| *** jhorstmann is now known as Guest5096 | 01:44 | |
| *** elodilles_pto is now known as elodilles | 08:34 | |
| *** marios_ is now known as marios | 12:47 | |
| fungi | we've gone a week without any backup failure notifications, so i guess the prune helped speed up weekly validation runs | 13:23 |
|---|---|---|
| *** sfinucan is now known as stephenfin | 14:06 | |
| clarkb | nice | 15:49 |
| clarkb | I need to focus on some beginning of the year paperwork stuff this week. I'll also work on getting a meeting agenda out today. If I can work through that stuff then maybe I can boot a noble paste late this week? | 15:53 |
| fungi | sounds good, i have similar tasks distracting me at the moment | 16:05 |
| fungi | clarkb: if we want to consider switching bindep over to pyproject.toml, it might make sense to couple that with the stack of fixes/updates we have for testing it on noble and whatnot | 16:13 |
| clarkb | oh ya that seems like a good tool to switch over | 16:15 |
| clarkb | doing my monday morning local package updates updated xfce. It set my background image to the default for xfce overriding what I Had before. I discovered a fun bug setting it back again. It seems that if I have the settings manager fullscreened over the desktop it won't change the image. But if I reduce its size so that part of thei mage is actually rendered it will | 16:18 |
| clarkb | xfce has taken a stance on "if a tree falls in a forest and no one is around to hear it?" philosophical questions | 16:18 |
| fungi | amusing | 16:19 |
| clarkb | fungi: ubuntu ports afs mirror is all happy now right? I can clear that off the meeting agenda? | 16:28 |
| fungi | checking | 16:29 |
| fungi | yeah, last updated 2 hours ago | 16:29 |
| fungi | it's all good | 16:29 |
| fungi | looks like mirror.epel hasn't updated for a couple of days, but the only other stale mirrors are empty volumes we never got around to deleting | 16:30 |
| fungi | rsync: failed to connect to pubmirror3.math.uh.edu (129.7.128.191): Connection timed out (110) | 16:31 |
| clarkb | great I'll drop that from the meeting agenda. We can add back if epel mirroring is persistently sad | 16:32 |
| fungi | noonedeadpunk: ^ you were the one to update the epel mirroring to that source a few years ago in https://review.opendev.org/862793 so maybe you have suggestions or contacts who can say whether we should move off it to something else? | 16:33 |
| fungi | i wonder if we should simply revert back to pubmirror1.math.uh.edu | 16:34 |
| fungi | i don't see any of the math.uh.edu mirrors listed at https://mirrormanager.fedoraproject.org/mirrors/EPEL currently | 16:38 |
| clarkb | osuosl might be a good choice as we know who to talk to there | 16:39 |
| clarkb | I also know who to takl to at the psu mirror | 16:39 |
| fungi | also dfw.mirror.rackspace.com claims to support rsync | 16:39 |
| fungi | and i can confirm it's listening on the port at least | 16:40 |
| clarkb | I think we started using rax's mirrors but they often fell behind on updates? but we know who to talk to there and it should be very local | 16:40 |
| noonedeadpunk | fungi: ugh, good question | 16:40 |
| noonedeadpunk | I can totally recall that and that some communication happened in IRC about it... | 16:40 |
| noonedeadpunk | and I think it was still tibbs who suggested it.... | 16:41 |
| noonedeadpunk | so note was still relevant after an update | 16:42 |
| fungi | yeah, i'm wondering if they took all their mirrors offline | 16:43 |
| fungi | even https://math.uh.edu/ itself times out in my browser | 16:43 |
| fungi | could be they restructured all the university of houston network resources or something | 16:44 |
| clarkb | maybe they decided to stop teaching math | 16:48 |
| fungi | zing! | 16:49 |
| fungi | clarkb: any objection to me rebasing your 816741 for bindep pyproject.toml on top of my stack of test updates? right now it's not passing and i think it would once they merge | 16:51 |
| clarkb | fungi: that would be great thanks | 16:54 |
| clarkb | I've got a first draft of a meeting agenda on the wiki now. Introduces planning for service coordinator election timing and a beginning of the year "meetup" (spiritual successor to the pre ptg we had last year) | 16:57 |
