| 4 Aug 2022 |
 Winter (she/her) | that should be something brought up with the python folks tbh | 03:43:18 |
| Winter (she/her) | exactly | 03:43:20 |
| Winter (she/her) | i'm sure it can be worked around somehow, or things can be put in place | 03:43:31 |
| Winter (she/her) | the python folks are trying to keep up with something so fast moving and prone to breakage, there's only so much they can do without communication | 03:44:00 |
 tpw_rules | like sandro said in the issue, it feels more like there's a disconnect (which i don't know about) between the general python maintainers and cuda maintainers, and the best and most prompt solution might be to downgrade the service level | 03:44:06 |
| Winter (she/her) | * the python folks are trying to keep up with something so fast moving and prone to breakage, there's only so much they can do | 03:44:09 |
| Winter (she/her) | precisely yeah | 03:44:21 |
| tpw_rules | i.e. focus on the packages around ZHF and accept that they're not stable and might take lots of work to get going if you follow unstable | 03:44:33 |
| Winter (she/her) | (well the latter doesn't apply to samuela since he's a maintainer, so he kinda has to keep up in that regard) | 03:45:12 |
| Winter (she/her) | (hope that doesn't sound rude, not trying to be) | 03:45:27 |
| tpw_rules | well that's the question, does he? | 03:45:35 |
| tpw_rules | there's too much value here to just give up, but the small focus and large effort might warrant saying that as the maintainer of this particular ecosystem | 03:46:38 |
| tpw_rules | idk if it's simpler to sort of batch fixes like that, the hope is that these headaches will be discovered and worked out by other distros | 03:47:05 |
| tpw_rules | and upstream collaboration/natural progression | 03:47:56 |
| tpw_rules | maybe more of a defined maintenance interval. i saw another discussion today at automatically marking packages as broken if they don't build for 3 months and stable releases are every 6 months | 03:50:07 |
| tpw_rules | but on the other hand "i logged into my system and upgraded my nixpkgs and now i can't do my job" as a daily occurrence seems to me like upgrading less frequently is a reasonable solution. i'm not sure how often this actually happens though, but it's definitely not rare | 03:50:59 |
| tpw_rules | i also feel a lot of breakages with my nixos-m1 work and cross compilation. i try to do a release and oh whoops some random commit no one thought much of broke everything. at least it's not a job thing but it's still frustrating | 03:52:31 |
| tpw_rules | but it not being a job thing it's easy for me to frame it internally as a health evaluation and to try again next week and see if it's still broken, or take a bit of time to file reports. that way it's more reliable for people for whom it IS a job thing | 03:54:00 |
| Winter (she/her) | yeah | 03:54:17 |
| Winter (she/her) | i do think a way forward could start with talking to the python folks, to figure that stuff out; if this is a frequent enough occurrence, it's definitely worth a shot. | 03:55:11 |
| Winter (she/her) | anyways i've said that like three times now so | 03:55:26 |
| tpw_rules | from what i understand that's happened before but i don't know if it went anywhere and i'm not sure why. maybe there needs to be a mediator | 03:56:02 |
| tpw_rules | * from what i understand (e.g. what sandro said on the issue) that's happened before but i don't know if it went anywhere and i'm not sure why. maybe there needs to be a mediator | 03:56:17 |
| tpw_rules | i have seen some messages here from some of the nixpkgs python maintainers as heads ups but idk if samuela saw them or they were helpful | 03:57:10 |
| tpw_rules | in any case, it's about my bedtime. thanks for your input | 03:57:47 |
| tpw_rules | (another semi-related case i just remembered, https://github.com/NixOS/nixpkgs/pull/174072#discussion_r879694247 . i don't love the idea of having a bot update those libimobiledevice packages for safety reasons and the fact that they're all a tightly integrated set, maybe some cuda stuff can get added to that list too) | 04:00:28 |
| tpw_rules | or a better method than a list in some repo no one's ever heard of :P | 04:00:40 |
| Winter (she/her) | i think people have heard of r-ryantm's repo :p (or, can find it pretty easily since it's linked in its PRs) | 04:02:16 |
| tpw_rules | ok maybe it would be more fair to say "source file" | 04:03:01 |
| Winter (she/her) | also: unrelated, but i also would recommend adding an update script (and adding it to the skiplist if you want), to make maintenance easier -- it really doesn't hurt to have update scripts for everything possible, even if not ran automatically, IMO | 04:03:29 |
