| tchaypo | adam_g: the change about handling the requirements pinning has me conflicted | 01:29 |
|---|---|---|
| *** pixelb has quit IRC | 02:24 | |
| * tchaypo writes words | 02:25 | |
| prometheanfire | ohai | 02:26 |
| *** jamielennox is now known as jamielennox|lunc | 02:32 | |
| *** jamielennox|lunc is now known as jamielennox|food | 02:32 | |
| *** jamielennox|food is now known as jamielennox | 03:01 | |
| adam_g | tchaypo, the spec was largely rejected when brought up in this weeks cross project meeting, tho no one seemed to have any better alternative other, but your comments are still welcome | 03:29 |
| tchaypo | aside from “improve the tooling" | 03:34 |
| tchaypo | https://github.com/nvie/pip-tools/issues/10 seems somewhat similar | 03:35 |
| tchaypo | at least in terms of the problems that it wants to solve | 03:35 |
| tchaypo | but like, that just led to pip-compile, which we already know about, and which isn’t a complete solution to what we need | 03:36 |
| tchaypo | compiling the file is the easy part | 03:36 |
| tchaypo | my thinking is that if we put the compiled everything-pinned version as requirements.txt we make life hard for downstream packagers; if we put it as requirements.gate we make life hard for people relying on package installs | 03:37 |
| tchaypo | either way, all downstream packagers are going to be making their own pinning decisions based on what’s in their distro, so what we test in the gate is not relevant to what end users do | 03:40 |
| tchaypo | unless they take our pip packages directly. It feels like a pip preference to read a “requirements-compiled” if it exists might be useful | 03:45 |
| * tchaypo continues with a stream of consciousness | 03:46 | |
| tchaypo | actually this reminds me of a point made on the discussion I linked that led to pip-compile: originally, setup.py had a list of top-level dependencies for pip to use when installing the package; while the requirements.txt was intended to be generated by “pip freeze”ing an environment and used to control deployment | 03:48 |
| tchaypo | if I understand correctly, our use of PBR gets rid of that distinction; the same requirements.txt file is used to control deployment *and* pushed into setup.py to control installation | 03:49 |
| tchaypo | perhaps we could have pbr copy the top-level deps into the package - so that people who “pip install” get what they expect, but people who want more precision have the option of installing from git | 03:51 |
| tchaypo | but agian, that would mean we’re not “doing what end users do” for more than a subset of end users | 03:51 |
| tchaypo | I should probably read the meeting notes and discover that everyone else has already trod this path.. | 03:51 |
| adam_g | "if we put it as requirements.gate we make life hard for people relying on package install" <- is that people relying on python packages, or distro packages? | 03:55 |
| adam_g | tchaypo, i dont have much of an argument aginst the notion that "we're not doing what end users do" other than 1) with aboslutely no data to prove it, i expect consumers of stbale branches dont rely on pip installations 2) we give a requirements.txt to end user that specifies capped version ranges.. gate, we'd be installing versions within those ranges, which is exactly what users get.. except we'd be specific about which versions, ins | 03:57 |
| adam_g | tead of rolling the dice and relying on pip (which is whats actually broken) | 03:57 |
| adam_g | my client split that into two msg's, hopefully thats legible | 03:58 |
| adam_g | but i need to run another van load to my new house. cya | 03:58 |
| tchaypo | I meant both - people installing python packagers may not get what we tested, and people installing distro packages won’t. But I was wrong, because no matter what we put in requirements.txt people using a distro package will always get the distro packager’s set of versions, not ours | 04:01 |
| tchaypo | much sympathy for moving house. I got that done two weeks ago and I still have most of my stuff in boxes | 04:02 |
| tchaypo | … packaging dilemmas intrude into real life | 04:02 |
| tchaypo | meh. So this seems like it’s mostly rejected for now, in favor of continuing to cause pain until we figure out how to change pip | 04:36 |
| tchaypo | so I’m just going to stop thinking about it | 04:36 |
| *** zerda has joined #openstack-stable | 04:43 | |
| *** jamielennox is now known as jamielennox|away | 06:35 | |
| *** mrunge has joined #openstack-stable | 06:52 | |
| *** e0ne has joined #openstack-stable | 07:51 | |
| *** e0ne has quit IRC | 08:19 | |
| *** pixelb has joined #openstack-stable | 08:43 | |