| clarkb | let me know if anything important is missing from there and I can edit/add/remove as necessary. I won't send it out until end of day today to give time for feedback on that | 16:58 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Update for newer Python, Tox and Hacking https://review.opendev.org/c/opendev/bindep/+/932190 | 17:12 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Support and prefer unittest.mock https://review.opendev.org/c/opendev/bindep/+/878733 | 17:12 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Fix Python version guessing logic for Debian Sid https://review.opendev.org/c/opendev/bindep/+/932191 | 17:12 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Add testing for Debian testing and unstable https://review.opendev.org/c/opendev/bindep/+/932192 | 17:12 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 17:12 |
| fungi | reworked the series while i was in there, and simplified the pyproject.toml change a bunch | 17:12 |
| fungi | though i suspect it's going to fail py36 testing, so this may be the point at which we ask whether we should drop support for that from bindep too | 17:13 |
| fungi | which i think also means dropping bionic testing | 17:15 |
| clarkb | why drop focal and jammy? is that just due to testing upper and lower bounds | 17:15 |
| fungi | yeah | 17:15 |
| fungi | maybe we can get by with just using an older min setuptools version for now | 17:18 |
| corvus | clarkb: fungi hi! happy new year! i'm declaring scrollback bankruptcy from before the holidays... has anyone set up a docker image mirror job yet? | 17:18 |
| fungi | i don't *think* so | 17:18 |
| corvus | i'll get started on that then | 17:20 |
| clarkb | I am not aware of one | 17:21 |
| clarkb | there was more discussion about options with openstack helm folks but I haven't heard if tehy did anything with that new role | 17:22 |
| corvus | did we decide to have jobs in the opendev tenant and put images in the opendev quay.io org for everyone, or did we want images needed by the zuul project to be handled by the zuul tenant and zuul-ci quay.io org? | 17:27 |
| clarkb | I want to say we said it was ok for common images (like mariadb) to be hosted by opendev but that the less common things should probably be project specific so that we (opendev) don't act as arbiters for that content | 17:28 |
| corvus | ack; i'll propose a change with all the images i think zuul-quick-start needs and we can prune those down and i'll move the excess to the zuul tenant | 17:29 |
| fungi | https://www.openwall.com/lists/oss-security/2025/01/06/1 is a great example of the sorts of bugs that continue to make nested kvm dicey | 17:34 |
| clarkb | are general protection faults generally considered security concerns? Not sure what the impact is that would lead to it being on that list | 17:35 |
| clarkb | (I would expect things to crash in the hypervisor and the host to kill the related processes and move on but maybe cleanup is dangerous?) | 17:35 |
| fungi | not sure, mostly just observing yet another situation where certain guest kernels can crash when doing nested kvm | 17:37 |
| clarkb | ++ | 17:37 |
| fungi | i meant "dicey" from the perspective of random job node crashes which we have no way to debug without access to the host | 17:38 |
| opendevreview | James E. Blair proposed opendev/system-config master: Add jobs to mirror dockerhub images https://review.opendev.org/c/opendev/system-config/+/938508 | 17:44 |
| corvus | i put that in system-config since it already had a secret, but we could make a new secret and move it | 17:45 |
| clarkb | corvus: I guess using a new secret might allow us to scope downt he access. but nothing is using that secret today so it should be fine for now | 17:47 |
| clarkb | there was a zuul syntax error fwiw | 17:47 |
| clarkb | corvus: also you should add the python images | 17:47 |
| corvus | ++ | 17:48 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 17:59 |
| opendevreview | James E. Blair proposed opendev/system-config master: Add jobs to mirror dockerhub images https://review.opendev.org/c/opendev/system-config/+/938508 | 18:06 |
| clarkb | I guess I can add ^ to the meetin agenda too | 18:06 |
