00:12 <d1b2> <esden> ok, I have pushed the glasgow, prom adapter addon that we were working on during the stream today: https://github.com/esden/glasgow-addons/tree/master/hardware/prom-addon

05:23 <ktemkin> whitequark: I don’t believe so, at least from a language-lawyer perspective

05:24 <ktemkin> IIRC, you can technically only reject set_interface if you only support the default or if the alt-setting doesn’t exist

05:25 <ktemkin> (in *practice* you can get away with more, since if a set_interface times out, I believe the major OSs will just keep the old alt-setting)

05:27 <ktemkin> e.g. https://usercontent.irccloud-cdn.com/file/JGU1EFie/IMG_0692.PNG

05:35 <ktemkin> the intention of the specification is that mutually exclusive components be collected either under parallel configurations or parallel altsettings; and that interfaces were considered orthogonal to each other

05:36 <ktemkin> (originally, there wasn’t intended to be a way for a logical function to span multiple interfaces; IIRC IADs actually came as an ECN)

05:42 <whitequark> ktemkin: right. so you can probably guess what problem i'm solving

05:43 <whitequark> WDF not supporting dynamic configuration changes

05:43 <whitequark> it's either STALL in response to Set Interface, or a vendor request that magically disables one of the interfaces

05:44 <whitequark> makes basically no difference in practice other than one of the choices being very slightly nicer from aesthetic perspective

05:44 <whitequark> (i think it makes no difference, anyway)

05:52 <ktemkin> yep; I know that problem well enough

05:54 <ktemkin> there isn’t a good answer; but I’ve leaned towards “overpopulate interfaces and use control requests to handle e.g. stalling other interfaces”, since I know that at least won’t change if the kernel decides to change how undefined behavior is handled

05:55 <ktemkin> depending on what you’re using the interfaces for, you might be able to use SET_FEATURE/CLEAR_FEATURE to turn the relevant interfaces on and off

05:58 <whitequark> ohhhh thanks

05:58 <whitequark> that works

17:22 <marcan> whitequark: remind me what _whitenotifier-f is running?

21:11 <miek> marcan: https://notifico.whitequark.org/