| 28 Jun 2026 |
 Lach | Hi
There is a pretty annoying regression on staging (and now in staging-next), which causes the systems with overcommit enabled to OOM during build: https://github.com/NixOS/nixpkgs/pull/535423
Python cryptography package expects allocation + filling of 4TiB of memory to fail with allocation error, but with overcommit enabled this just causes OOM
Can this patch be accepted to staging-next? | 20:25:09 |
| Lach | This patch was already merged upstream | 20:25:51 |
 hexa | 
Download | 20:27:54 |
| hexa | roadblock | 20:27:58 |
| Lach | (Didn't knew anyone looked at checkmarks)
Marked first, llms were not used so the second | 20:29:52 |
| hexa | we absolutely do | 20:30:42 |
 Winter | why would we have them if nobody looks at them? 😅 | 20:42:59 |
| Lach | Most of the time I'm seeing them empty :D | 20:45:09 |
| hexa | we should just autoclose if nothing is checked tbh | 20:45:36 |
 emily | have had thoughts about splitting the checklist into "things relevant for every PR" / "things new contributors should look at" | 21:01:07 |
| emily | and hooking automated triage off that | 21:01:14 |
| emily | tbh we could probably drop a few of the extraneous checklist items | 21:01:36 |
| emily | everything between built on platform and CONTRIBUTING.md is pretty inconsistent value | 21:02:35 |
 Grimmauld (any/all) | I'd love a "review template" for that stuff, but i don't think github can do that... | 21:07:37 |
| emily | the thing about the CONTRIBUTING.md checkbox is that nobody is actually going through every provision in it every PR to check whether they're in compliance | 21:08:34 |
| emily | but we definitely want new contributors to read it | 21:08:42 |
| emily | my thinking is that the PR checklist itself should be minimal and focused on the high-importance stuff you should be checking every PR, and then first-time contributors get a bot reply with extra checkboxes for stuff we want people to learn about at the start | 21:09:36 |
 kuflierl | To be fair, i did the first 2 times or so. I don't anymore since i think i now most of it | 21:09:37 |
| emily | and the reward for those is you get the PR assigned to a round-robin new user contribution triage team | 21:09:49 |
| Grimmauld (any/all) | how would the new user interact with the bot checkboxes? You need committer to edit things | 21:10:42 |
| kuflierl | I like this. Reading it the first time was a hasle. I would have loved having a checklist + links to further questions | 21:10:48 |
| kuflierl | * I like this. Reading it the first time was a hasle. I would have loved having a checklist + links to further questions and deep dives | 21:11:28 |
| emily | yeah, the actual docs also ideally need reworking to be more readable and have a better hierarchy of information / gradual disclosure of detail according to importance… | 21:12:03 |
| emily | true. we could probably figure something out though. links or whatever. GHA can do custom buttons and stuff | 21:12:26 |
| emily | or the bot could just edit it in to your PR message | 21:12:36 |
| kuflierl | We also might want to have better linking. Something like a central file for all questions that branches out into the correct files | 21:13:15 |
| kuflierl | I always have issues finding something and rely on search engine foo | 21:13:39 |
| kuflierl | The answer tends to be "somewhere in the repo" | 21:14:31 |
| hexa | python-updates was mostly fine | 21:14:31 |
| hexa | one numpy 2.5.0 merge that didn't test jack shit is now causing all kinds of deprecationwarnings and api breakages | 21:14:59 |
