#asahi · 2023-02-13 03:05
marcannrmntq: just log out and back in. render scale does not get updated for running apps, that's why it look blurry.#asahi · 2023-02-12 06:57
marcanI would assume any recent dock would use the USB 3.0 chips, so no upstream support#asahi · 2023-02-12 06:57
marcanyeah, for USB 2.0 displaylink only#asahi · 2023-02-12 06:56
marcancorrection, there is some support... not sure if for the latest chips though.#asahi · 2023-02-12 06:55
marcanI would not recommend it#asahi · 2023-02-12 06:55
marcanchadmed: there is no displaylink driver upstream I think, so getting whatever downstream modules you need for that would be its own adventure#asahi · 2023-02-12 02:17
marcannrmntq: also will never work on Xorg due to legacy X11 limitations, you need Wayland to see the Fn key#asahi · 2023-02-10 04:51
marcanwe probably need to ctypes setxattr since the macOS version has an extra argument we probably need#asahi · 2023-02-10 04:39
marcanmostly just 64K chunking and a little header, all stashed into the resource fork#asahi · 2023-02-10 04:39
marcanlooks pretty simple tbh, we can probably just do it in python#asahi · 2023-02-10 04:35
marcanwe already use ctypes to poke Apple's LZFSE implementation to extract kernelcaches. if AFSC using LZFSE isn't much more complicated on top of that, we could implement it in pure python#asahi · 2023-02-10 04:34
marcanso it sounds like it's probably better to reimplement exactly what we need into a separate tool, or better, python module#asahi · 2023-02-10 04:34
marcanfrom what I can tell it can't, without some roundabout thing using archives and another (ramdisk?) APFS partition which seems really silly#asahi · 2023-02-10 04:27
marcannot sure if it can do that#asahi · 2023-02-10 04:27
marcanin-place compression is useless because you still need to stage the uncompressed file first, which defeats the purpose#asahi · 2023-02-10 04:26
marcanrequirements for that to happen: a build of that tool we can bundle with the installer (must also work from recoveryOS), and that the tool supports copy-compression#asahi · 2023-02-10 04:26
marcanhowever, I *do* want to implement APFS compression because that will allow us to stage version upgrades safely by having both versions at once#asahi · 2023-02-10 04:25
marcanjmr2, dgb: I am not comfortable shrinking the recovery partition further, because APFS seems to really hate small filesystems and will fail volume creation and things like that. So it won't help with that.#asahi · 2023-02-09 12:54
marcanannie444: Nothing is taken down#asahi · 2023-02-06 07:47
marcanso I'm in no rush, more interested in getting the speaker safety stuff worked out this week ;)#asahi · 2023-02-06 07:46
marcanmy M2 Pro Mac Mini arrived this morning and so did Lina's M2 Max MBP, though I expect most of the work this time will be the new DCP stuff that jannau's been working on with the M2 Mini already#asahi · 2023-02-06 06:46
marcan(more if there are major changes, less if it's the same SoC and just DT changes)#asahi · 2023-02-06 06:46
marcansamurai: no, you should generally expect about a month before new models are supported (give or take)#asahi · 2023-02-03 05:45
marcanit will work some day, but there is no ETA.#asahi · 2023-02-03 05:44
marcanblender requires a newer GL version than is currently supported (and in particular it requires features which absolutely aren't supported and won't be for a while), so it is not expected to work#asahi · 2023-01-22 04:59
marcanHotswap: https://github.com/AsahiLinux/asahi-scripts/blob/main/initcpio/hooks/asahi#asahi · 2023-01-22 02:48
marcanso chromium is presumably also broken on everything else that isn't amd/intel/nvidia#asahi · 2023-01-22 02:48
marcanI'm assuming the "funny" behavior there is just standard kmsro behavior (i.e. the same issue affects ~every embedded GPU)#asahi · 2023-01-22 02:47
marcantpw_rules: they are hardlinks, it's just the way mesa works (all the dri drivers are hardlinks to the same file)#asahi · 2023-01-22 02:46
marcanI don't understand why they even whitelist individual dri drivers? Just whitelist the whole directory?#asahi · 2023-01-21 06:01
marcanthen we'll need ublock.ko#asahi · 2023-01-21 06:00
marcanchadmed: technically apple themselves could do that via DCP syslogs :p#asahi · 2023-01-21 03:19
marcanwe don't develop desktop environments#asahi · 2023-01-21 03:18
marcanwhat about it?#asahi · 2023-01-21 03:18
marcanwhat canonical thing?#asahi · 2023-01-21 03:18
marcanit doesn't make much sense#asahi · 2023-01-21 03:18
marcanwhat are they going to do, pressure me to put ads into... kernel drivers?#asahi · 2023-01-21 03:18
marcanthis is an OS porting project#asahi · 2023-01-21 03:16
marcanI fail to see the connection#asahi · 2023-01-21 03:15
marcanwhy would that have any influence over the project?#asahi · 2023-01-21 03:15
marcan...adtech?#asahi · 2023-01-21 02:33
marcanCyrinux: it just needs the ifdef, just haven't had a chance to look at it again.#asahi · 2023-01-19 17:12
marcanI should get some sleep then :) (was about to go to bed)#asahi · 2023-01-19 17:12
marcanyeah, that should be enough (might want to restart your login manager too since that itself might have some components stuck on the old broken mesa)#asahi · 2023-01-19 17:04
marcan(you also get a slightly newer mesa, I *think* there's no weird regressions on that one since it's been in -dev for a while)#asahi · 2023-01-19 17:03
marcanmkurz: done, please update and confirm it works#asahi · 2023-01-19 16:58
marcanon it.#asahi · 2023-01-17 16:42
marcankjoonlee: key repeat works fine here on a TTY on an M2 MBP (not sure if you can adjust it, haven't tried).#asahi · 2023-01-17 09:36
marcan(watch the DTs for the iack-offset once M2 Pro/Max drop, that will tell us the max die count supported :p)#asahi · 2023-01-17 09:35
marcansomething something 4x M2 Max