pmxbot IRC Log Viewer

[01:19:00] <lifeless> hmm, I think I'm nearly in a position to get back to pip work

[01:19:15] <lifeless> Nakato: / tchaypo: how are things on your tickets?

[01:20:45] <tchaypo> Good morning!

[01:20:58] <Nakato> lifeless: I'm down the rabbit hole of path in constraints and making progress.

[01:21:19] <tchaypo> unless I’ve dropped something, I don’t think I have anything in flight except the PEP (and the accompanying marker implementation)

[01:22:01] <Nakato> And still have the tox constraints on the plate, of course

[01:23:09] <tchaypo> the PEP bit is easy (I think it’s largely done[1] except for needing to fix the BNF already) [1] for values of ‘done’ that mean ‘drafted and ready for feedback’); the implementation is a bit scarier - I’m going to have to figure out how to do the parsinga

[01:23:56] <tchaypo> (some of it is easy - the existing string comparison stuff I should be able to steal from the existing implementation; and packaging already parses PEP-440 versions and does comparisons on them, so I don’t need to do that bit)

[01:25:56] <lifeless> tchaypo: ok cool

[01:26:30] <lifeless> Nakato: \o/

[01:26:39] <lifeless> ok so I think I'm going to check the trello

[01:26:59] <lifeless> I've backported all the bits to juno and kilo and am just waiting on reviews

[01:27:39] <tchaypo> oh

[01:27:54] <tchaypo> https://review.openstack.org/#/c/187823/ - nakato’s bit has been released, so I’ll redo the checks as I think that should work now

[01:28:22] <Nakato> Having a path in constraint's really breaks the workflow in req_set

[01:29:00] <tchaypo> This could be completely off base, but the first thought that floats into my head is “well don’t do that then"

[01:29:52] <lifeless> tchaypo: see devstack

[01:30:03] <tchaypo> more specifically - the machinery that parses requirements.txt drops the paths from “-e blah#egg_info=packagename” pretty early and just works with “packagename” for the most part - the path is only referred to for the bits that actually need to handle the installation

[01:30:20] <lifeless> tchaypo: we edit-constraint to change oslo.versionedobjects >=,<= -> file:///opt/stack/oslo.versionedobjects

[01:30:25] <tchaypo> I don’t know what you’re trying to do right now, but maybe something similar would work for handling the contraints?

[01:46:26] <lifeless> Nakato: https://review.openstack.org/198188 will perhaps help a bit

[01:46:37] <lifeless> Nakato: it will give a guaranteed package name to compare against

[01:46:47] <lifeless> Nakato: and its the only syntax we needto support in openstack today

[01:47:04] <lifeless> Nakato: so it might let you divide and conquer into a today plus future set of work

[01:47:12] <lifeless> where future can depend on my resolver branch

[01:51:32] <Nakato> tchaypo: Hym, I think we need to look into 'gate-tempest-dsvm-neutron-src-pbr'

[01:56:01] <Nakato> http://paste.openstack.org/show/346956/

[01:56:43] <tchaypo> my eyebrows are doing all manner of things to indicate my level of WTF

[01:57:38] <tchaypo> yep, I think you’re right.

[01:59:32] <Nakato> tchaypo: I think I've got how to reproduce it local. run $(edit-constraints constraints.txt -- NameNotInConstraints "whatever"

[01:59:35] <Nakato> )

[02:00:10] <tchaypo> also the py34 failure; virtualenv should be installed for python >=2.7 but either it’s not, or this is one of those occasions where `pip freeze` lies about what’s in the venv - perhaps pip list is going to be a better choice

[02:01:43] <Nakato> The gate-tempest-dsvm-neutron-src-pbr seems to be general gate borkage. Not releated to your patch.

[02:01:43] <Nakato> https://review.openstack.org/#/c/193927/

[02:04:13] <lifeless> Nakato: tchaypo: tahats fixed in my patches

[02:05:09] <lifeless> https://review.openstack.org/#/c/198188/ (merged, yay), and https://review.openstack.org/#/c/198186/

[02:07:02] <Nakato> Cool

[02:08:24] <lifeless> I'm fairly sure that the clone of requirements is -not- ZUUL_REF ready

[02:08:27] <lifeless> this is arguably a bug

[02:08:32] <lifeless> but also makes your Depends-On meaningless

[02:08:53] <tchaypo> booo, hisss

[02:09:55] <lifeless> oh oof, coinor.pulp. Stabbbbb

[02:14:48] <lifeless> tchaypo: did you merge https://review.openstack.org/#/c/187846/ into your experiment ?

[02:14:52] <lifeless> tchaypo: if so I can abandon it

[02:28:51] <tchaypo> I thought I had, but I can’t see where I did it..

[02:30:11] <tchaypo> I vaguely remember expecting my patch to fail but 187846 to succeed. So let me see how that works out

[02:33:29] <tchaypo> nope, it seems that pip freeze doesn’t hide virtualenv, so that py34 failure is real

[02:33:42] <tchaypo> (and possibly fixed by 187846)

[03:18:17] <tchaypo> okay, 187846, even updated to make sure setuptools>=17.1 is installed, doesn’t get virtualenv installed

[03:19:32] <tchaypo> on py34, it seems to be fine on py27

[03:19:45] <tchaypo> I’ll come back to this later.

[03:19:57] <tchaypo> .. and suddenly on this run I’m getting setuptools-18.0.1 installed.

[03:22:04] <lifeless> still liking your devpi?

[03:22:54] <lifeless> or is this in infra?