| fungi | looks like pyproject.toml use with pbr under python 3.6 isn't going to work without more fiddling in pbr itself, i think? https://zuul.opendev.org/t/opendev/build/16d00668281b4857a8c933e0a063b307/log/job-output.txt#615-617 | 18:34 |
| clarkb | fungi: thats the thing that requires you to not use develop | 18:35 |
| clarkb | fungi: I worked around that in pbr with usedevelop=false for that specific environment | 18:35 |
| clarkb | its the whole pep660 support in pip but not in setuptools for python3.6 versions of those two tools problem | 18:36 |
| fungi | right, maybe nox does editable by default then? | 18:36 |
| clarkb | fungi: its in the noxfile.py: session.install("-e", ".") | 18:37 |
| clarkb | I think we can either try to lookup the session python version and drop the -e or add a new session that matches 3.6 specifically and install . without the -e | 18:38 |
| clarkb | oh no we can't redefined tests() so need to check the version and drop the -e | 18:40 |
| fungi | i guess you do it like a normal python version check you'd do in other python code | 18:41 |
| clarkb | https://nox.thea.codes/en/stable/config.html#nox.sessions.Session.python | 18:41 |
| clarkb | yup exactly something like if session.python == '3.6': no -e else: use -e | 18:41 |
| clarkb | or just stop using -e entirely | 18:43 |
| clarkb | the benefits for bindep are minimal since it installs so quickly you can just reinstall it | 18:43 |
| clarkb | in theory we could write a little wrapper function for session.install too that did those checks then we could share that around all the nox projects | 18:51 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 18:52 |
| fungi | i'd rather just clean up the workarounds once everything requires python 3.7 | 18:53 |
| clarkb | fungi: I think that comparison doesn't work when you're running nox interactively because session.python will be "3" in that case (matching the decorator parameter value). Things are setup that way so that you can run it locally and it will just use your default python3 version. but then in CI (or if you are explicit locally) you can force the python version separately | 18:55 |
| clarkb | fungi: maybe it should be if session.python != "3" and session.python <= "3.6" ? | 18:55 |
| clarkb | we can wait to see what CI says about this first though. But ya if you just run nox -s tests locally I think it will not use -e ? | 18:56 |
| clarkb | assuming your have a newer python | 18:56 |
| clarkb | corvus: I think now there is an indentation issue? | 19:08 |
| opendevreview | James E. Blair proposed opendev/system-config master: Add jobs to mirror dockerhub images https://review.opendev.org/c/opendev/system-config/+/938508 | 19:11 |
| corvus | oops | 19:11 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 19:32 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 19:51 |
| fungi | increasingly worried that newer setuptools is requiring a [project] table in pyproject.toml, but not sure why it didn't surface in the pbr changes if so | 19:54 |
| fungi | though before i go making assumptions, i'm trying to figure out why this didn't do an editable install: https://zuul.opendev.org/t/opendev/build/1841505ad6364d2883c2a1f76eddfefe/log/job-output.txt#622 | 20:04 |
| clarkb | I guess whatever the python version is is <=3.6? | 20:05 |
| fungi | oh, 3.12 < 3.6 yep! | 20:05 |
| * fungi grumbles | 20:05 | |
| clarkb | fungi: you should be able to print debug there. re [project] in pyproject.toml that is really odd considering the pbr changes never hit that | 20:05 |
| clarkb | maybe setting py-modules=[] overrode that? | 20:06 |
| fungi | but pbr does that in its own packaging and didn't hit the same error | 20:06 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 20:07 |
| clarkb | right I'm wondering if telling setuptools to not determine the moduels skops the need for a [project] block | 20:09 |
| fungi | i already have another edit ready for moving just about everything out of setup.cfg, but really hoping it's not necessary | 20:09 |
| clarkb | because maybe module auto detection requires a project name or something? | 20:09 |
| fungi | unfortunately as soon as there's a [project] table a lot of tools assume all metadata is either set there or declared as dynamic | 20:09 |
