#asahi-dev · 2026-10-15 11:28
chadmedhahaha#asahi-dev · 2026-10-15 11:26
chadmedit just so happens in the case of our webapp the "frontend" is the backend to another frontend :P #asahi-dev · 2026-10-15 11:25
chadmedthe core is going to have all the stuff required to mutate the state of the machine and each frontend to it can implement as much or as little of that as it wants#asahi-dev · 2026-10-15 11:25
chadmedDo It Once, Do It Right(tm)#asahi-dev · 2026-10-15 11:24
chadmedchaos_princess: dont think its important rn but i am pulling all of that out of the cli frontend so that its easy to do eventually#asahi-dev · 2026-10-15 11:24
chadmedsofus: thought that stuff might belong in a subsequent PR after you cleaned it up and fixed the commit messages, but i can do that myself if you want a la your fixup commit#asahi-dev · 2026-10-15 11:22
chadmedoops missed some drm api changes #asahi-dev · 2026-10-15 11:20
chadmedPartition probably should not be a pure dataclass when a bunch of methods on InstallerMain are dedicated to getting/setting data on an instance of Partition #asahi-dev · 2026-10-15 11:19
chadmedinstaller refactor is coming along. there are a lot of "interesting" choices#asahi-dev · 2026-10-15 11:18
chadmedjannau: rebased interchange and opened a PR#asahi-dev · 2026-10-14 12:16
chadmedwhich youre going to need regardless #asahi-dev · 2026-10-14 12:16
chadmedit seems like its only major direct dependencies are fastapi, pydantic and websockets#asahi-dev · 2026-10-14 12:15
chadmedwe're gonna have to pull a bunch of nonsense in anyway#asahi-dev · 2026-10-14 11:39
chadmedyou just throw json messages in one end and they come out the other and you can do whatever you want with them #asahi-dev · 2026-10-14 11:38
chadmedfastapi comes with one#asahi-dev · 2026-10-14 11:29
chadmedits dead simple#asahi-dev · 2026-10-14 11:29
chadmedstart with a small driver for something simple and work your way up. in the context of these devices, the usual advice is to take the m1n1 experiment for the keyboard led pwm and try to build a linux driver for it without looking at ours#asahi-dev · 2026-10-14 11:18
chadmed"we designed this for a very very specific problem only a company like google could have, anyway we're gonna force it on you all and theres nothing you can do about it"#asahi-dev · 2026-10-14 11:18
chadmedlook at all the places protobuf has managed to find itself in #asahi-dev · 2026-10-14 11:18
chadmedyeah its pervasive, not just a frontend thing#asahi-dev · 2026-10-14 11:14
chadmedeverything is over-abstracted, over complicated, heavy, homogeneous, theres a new cargo cult every 3 minutes #asahi-dev · 2026-10-14 11:13
chadmedi have never been a fan of how googlers design software, and googles utter dominance over the web for the past ~15-20 years has left their dna all over everything#asahi-dev · 2026-10-14 11:09
chadmedmy hate for modern web dev would not fit in TON 618 #asahi-dev · 2026-10-14 11:02
chadmedhere be dragons :P #asahi-dev · 2026-10-14 10:41
chadmed^#asahi-dev · 2026-10-14 09:05
chadmedlets go#asahi-dev · 2026-10-14 08:40
chadmedthat product also will return an empty 200 ok even if nothing happened. you then have to go to its own error logs in the product and see what you did wrong#asahi-dev · 2026-10-14 08:39
chadmedyeah another product we deal with does something like that#asahi-dev · 2026-10-14 08:38
chadmedreal nonsense i have to deal with daily#asahi-dev · 2026-10-14 08:38
chadmeds/instance/reference but whatever you get the idea#asahi-dev · 2026-10-14 08:38
chadmedif youre a backend dev and you do this fuck you #asahi-dev · 2026-10-14 08:37
chadmed200 OK "{"error_code": "ISM_4000", "message": "Object instance not an instance of an object"}"#asahi-dev · 2026-10-14 08:34
chadmedwhy yes i would like my api to return 201 when nothing was created thank you $product#asahi-dev · 2026-10-14 08:34
chadmednot the $dayjob way#asahi-dev · 2026-10-14 08:34
chadmedwe do things properly#asahi-dev · 2026-10-14 08:34
chadmedno #asahi-dev · 2026-10-14 08:33
chadmedfor the web one it will yield some bytes that look like json or whatever needs to happen #asahi-dev · 2026-10-14 08:33
chadmedfor the cli it will be the mess of p_error() calls#asahi-dev · 2026-10-14 08:32
chadmedyeah the python side of the frontend will catch the core's exception and then turn that into whatever#asahi-dev · 2026-10-13 11:40
chadmedill add the "required" frontend functions to the core which if not implemented will raise an exception #asahi-dev · 2026-10-13 11:39
chadmedadding the web one will require flask or whatever but also not that hard#asahi-dev · 2026-10-13 11:39
chadmedthe cli frontend is easy, we just copy paste the existing functions from the main class into the new one and move some stuff around to call back into the core instead of do things like os.system() directly#asahi-dev · 2026-10-13 11:38
chadmedim going to rip the core out and make it a skeleton-ish superclass with all the base functionality that will be subclassed by a particular frontend#asahi-dev · 2026-10-13 11:37
chadmedthats okay its not hard to decouple #asahi-dev · 2026-10-13 11:37
chadmedthere is.... a lot more coupling than i remember :P #asahi-dev · 2026-10-13 11:37
chadmedim looking at refactoring the core#asahi-dev · 2026-10-13 08:17
chadmedthey blew some fuses on your machine when they knew it was being shipped to you #asahi-dev · 2026-10-13 08:15
chadmedour installer might be better than the macos one atp #asahi-dev · 2026-10-13 08:08
chadmedthe fix is in main (it just disables that optimisation pass for windows targets lol) #asahi-dev · 2026-10-13 08:08
chadmedhopefully the next llvm point release fixes the seh data alignment thing