[03:25:23] <tchaypo> on my machine

[03:26:57] <lifeless> yah

[03:36:44] <tchaypo> excellent. I used `tox —recreate -epy27` to get a fresh venv; and now I have the fun that when it’s trying to install pbr, it first uninstalls pbr-0.0, which seems to involve it deleting $TMPDIR/venv/bin/pbr twice; when the second one fails it gives up and errors

[03:39:48] <tchaypo> http://paste.openstack.org/show/ZMIKLLRF1k0GPBbxl77m/ - lines 36 and 126

[03:45:06] <lifeless> wat -

[03:45:08] <lifeless> /var/folders/_n/ft3sz29979590_nj7tj1dxhm0000gn/T/tmphMmQvy/tmpkSv6Ru/

[03:51:19] <lifeless> tchaypo: if your venv is affecting your tests, the tests aren't isolated enough

[03:51:23] <lifeless> tchaypo: (for these tests)

[03:51:34] <lifeless> tchaypo: I thought we were making our own venv and upgrading within it ?

[03:57:45] <tchaypo> the /var/folders stuff is OS X specialness

[03:58:09] <tchaypo> but i think the venv/ folder there is the root of our venv

[03:58:32] <lifeless> what does tox have to do with our tests?

[03:58:50] <tchaypo> it runs them?

[03:59:02] <lifeless> it does not

[03:59:29] <tchaypo> okay

[03:59:35] <lifeless> it makes an sdist

[03:59:43] <lifeless> it makes venvs to contain the deps for us

[03:59:43] <tchaypo> it makes a venv in which to run testr (which runs the tests), and then runs testr (which runs the tests)

[03:59:47] <lifeless> and it installs the sdist into the venv

[03:59:50] <lifeless> then runs testr

[03:59:56] <lifeless> testr doesn't run the tests

[04:00:06] <lifeless> testr runs the test runner(s)

[04:00:10] <tchaypo> right

[04:00:22] <lifeless> the tests you've been hacking on make their own venvs

[04:00:56] <lifeless> anyhow, I'd just rm -rf .tox

[04:01:11] <lifeless> if you want to make tox remake the venv with hopefully less confusion

[04:04:08] <tchaypo> no difference

[04:04:25] <tchaypo> or, to use full sentences instead of relying on telepathy

[04:04:34] <tchaypo> I’ve given that a go but I get the same result

[04:08:44] <lifeless> thats from the tests or from tox?

[04:09:01] <lifeless> (use tox --notest) to separate the two thing out

[04:11:42] <tchaypo> *double-checks* the error is reported in a from the tests - it’s inside the stdout {{{ }}} block from FAIL: pbr.tests.test_packaging.TestRequirementParsing.test_requirement_parsing(not-editable,tree,pbr_installed)

[04:12:05] <tchaypo> and one other test also

[04:12:35] <tchaypo> and tox -epy27 -notest runs fine

[04:15:37] <Nakato> lifeless: https://github.com/pypa/pip/pull/2937 Added tests, with file://...#egg=name syntax.

[04:15:46] <lifeless> tchaypo: ok, so tox has nothing to do with it then :)

[04:15:50] <lifeless> Nakato: \o/

[04:17:05] <tchaypo> right. perhaps I could have said “I recreated the venv in which testr is running the test runner” rather than “I used tox to recreate the venv"

[04:19:49] <lifeless> tchaypo: it sounded like you were saying tox was erroring

[04:20:12] <tchaypo> yeah

[04:20:28] <tchaypo> mentioning tox and then referring to “when it’s trying to install pbr"

[04:20:39] <tchaypo> not very clear when “it” is something I haven’t mentioned in the sentence

[04:27:50] <lifeless> tchaypo: Your paragraph was very clear. It was also not what you meant :)

[04:35:04] <tchaypo> :)

[04:37:12] <tchaypo> it may have been clearer in german, where I would have had to use the dative for the indirect object (reducing the risk of confusing it (the tests) with it tox, the direct object)

[04:49:51] <lifeless> if you wish to practice your german here, thats fine. I can't promise to follow along

[04:52:11] <tchaypo> I’ll just say that my favorite new german word last week was “zahnradbahn” - tooth-wheel train

[13:38:00] <sigmavirus24> So, gihttps://github.com/certifi/python-certifi/issues/24 just made me realize that wheels apparently don't respect MANIFEST.in for including stuff like a LICENSE file, etc. Is that a bug or intentional?

[13:53:57] <dstufft> MANIFEST.in says what to include in the tarball, wheels are just a bundle of things to actually install

[13:56:33] <sigmavirus24> So, should wheels have a LICENSE included or no? I'm guessing the answer is no

[13:56:36] <sigmavirus24> Which seems wrong

[13:58:51] <dstufft> as written, No. I believe that Metadata 2.0 will add them though

[14:02:18] <sigmavirus24> okay

[14:02:27] <sigmavirus24> just wanted to make sure

[22:26:27] <lifeless> yay one card green

[22:26:45] <lifeless> Nakato: tchaypo: are you in need of reviews?

[22:30:31] <Nakato> lifeless: Only thing outstanding currently is the pip pull request. https://github.com/pypa/pip/pull/2937

[22:30:31] <Nakato> Can look at some if you need any.

[22:31:30] <lifeless> I've a bunch pending review in openstack but they're all very shallow

[22:31:59] <lifeless> https://review.openstack.org/#/q/owner:lifeless+status:open,n,z

[22:35:25] <lifeless> Nakato: reviewed

[22:35:40] <lifeless> Nakato: I think one extra condition in the if block should do it

[22:46:56] <tchaypo> I have a PR open on the interoperability-peps repo with my pep

[23:12:23] <tchaypo> https://github.com/pypa/interoperability-peps/pull/29

Log file Viewer

Help | Karma | Search:

#pypa-dev logs for Monday the 6th of July, 2015