'm a transplant

[16:52:10] <sumanah> cooperlees: as long as I have you here, a few more questions

[16:52:27] <cooperlees> shoot

[16:52:51] <sumanah> cooperlees: do you know of any other packaging/distro people who will be at the PyCon sprints who are not already in the list at https://wiki.python.org/psf/PackagingSprints ? or at EuroPython?

[16:54:32] <cooperlees> I know Lukasz Langa and Jason Fried will be there for part of the time. Lukasz will be doing core related stuff but not sure what my team mate Jason is planning to work on.

[16:54:47] <cooperlees> I'll ask him if he graces me with his presence soon

[16:54:53] <sumanah> cooperlees: Thanks :)

[16:55:12] <sumanah> cooperlees: https://mail.python.org/pipermail/python-list/2018-March/732138.html would you mind forwarding this around to the internal-to-Facebook Python world? we really do want as much testing as possible during the beta

[16:55:48] <cooperlees> All of Facebook hits my internal mirror :P

[16:55:59] <EWDurbin> btw, i don't miss GlusterFS

[16:56:06] <cooperlees> EWDurbin: haha

[16:56:08] <EWDurbin> PyPI used to use Gluster internally as well

[16:56:10] <cooperlees> Least mine is on Flash :P

[16:56:21] <EWDurbin> i miss ceph a bunch

[16:56:37] <dstufft> S3!

[16:56:49] <cooperlees> dstufft: coolaide drinker

[16:56:51] <EWDurbin> but ceph exposes RBD (Rados Block Device)

[16:56:52] <sumanah> cooperlees: for Facebook work I'm sure they do, but I presume a lot of Facebook folks also have their own personal side projects, etc. and upload to PyPI sometimes. Am I wrong?

[16:57:13] <cooperlees> sumanah: yeah I agree - I posted yesterday actually about Warehouse in our internal Facebook group

[16:57:13] <EWDurbin> RBD is super fabulous! and surprisingly quick for VM hard drives and the like.

[16:57:20] <sumanah> ahhhhh ok

[16:58:24] <EWDurbin> cooperlees: how often does internal facebook mirror sync from PyPI?

[16:58:44] <EWDurbin> i'm kind of surprised that hashing goofiness doesn't bother it more even with flash storage.

[16:58:52] <sumanah> cooperlees: and finally, if you think of someplace else we should be signal-boosting the Warehouse beta, let us know? I particularly am keen to get the word out to people who don't do *that* much Python and people who aren't strong English readers

[16:58:54] <cooperlees> EWDurbin: every 2 hours

[16:59:03] <EWDurbin> cooperlees: why not minutely?

[16:59:08] <cooperlees> EWDurbin: yes your sharing is a PITA

[16:59:13] <pombreda> Hola pypa-dev'ers :)

[16:59:16] <cooperlees> EWDurbin: I have no need to be that up to date

[16:59:17] <sumanah> hi pombreda :)

[16:59:25] <cooperlees> EWDurbin: I was trying to be nice to your infra

[16:59:28] <EWDurbin> cooperlees: honestly more smaller syncs are less taxing and spread out more easily

[16:59:35] <pradyunsg> Hi, I'm late.

[16:59:39] <sumanah> and that's about it for today's livechat on the PyPI beta

[16:59:56] <sumanah> next one: Friday, March 30th, 10-11am EDT, 16:00-17:00 CEST, 7:30pm-8:30pm India, 14:00-15:00 UTC https://www.timeanddate.com/worldclock/fixedtime.html?msg=Warehouse/PyPI+beta+live+chat&iso=20180330T14&p1=1440&ah=1

[16:59:58] <cooperlees> I also then have to sync the packages to 2 other regions - I have a mirror in west coast, east coast and Sweden :)

[17:00:00] <EWDurbin> cooperlees: we've slowly gotten to the point where fewer and fewer bandersnatch mirrors hammer us at UTC midnight

[17:00:19] <sumanah> this is a 24/7 channel so folks should feel free to continue talking here but Ernest and others may drift away :)

[17:00:33] <dstufft> I remember when we'd start erroring at UTC midnight because of xmlrpc kicking our ass

[17:00:40] <EWDurbin> then i fixed it

[17:00:41] <EWDurbin> :-p

[17:00:46] <EWDurbin> one of my most proud accomplishments

[17:00:49] <cooperlees> LoL - YAY for UTC

[17:00:59] <cooperlees> I was stoked have beer in Greenwich @ Meantime brewing company 2 days ago :P

[17:01:06] <cooperlees> Nerdy Beers!

[17:01:08] <cooperlees> *having

[17:01:33] <dstufft> EWDurbin: it's strange the things we're proud of, one of my happiest PyPI changes was getting it onto mandatory TLS :P

[17:01:39] <sumanah> :)

[17:01:55] <EWDurbin> i just hate 500s

[17:01:56] <cooperlees> Is it predominantly you two maintaining these days ?

[17:02:02] <EWDurbin> cooperlees: NO!