| clarkb | maybe it is enough to make things dynamic? | 20:10 |
| fungi | it's a list of around 10 different keys | 20:11 |
| fungi | i'll push that up wip just to see what we get | 20:11 |
| clarkb | and I guess all of these changes are newer than when we first starting poking at this with bindep and pbr | 20:11 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 20:12 |
| clarkb | fungi: note the key used in pbr pyproject.toml is py-modules not packages | 20:15 |
| clarkb | looks like the first pyproject.toml change for bindep is still failing | 20:15 |
| fungi | the error indicates both are synonymous, but maybe they behave subtly differently | 20:16 |
| clarkb | I don't see py-modules in the error output. Maybe you're looking at the docs? | 20:17 |
| opendevreview | James E. Blair proposed opendev/system-config master: Add jobs to mirror dockerhub images https://review.opendev.org/c/opendev/system-config/+/938508 | 20:17 |
| fungi | the original error about being a flat package with multiple modules, saying you should override setuptools' module search function in one of several ways | 20:18 |
| clarkb | ah | 20:19 |
| clarkb | the py36 job did pass | 20:19 |
| fungi | yeah, it's been passing | 20:19 |
| fungi | which suggests we're dealing with a behavior change in newer setuptools | 20:19 |
| clarkb | so ya this must be in newer setuptools. But PBR did test py312 so thats odd | 20:19 |
| clarkb | I think we may also need a __main__.py for when bindep is used as a module (that apepars to by what failed in unittesting?) | 20:22 |
| clarkb | no maybe its something else. Says bindep --version failed in your pyproject takes over change | 20:23 |
| clarkb | with no output so maybe that change is simply not correct for installing the console scipt? | 20:23 |
| fungi | well, the latest revision of 816741 is at least doing editable mode for py312, so there's one less question to resolve | 20:24 |
| fungi | the old setup.cfg still specified the entrypoint as bindep = bindep.__main__:main | 20:25 |
| fungi | i didn't change that, just moved it into pyproject.toml | 20:26 |
| fungi | and yes, bindep already has a __main__.py | 20:26 |
| fungi | but maybe it's not getting included in the package or something | 20:28 |
| fungi | if it's bindep --version failing, i suspect we're not installing pbr (listed in requirements.txt), so maybe the dynamic set for dependencies isn't working as expected | 20:30 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 20:35 |
| fungi | curious to see if that changes the picture | 20:35 |
| fungi | yeah, running build is failing to include any of the actual project python files | 20:47 |
| fungi | like pbr isn't actually supplying the manifest | 20:48 |
| fungi | so it's actually the entire project missing | 20:49 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 20:54 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Move stray release note into the correct location https://review.opendev.org/c/opendev/bindep/+/938522 | 20:54 |
| clarkb | fungi: thats odd because it did work when we originally pushed the code and pbr itself is able to usepbr to package itself | 20:55 |
| fungi | yeah, agreed | 20:56 |
| clarkb | is it possibly related to not requiring setuptools >=64? | 20:56 |
| clarkb | I wonder if it works if you do that | 20:57 |
| fungi | seems unlikely, since it would use the newest available version, but worth a try | 20:57 |
| clarkb | does it log what version it is trying to use /me looks | 20:58 |
| clarkb | doesn't look like it | 20:58 |
| fungi | no, pip is very thorough about hiding all possible output from build backend stuff | 20:58 |
| clarkb | the only other thought I've got is maybe require latest pbr | 21:00 |
| clarkb | since pbr itself is using the latest version of pbr when packaging itself | 21:00 |
| fungi | huh, okay now that's passing for 3.12 but not 3.6 | 21:05 |
| fungi | looks like it's missing the entrypoint script | 21:06 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 21:08 |
| clarkb | another event horizon between setuptools releases and pyproject.toml support? | 21:08 |
| clarkb | I really hope not because having one was plenty | 21:08 |
