Sven Slootweg (soft-deprecated) @joepie91@pixie.town

@UntaggedTransmitter Do you know of any particularly good switch options?

@UntaggedTransmitter I mean, I can do the soldering part, but I dread accidentally breaking something in the process of opening it up, given how aggressively anti-repair a lot of hardware is nowadays

@raine (But then that would've required FOSS folks as a category to understand that there's a broader ethical component to the whole FOSS thing and well, we've seen how well that went)

@raine Yeah it does, public licenses were a neat little legal hack in their time but they're *really* not a very good tool for the long term, and the FOSS world really should've worked to build better structures and tools for this sort of stuff over the years 😕

@raine I'm not sure this would be possible; the GPL, like any public license, can only impose restrictions on usages that are normally prevented by copyright, so either "AI" training is a derivative work under copyright (and then the license already covers it) or it isn't (and then there's no way the license can make demands in the first place)

@starfire Sounds like a similar situation to mine, I guess 😅

@starfire But I definitely would not be buying Logitech mice if it weren't for that

@starfire I mean, I've had generally much better outcomes with mice from other brands, just none of them make trackballs with a viable ergonomic configuration for me (and are also for sale in NL)

@FunkyBob@chaos.social I've done this in the past for some personal projects, though possibly that was a smaller scale than you're looking for.

It mostly involved 'converging' the styles first (identifying similar things and combining them into a set of shared semantic rules that's smaller and less duplicated in total), and then once that was done, changing the implementation of those rules.

The convergence step served mainly to prevent "oops I overlooked this one place that only partly shares this set of rules and now I accidentally broke it" by making sure that semantically similar elements actually use a similar set of class names.

The process is not that different from how I do this for other code refactoring, though in this case 'similarity' is determined by an element's purpose in the layout or design and visual style, rather than the logic of a function.