| *** jeffmaury has joined #openstack-stable | 08:52 | |
| *** jeffmaury has quit IRC | 09:03 | |
| *** jamielennox|away is now known as jamielennox | 09:14 | |
| *** jeffmaury has joined #openstack-stable | 09:22 | |
| *** derekh has joined #openstack-stable | 09:28 | |
| *** jeffmaury has quit IRC | 09:32 | |
| *** jeffmaury has joined #openstack-stable | 09:32 | |
| *** jeffmaury has quit IRC | 10:01 | |
| *** jeffmaury has joined #openstack-stable | 10:04 | |
| *** sileht has quit IRC | 10:07 | |
| *** e0ne has joined #openstack-stable | 10:08 | |
| *** sileht has joined #openstack-stable | 10:22 | |
| *** zerda has quit IRC | 10:40 | |
| *** e0ne is now known as e0ne_ | 10:55 | |
| *** e0ne_ is now known as e0ne | 10:57 | |
| *** jeffmaury has quit IRC | 11:01 | |
| *** jeffmaury has joined #openstack-stable | 11:35 | |
| *** e0ne is now known as e0ne_ | 11:49 | |
| *** jeffmaury has quit IRC | 11:51 | |
| *** e0ne_ is now known as e0ne | 11:59 | |
| *** eharney has quit IRC | 12:07 | |
| *** ihrachyshka has joined #openstack-stable | 12:11 | |
| *** jeffmaury has joined #openstack-stable | 12:34 | |
| *** e0ne is now known as e0ne_ | 12:41 | |
| *** e0ne_ is now known as e0ne | 12:44 | |
| *** jeffmaury has quit IRC | 12:52 | |
| *** ihrachyshka has quit IRC | 13:14 | |
| *** eharney has joined #openstack-stable | 13:15 | |
| *** eharney has quit IRC | 13:25 | |
| *** eharney has joined #openstack-stable | 13:26 | |
| *** eharney has quit IRC | 13:33 | |
| *** jamielennox is now known as jamielennox|away | 13:36 | |
| *** eharney has joined #openstack-stable | 13:49 | |
| *** ihrachyshka has joined #openstack-stable | 13:51 | |
| prometheanfire | adam_g: tchaypo: I only have one or two exceptions to the pip requirements in the versions I install | 14:24 |
| prometheanfire | adam_g: tchaypo: this is packaging for gentoo, we are able to do more fancy dep resolution :P | 14:24 |
| *** e0ne is now known as e0ne_ | 14:26 | |
| *** e0ne_ is now known as e0ne | 14:27 | |
| *** jeffmaury has joined #openstack-stable | 14:28 | |
| *** jeffmaury has quit IRC | 14:38 | |
| *** xaeth_afk is now known as xaeth | 14:40 | |
| *** mrunge has quit IRC | 14:56 | |
| *** jeffmaury has joined #openstack-stable | 15:03 | |
| *** jeffmaury has quit IRC | 15:15 | |
| *** apevec has joined #openstack-stable | 15:31 | |
| *** apevec has joined #openstack-stable | 15:31 | |
| *** med_ has joined #openstack-stable | 15:44 | |
| med_ | What's the procedure for getting a Kilo fix that just landed backported to stable/juno. Does someone just need to do the work and then propose it? | 15:45 |
| med_ | (in particular this is in reference to the oslo.messaging heartbeat work that just landed) | 15:45 |
| med_ | ^ adam_g jamespage Daviey | 15:45 |
| prometheanfire | med_: https://wiki.openstack.org/wiki/Fuel/How_to_contribute#Backport_bugfixes_to_stable_release_series | 15:47 |
| * med_ reviews https://wiki.openstack.org/wiki/StableBranch and notes that it definitely meets the criteria | 15:47 | |
| med_ | thanks prometheanfire | 15:47 |
| prometheanfire | that was the next place :P | 15:47 |
| prometheanfire | guess you are suposed to submit the review directly from the stable/blah branch | 15:48 |
| prometheanfire | https://wiki.openstack.org/wiki/Documentation/HowTo#How_to_a_cherry-pick_a_change_to_a_stable_branch | 15:48 |
| prometheanfire | that last link seems best | 15:49 |
| prometheanfire | there are fixes I should probably backport too :| | 15:49 |
| *** jeffmaury has joined #openstack-stable | 16:48 | |
| *** jeffmaury has quit IRC | 16:49 | |
| *** e0ne has quit IRC | 17:11 | |
| *** e0ne has joined #openstack-stable | 18:04 | |
| *** e0ne has joined #openstack-stable | 18:05 | |
| *** derekh has quit IRC | 18:11 | |
| *** e0ne has quit IRC | 18:57 | |
| *** e0ne has joined #openstack-stable | 18:58 | |
| *** e0ne has quit IRC | 19:08 | |
| *** e0ne has joined #openstack-stable | 19:10 | |
| *** apevec has quit IRC | 19:15 | |
| *** xaeth is now known as xaeth_afk | 19:15 | |
| *** e0ne has quit IRC | 19:19 | |
| *** jeffmaury has joined #openstack-stable | 19:26 | |
| *** ihrachyshka has quit IRC | 19:44 | |
| *** xaeth_afk is now known as xaeth | 19:46 | |
| *** eharney has quit IRC | 20:38 | |
| *** jamielennox|away is now known as jamielennox | 21:49 | |
| *** jeffmaury has quit IRC | 21:59 | |
| *** xaeth is now known as xaeth_afk | 22:02 | |
| *** jamielennox is now known as jamielennox|away | 22:05 | |
| *** jeffmaury has joined #openstack-stable | 22:08 | |
| *** jeffmaury has quit IRC | 23:14 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!