00:00 <romangg> But there is a tendency to water down that split and let the daemon talk to the compositor via proprietary other means (config file writes etc) in order to quickly fix issues.

00:05 <romangg> I'm trying to argue against and for a more long-term approach with clear interfaces to also make integration with other indepedent compositors more and more possible.

00:13 <romangg> So that's also the reason why I was asking here what other people are thinking about such interfaces and if there was discussion in the past about the topic.

06:57 <whot> romangg: just ftr, I floated the idea of a libinput-config helper library in montreal and it wasn't that well received, primarily because at this point every compositor already has some input configuration interface, so it'd be like slotting yet another later in. it would've made sense 5-6y ago

06:59 <whot> romangg: that's also the reason why it has diminishing return, you'd be putting in the effort to switch the compositors over(or add support for it) and OTOH get all the corner cases supported that aren't handled by the current approach

07:00 <whot> romangg: and, and this is the real crux here - you'd also have to figure out what the GUI to it will be. this *may* be omitted if you just re-use the current guis but then it's a lot of work just to stand still because you won't get any new features this way

13:12 <pq> romangg, good luck! I'll be casually interested in what you achieve there, but I can't participate. Weston does not have any runtime configurability, but some would be nice.

13:16 <pq> ad hoc solutions are likely awful to use and hard to maintain, which why I'd prefer sharing some infrastructure or tools there, but compromising flexibility may be a hard sell - if the "standard" solution is not good enough, then one would be stuck with a custom one anyway and have to maintain two systems.

17:07 <romangg> pq: haha, thank you. Indeed it's difficult. But I think at some point one has to start at least with it and it will become important long-term. Either that or there will be a single standard desktop in the end on Linux. Which might have advantages, but also disadvantages.

17:37 <romangg> Ok, that's a bit exaggerated. But in any case it does interfere with the dynamic and flexibility of the platform. Which is a great asset.

17:48 <autistic> romangg: I'm just wondering, as someone absent when this conversation started, what are you proposing? If you don't mind my curiosity