[17:02:07] <dstufft> And di_codes1

[17:02:09] <dstufft> are the "core" maintainers

[17:02:13] <dstufft> plus a bunch of contrib

[17:02:16] <cooperlees> cool

[17:02:23] <EWDurbin> there was a time when i used to say "the PyPI team" and it was just a veiled cover of myself and donald

[17:02:25] <EWDurbin> now it's really a team

[17:02:28] <EWDurbin> and it warms my heart

[17:02:29] <EWDurbin> muchly

[17:02:40] <dstufft> I am not good at keeping track of who all is sending PRs :(

[17:02:57] <sumanah> cooperlees: I send a weekly email to pypa-dev & distutils-sig with progress on Warehouse

[17:02:59] <cooperlees> yeah but PRs and deploying to drop are two different things imo

[17:03:07] <cooperlees> sumanah: yeah I read them

[17:03:12] <dstufft> (turns out I'm not good at much except yelling on mailing lists and issues)

[17:03:23] <cooperlees> lol

[17:03:38] <dstufft> luckily this is important skill!

[17:03:42] <EWDurbin> https://usercontent.irccloud-cdn.com/file/rZDQvmeH/Screen%20Shot%202018-03-27%20at%201.03.34%20PM.png

[17:03:48] <sumanah> cooperlees: It's Nicole, Dustin, and Ernest comaintaining basically right now, and I'm really happy about how much space we took up in the recent mail with volunteer PRs that got merged in the last week! https://groups.google.com/forum/#!topic/pypa-dev/w7IMrNiiEuQ

[17:03:51] <EWDurbin> so stoked, can't pick out UTC midnight on that at all anymore!

[17:04:12] <pombreda> one tiny stuff: I am still not sure how to get images properly show up from my description (this is both on old and new pypi ) eg here: https://pypi.org/project/scancode-toolkit/

[17:04:15] <sumanah> cooperlees: hope the mail is useful to you - let me know if there are changes that would make it more useful

[17:04:34] <cooperlees> Will do

[17:04:37] <sumanah> pombreda: looking. I see that there's an image placeholder

[17:04:48] <EWDurbin> pombreda: where is the source for that README?

[17:04:51] <sumanah> pombreda: after "See the roadmap for upcoming features: https://github.com/nexB/scancode-toolkit/wiki/Roadmap" -- that's what you mean?

[17:05:25] <pombreda> https://github.com/nexB/scancode-toolkit/blob/develop/README.rst

[17:05:32] <pombreda> sumanah, yes

[17:06:01] <pombreda> my rst hass this: .. image:: samples/screenshot.png

[17:06:08] <pombreda> which of course is local only

[17:06:16] <pombreda> and relative

[17:06:23] <pombreda> I guess this should be made absolute?

[17:06:31] <EWDurbin> pombreda: yes, that'd do it

[17:06:34] <pombreda> thx

[17:07:06] <cooperlees> sumanah: Where is my best link to point my VP at to see your potential road map for our potential grant? Just finishing email

[17:07:23] <EWDurbin> okay all, i'm out!

[17:07:34] <pombreda> and now for something completely different... at some point of time in a near future, I would like to tackle with a PEP and with Pypi the problem of licensing clarity in Python packages and on Pypi.

[17:08:25] <pombreda> it is not clear how to state a proper license in a package (classifier, license field?) and that should bubble up more clearly in the the pypi in the future

[17:08:42] <sumanah> cooperlees: MarkMangoba and Ernest and I are working on that grant application and can get a ~400-word thing to you later today. Or I can email you now with a one-paragraph thing.

[17:09:02] <cooperlees> sumanah: One paragraph thiong would be sweet

[17:09:06] <cooperlees> cooper@fb.com

[17:09:10] <sumanah> got it cooperlees

[17:09:11] <cooperlees> *thing

[17:09:24] <cooperlees> I've already quoted IRC and your desire for the grant

[17:10:07] <pombreda> re:licensing I am involved a in few projects that deal with this https://spdx.org/ and https://clearlydefined.io/ as well as scancode

[17:10:18] <sumanah> pombreda: interesting! I'm looking around and bet I have seen other questions/issues about related things, like https://github.com/pypa/packaging-problems/issues/41

[17:10:28] <pombreda> :P

[17:10:54] <sumanah> pombreda: I'm sorry, I don't know what you mean by that

[17:11:11] <sumanah> I am not trying to dismiss what you've said, just to find more context

[17:11:20] <pombreda> sure :)

[17:11:51] <pombreda> sumanah, the way we document licensing in packages is a tad messy ATM

[17:11:56] <pombreda> that's it

[17:12:09] <sumanah> Yes, I know that :)

[17:12:39] <pombreda> and I started working on a pep at a very slow pace to address first the metadata side of things https://github.com/pombredanne/spdx-pypi-pep/issues/1

[17:12:43] <pombreda> but I am slow

[17:13:11] <pombreda> now the clearlydefined project is another tack to actually scan, review and help every project provide proper info