| fungi | yeah, i think so, that's another pep that landed around the 3.6-3.7 timeframe | 21:08 |
| clarkb | I guess the main thing to sort out now is why pbr is ok so that we're comfortable releasing pbr? | 21:11 |
| fungi | yeah, that's making me a little more weird, like maybe pbr testing still isn't entirely using the patched version | 21:11 |
| fungi | scarily, the latest 938520 is going all green | 21:12 |
| clarkb | pbr does refer to itself using backend-path | 21:12 |
| clarkb | rather than intsalling pbr via requires. Possibly something is happening via that to make it just work? | 21:13 |
| fungi | yeah, so 938520 is testing clean | 21:13 |
| clarkb | fungi: https://review.opendev.org/c/opendev/bindep/+/816741/11 this one passed except for py36 | 21:14 |
| clarkb | that implies to me that with a working noxfile to disable -e installs for py36 the rest of the content from ps11 should be fine? | 21:15 |
| clarkb | fungi: can we revert to the ps11 state but with the updated noxfile? | 21:15 |
| fungi | yeah, let's try splitting the rest of that into the later wip change, just a sec | 21:16 |
| clarkb | hrm build release also failed | 21:17 |
| clarkb | error: Multiple top-level packages discovered in a flat-layout: ['bindep', 'playbooks', 'releasenotes']. | 21:17 |
| clarkb | but why didn't that fail when running py312 unittests? | 21:17 |
| fungi | fwiw, teh only difference between ps 11-15 is tools.setuptools py-modules vs packages, i wonder if they're really not synonymous | 21:18 |
| clarkb | its certainly looking that way | 21:18 |
| clarkb | its almost like older platforms only respect the one key and newer platforms only respect the other | 21:20 |
| clarkb | I wonder if we can set them both and get away with both sets of platforms being happy | 21:20 |
| clarkb | (but also thats just a hack and my guessing at the problem) | 21:20 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Move stray release note into the correct location https://review.opendev.org/c/opendev/bindep/+/938522 | 21:20 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 21:20 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 21:21 |
| clarkb | the error says py_modules or packages | 21:21 |
| clarkb | maybe py-modules is not correct (it is what pbr used though) | 21:21 |
| clarkb | perhaps py-modules is old/deprecated and that explains it working on the older platform though | 21:21 |
| fungi | scarily, https://stackoverflow.com/questions/76663521/where-does-py-modules-go-in-pyproject-toml-setuptools has both doing different things | 21:22 |
| fungi | https://setuptools.pypa.io/en/latest/userguide/package_discovery.html#automatic-discovery | 21:23 |
| clarkb | I remember reading those docs a year or two ago and feeling just as confused when I was done /me checks if the feeling is any different this time around | 21:24 |
| fungi | yeah, i'm descending into the same madness again | 21:25 |
| clarkb | at a high level I feel like using the src/ layout would probably fix things for us | 21:26 |
| clarkb | but we'd have to fix all our imports within the project too likely | 21:26 |
| fungi | we'd have to change the contents of all our git repos to move everything into a subdirectory, yes | 21:27 |
| clarkb | https://setuptools.pypa.io/en/latest/userguide/package_discovery.html#custom-discovery seems like what we want if we want to keep supporting the existing structure but the docs there don't really help me undersatnd how to configure it | 21:28 |
| clarkb | "disable implicit namespaces" in their example doesn't map to the human readable text in the paragraph above | 21:29 |
| clarkb | or maybe it does but in a non obvious to me manner | 21:29 |
| clarkb | that said you updated your change to set the pcakage to bindep right? maybe thats the minimal change we need? | 21:30 |
| clarkb | fungi: in your latest patchset you renamed tools.setuptools to tool.setuptools. fwiw pbr uses tools.setuptools but the docs you just linked to use tool.setuptools. Maybe that is the problem? | 21:31 |
| fungi | oh, huh... | 21:31 |
| clarkb | rather than py-modules vs packages vs py_modules could it be the header name lookup instead | 21:31 |
| clarkb | if so in theory we can set both values maybe? | 21:31 |
