| *** mhen_ is now known as mhen | 01:49 | |
| *** tosky_ is now known as tosky | 07:33 | |
| opendevreview | Stephen Finucane proposed openstack/pbr master: Do not use the onerror parameter in shutil.rmtree() https://review.opendev.org/c/openstack/pbr/+/924803 | 09:11 |
|---|---|---|
| opendevreview | Stephen Finucane proposed openstack/pbr master: Silence Python warnings https://review.opendev.org/c/openstack/pbr/+/954708 | 09:14 |
| hberaud[m] | dtantsur: Nice works, congrats. A side topic that you may want to heard of https://review.opendev.org/c/openstack/nova/+/948340 | 09:14 |
| hberaud[m] | I just made some suggestions around releases notes on your both patches, else, LGTM | 09:32 |
| dtantsur | Looking now | 09:34 |
| dtantsur | hberaud[m]: not sure what you mean there: the green pool also does not shrink. It just matters less with lightweight green threads. | 09:35 |
| dtantsur | The underlying greenpool has a resize method, but we don't seem to ever call it | 09:37 |
| opendevreview | Dmitry Tantsur proposed openstack/futurist master: Add DynamicThreadPoolExecutor that resizes itself https://review.opendev.org/c/openstack/futurist/+/955217 | 09:43 |
| dtantsur | hberaud[m]: here are some reno and docstrings improvements ^^^ | 09:43 |
| hberaud[m] | ack thanks | 10:03 |
| hberaud[m] | perfect | 10:03 |
| hberaud[m] | I'd also suggest to add a release note there https://review.opendev.org/c/openstack/futurist/+/955478 | 10:04 |
| dtantsur | hberaud[m]: it's not user-visible, the API is not exposed via public interfaces. | 10:05 |
| dtantsur | Which behavior changes do you have in mind? | 10:05 |
| hberaud[m] | Indeed sorry | 10:05 |
| hberaud[m] | I was thinking about tombstone on the queue, causing all workers to eventually stop | 10:06 |
| dtantsur | Yeah, but stop() is already only called in situations where a total stop is required. In all instances I could find, we iterate over all workers anyway. | 10:07 |
| dtantsur | So unless I"m missing something important, it's not actually a change. | 10:07 |
| hberaud[m] | Yeah you are right | 10:07 |
| hberaud[m] | ignore my comment | 10:07 |
| dtantsur | TheJulia: btw we need to consider running our CI job on futurist. | 10:11 |
| hberaud[m] | dtantsur: concerning your previous request about pep8-incompatible recommendation, is it still relevant? Apparently you fixed it. | 11:38 |
| dtantsur | hberaud[m]: I have not, futurist still prefers W503 over W504, the former being pep8-incompatible | 11:43 |
| hberaud[m] | Ok, then I'll try to move to W504 | 11:44 |
| hberaud[m] | I see many projects ignore both | 11:46 |
| dtantsur | heh | 11:46 |
| hberaud[m] | do you have a preference? | 11:46 |
| hberaud[m] | https://codesearch.opendev.org/?q=W503&i=nope&literal=nope&files=&excludeFiles=&repos= | 11:46 |
| dtantsur | W504 matches pep8, which is why it's my preference | 11:46 |
| hberaud[m] | ok, sold | 11:46 |
| dtantsur | Has anyone seen tox just hard hanging on start-up? Oo | 11:48 |
| dtantsur | is it for everyone or just me? | 11:49 |
| hberaud[m] | yeah, it hang on on requirements | 11:50 |
| dtantsur | neat | 11:51 |
| dtantsur | okay, so I'm splitting the properties per gibi request. But I'll wait a bit until I can run the unit tests locally before submitting. | 11:51 |
| * dtantsur has lunch in the meantime | 11:51 | |
| opendevreview | Hervé Beraud proposed openstack/futurist master: prefer using flake8 W504 instead of W503 https://review.opendev.org/c/openstack/futurist/+/955593 | 12:07 |
| hberaud[m] | dtantsur: it will impact your patch, either we merge yours first and then I go back with mine, or the contrary, its up to you | 12:08 |
| hberaud[m] | ^ | 12:08 |
| dtantsur | Sigh, it will | 12:09 |
| dtantsur | hberaud[m]: I'd slightly prefer to get my series in first. I'm starting a long PTO on Friday. | 12:09 |
| dtantsur | or I can rebase the entire chain if you don't want to wait | 12:10 |
| hberaud[m] | Yeah sure no problem, I propose to go first with your patch, and I'll rework mine once done | 12:10 |
| hberaud[m] | No rush from my side | 12:10 |
| dtantsur | thx! | 12:10 |
| opendevreview | Merged openstack/pbr master: Silence Python warnings https://review.opendev.org/c/openstack/pbr/+/954708 | 12:42 |
| opendevreview | Merged openstack/pbr master: Do not use the onerror parameter in shutil.rmtree() https://review.opendev.org/c/openstack/pbr/+/924803 | 12:42 |
| opendevreview | Dmitry Tantsur proposed openstack/futurist master: Fix ThreadWorker.stop to stop only this worker https://review.opendev.org/c/openstack/futurist/+/955478 | 12:48 |
| opendevreview | Dmitry Tantsur proposed openstack/futurist master: Add DynamicThreadPoolExecutor that resizes itself https://review.opendev.org/c/openstack/futurist/+/955217 | 12:48 |
| opendevreview | Dmitry Tantsur proposed openstack/futurist master: Provide insights into the state of ThreadPoolExecutor https://review.opendev.org/c/openstack/futurist/+/955595 | 12:48 |
| dtantsur | hberaud[m], gibi, this is the chain now ^^^ | 12:48 |
| hberaud[m] | ack | 12:52 |
| TheJulia | dtantsur: already considered/pondered on my end. I'm going to pin the stack size down, and even then we really don't *appear* to chew that much memory. I think it should, realistically, be fine in the grand scheme of things | 13:10 |
| * dtantsur nods | 13:12 | |
| TheJulia | dtantsur: I'm under no illusion there will be an increased footprint, but its the virtual which skyrockets due to the free memory allocations, not the actual resident amount. | 13:27 |
| TheJulia | (at least, when the default stack size is pinned down) | 13:28 |
| gibi | dtantsur: ack thanks | 13:28 |
| TheJulia | dtantsur: eh, slightly closer to 8 minutes on the current rev, but generally I'll just let it keep going since it is still early on | 17:13 |
| TheJulia | but overall, still seeming consistent and generally good | 17:14 |
| TheJulia | dtantsur: hovering right around 7:45 consistently | 18:05 |
| TheJulia | SGTM | 18:05 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!