| gibi | damani[m]: ralonsoh: re cotyledon dependency. Yeah it feels strange that. Does eventlet also made an optional dependency of oslo.service? | 11:59 |
|---|---|---|
| gibi | I think the main problem with optional dependencies that pip's requirement.txt does not support expressing that I need oslo.service with cotyledon backend | 12:00 |
| gibi | I need to independently list oslo.service and cotyledon | 12:00 |
| gibi | hm I might be mistaken about the requirements.txt support for optional dependencies | 12:01 |
| gibi | https://www.python.org/dev/peps/pep-0508/#extras | 12:02 |
| gibi | let me try it... | 12:02 |
| gibi | OK locally oslo.service[threading] >= 4.2.0 works from a requirements.txt file. I will update https://review.opendev.org/c/openstack/nova/+/948311/23/requirements.txt to see if it works on the gate too | 12:07 |
| ralonsoh | gibi, sorry, I was having lunch | 12:23 |
| ralonsoh | but how pip checks the option? | 12:23 |
| ralonsoh | checking your patch now | 12:23 |
| gibi | the patch is not updated yet | 12:36 |
| gibi | but I think it we should add `oslo.service[threading] >= 4.2.0` to our requirements.txt | 12:36 |
| gibi | to request the extra installs | 12:37 |
| gibi | https://review.opendev.org/c/openstack/nova/+/948311/25/requirements.txt now it is updated, lets see if CI likes it or not | 12:49 |
| damani[m] | gibi, i have not put eventlet optonal because it's the default backend | 12:50 |
| damani[m] | but it's a good question do we put eventlet as an optional requirements too | 12:51 |
| damani[m] | and yes that is is the good syntax oslo.service[threading] >= 4.2.0 | 12:51 |
| gibi | damani[m]: ack. If the above requirements.txt allows to pass CI then I have nothing further :) | 12:54 |
| gibi | damani[m]: ralonsoh: OK that requirements.txt line works in CI too https://zuul.opendev.org/t/openstack/build/ab610d74de4e4adcbc22f188431fbfb9/log/job-output.txt#2496 so I rest my case | 13:50 |
| ralonsoh | gibi, cool thanks! | 13:56 |
| JayF | oh awesome | 14:12 |
| JayF | when did requirements.txt learn about package[optional] syntax!? | 14:12 |
| JayF | Thank you \o/ | 14:12 |
| gibi | JayF: you are welcome. This eventlet removal excersise is a great learning opportunity for me so far. | 14:16 |
| JayF | So much of it is learning how many things I've known for so long somehow stopped being true in the interim LOL | 14:16 |
| gibi | time is a harsh mistress | 14:17 |
| damani[m] | it's what i have mentioned yesterday about the syntax | 14:18 |
| damani[m] | i have made an example yesterday with the syntax | 14:18 |
| * JayF updated networking-baremetal patch's requirements.txt | 14:25 | |
| JayF | I think at some point, these are going to hit the NGS/NBM repos (whatever runs the ML2 plugin agents?) -- when that time comes let me know and I can line up patches so we can try and keep everything happy | 14:26 |
| JayF | We've been fighting CI in Ironic for the last week, some devstack neutron changes cratered our job and I'd like to avoid a similar thing happening again | 14:27 |
| JayF | gibi: that same requirements line is blowing up NBM: https://review.opendev.org/c/openstack/networking-baremetal/+/947985/6/requirements.txt https://www.irccloud.com/pastebin/uHifK6Z7/ | 15:18 |
| * JayF wonders what he did wrong | 15:20 | |
| gibi | strange it is almost all green in https://zuul.opendev.org/t/openstack/status?change=948311 | 15:20 |
| gibi | with the same req line :/ | 15:21 |
| JayF | no difference in pbr/setuptools requirements in pyproject.toml, either | 15:27 |
| JayF | gibi: it's reproducable locally with both networking-generic-switch and networking-baremetal executed via tox. I cannot explain why it works in nova, everything I'm seeing points to it shouldn't work | 19:36 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!