[17:13:28] <pombreda> and has a nice roaster involved

[17:14:19] <pombreda> again just a heads up at this stage. Bu there are things to do imho at the pep level, then also help every author provide better licensing ingo and then eventually surface more of it in pypi :P

[17:14:48] <pombreda> s/ingo/info

[17:15:47] <pombreda> sumanah, I will chime in on #41 you listed too

[17:19:48] <cooperlees> sumanah: Should I hold off sending this email for your paragraph? :D

[17:20:03] <sumanah> cooperlees: yeah, 5 min, thanks

[17:25:32] <sumanah> cooperlees: sent

[17:31:23] <cooperlees> sumanah: ack - got it

[17:32:02] <cooperlees> sumanah: let me check if you should bother applying once I hear from my VP

[17:33:50] <sumanah> cooperlees: good to know, thank you

[17:35:22] <waseem18> sumanah: Quick update: I'm getting up to speed on Mypy. Will soon start with small PR'S to Twine.

[17:35:53] <cooperlees> O, type hinting twine :D Nice

[17:36:39] <sumanah> cooperlees: yeah I think it'd be great! I saw https://github.com/pypa/twine/issues/231 and thought "oooh". thanks waseem18.

[17:37:00] <sumanah> cooperlees: y'all do type annotations or similar in your Python codebases?

[17:37:13] <cooperlees> The PEP author sits next to me .... :)

[17:37:38] <waseem18> Amazing cooperlees :)

[17:37:45] <sumanah> cooperlees: I figured it was a pretty strong chance. But I also know how institutions can be sometimes

[17:38:00] <cooperlees> Facebook is pretty different in that regard

[17:38:05] <cooperlees> Instagram even more so

[17:38:05] <sumanah> "a very good thing! the author of the standard works here! but still"

[17:38:17] <cooperlees> They are leading the way with Type annotating on a large code base

[17:38:22] <cooperlees> I think Carl has a talk @ PyCon about it

[17:38:28] <cooperlees> Carl == Carl Meyer

[17:38:43] <sumanah> yeah I see Carl at PyCon usually, I think. Instagram Carl?

[17:39:07] <cooperlees> Yeah that's him

[17:39:18] <cooperlees> O wow - Yannick, my old manager opened the twine Type Hinting issue. haha

[17:39:19] <cooperlees> Small workd

[17:39:25] <cooperlees> *small world

[17:39:32] <sumanah> cooperlees: I worked on Zulip for a while and saw how type annotations helped us move faster, so now that I'm comaintaining Twine, why not

[17:40:05] <cooperlees> All my internal Python 3 is type annotated and checked via mypy

[17:40:44] <sumanah> cooperlees: cool. if you have any resources you think waseem18 and I should look at, other than the PEP & mypy & http://blog.zulip.org/2016/10/13/static-types-in-python-oh-mypy/ , lemme know

[18:20:44] <cooperlees> The typing library docs. mypy docs and running mypy should get them there :)

[18:35:31] <sumanah> pombreda: will you announce your license clarification work on distutils-sig?

[18:46:44] <pombreda> sumanah, of course :P

[18:46:50] <sumanah> :)

[18:48:18] <pombreda> sumanah, when this is starting to take more shape. If you are interested in the details for the community, peer review and curation/license clarity side, there was a prese made recently there https://osls18.sched.com/event/Djsx/clearlydefined-enabling-project-success-through-metadata-jeff-mcaffer-microsoft-rashmi-chitrakar-qualcomm

[18:48:32] <pombreda> the pep side, I just need time :P

[18:48:42] <pombreda> so many fun things to do and so little time :P

[18:50:13] <pombreda> sumanah, and this is a project we do under the the OSI umbrella https://opensource.org/clearlydefined

[18:50:31] <pombreda> I am proud to have coined the term FWIW :P

[18:50:38] <pombreda> clearlydefined

[18:51:01] <pombreda> afk for a few, will catch up on logs

[20:36:15] <thea> hrm. Should readme_renderer warn about relative URLs (especially for images)?

[20:39:39] <sumanah> thea: hmmm

[20:40:08] <sumanah> thea: you mean because relative URLs, especially for embeds, will not render properly on PyPI?

[20:40:12] <sumanah> thea: my take is: yes

[20:42:21] <thea> yeah

[20:42:42] <thea> I'll file a bug. The other readme_renderer stuff needs to get merged in first.

[20:43:58] <thea> https://github.com/pypa/readme_renderer/issues/71

[20:49:10] <sumanah> nod nod

[21:08:45] <sumanah> agronholm: I'll re-up my offer if there is anything I can do (testing, for instance) to help you get the next `wheel` out

[21:17:34] <agronholm> sumanah: I've outlined the issues I want to fix for v0.31

[21:17:41] <agronholm> sumanah: just check the milestone there

[21:18:02] <agronholm> sumanah: would a release this week work?

[21:18:14] <sumanah> agronholm: indeed it would!

[21:18:20] <sumanah> looking at https://github.com/pypa/wheel/milestone/2 now.

[21:18:38] <sumanah> agronholm: thank you.

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Tuesday the 27th of March, 2018