| clarkb | and have a two sections that are redundant with one another | 21:32 |
| fungi | fwiw https://packaging.python.org/en/latest/guides/writing-pyproject-toml/ says it's "tool" not "tools" | 21:32 |
| clarkb | but also your latest ps fails py312 and not py36 | 21:33 |
| clarkb | ps11 was the other way around. This is super confusing | 21:33 |
| fungi | https://setuptools.pypa.io/en/latest/userguide/pyproject_config.html#setuptools-specific-configuration also says "tool.setuptools" | 21:34 |
| clarkb | Emit a warning when ``[tools.setuptools]`` is present in ``pyproject.toml`` and will be ignored. -- by :user:`SnoopJ` (#4150) | 21:35 |
| clarkb | tools.setuptools appears to be the old thing and tool.setuptools is the new thing. I suspect that py312 is working without that section entirely | 21:35 |
| fungi | huh | 21:35 |
| clarkb | but when you "fix" it to use the correct name tool.setuptools it breaks | 21:35 |
| clarkb | but then release-build still fails | 21:35 |
| clarkb | ok theory: when you drop the tool*section entirely it makes things "work" possibly because we don't need to pip install -e . when we're running all the code relative to . anyway | 21:37 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 21:37 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 21:37 |
| fungi | seeing what happens when it's removed | 21:37 |
| clarkb | but then release build breaks because the resulting package build is of nebulous utility as a package | 21:37 |
| clarkb | but ya lets reset to not having that section at all and see what ahppens | 21:37 |
| fungi | increasingly happy i keep bubbling the unrelated fixes ahead of these two | 21:38 |
| clarkb | I'm going to pop out for a bike ride now to take advantage of some nice weather. Interested to see what you learn. I'll also get the meeting agenda out sometime after I return and before the end of my day today. Feel free to leave notes here if you want additional edits. | 21:38 |
| clarkb | I think I'll add a note about corvus' change to mirror container images | 21:39 |
| fungi | sounds good. catch you later! | 21:39 |
| clarkb | fungi: do we also need to WIP the pbr release until we understand tools.setuptools weirdness better? | 21:39 |
| fungi | it's a pre-release, so i'm not worried about it | 21:39 |
| fungi | and it's still waiting on an openstack release team discussion about independent model deliverables being allowed pre-releases | 21:40 |
| clarkb | ack | 21:40 |
| fungi | if we decide pbr needs another commit, i'll either update the pre-release request or make a new one depending on what lands first | 21:41 |
| fungi | and if we decide it doesn't need any new commits, then no adjustments are needed to the pre-release request anyway | 21:41 |
| fungi | yeaj, looks like it's build-python-release that's going to be unhappy about things | 21:52 |
| Clark[m] | That probably implies the install we do in unit tests isn't really installing anything and we're relying on the local code path instead from th git checkout | 21:54 |
| fungi | editable installs, yes | 21:55 |
| fungi | we used to force at least one test to be non-editable in order to catch that, if memory serves | 21:55 |
| fungi | maybe build-python-release is serving that purpose now | 21:56 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 21:58 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 21:58 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 22:03 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 22:03 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Use PBR's pyproject.toml build-backend support https://review.opendev.org/c/opendev/bindep/+/816741 | 22:11 |
| opendevreview | Jeremy Stanley proposed opendev/bindep master: Evacuate most metadata out of setup.cfg https://review.opendev.org/c/opendev/bindep/+/938520 | 22:11 |
| fungi | i'm not sure i love that solution, but it's working | 22:23 |
| fungi | i guess i should stack another change dropping py36 support to see how it works out | 22:23 |
| Clark[m] | fungi: do we run build python release in PBR? If nnot adding that as a sanity check might be a good idea | 23:39 |
| Clark[m] | I'll look at the fix above soon | 23:39 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!