Partitions would definitely be the most logical approach, but I figure, if a user has trouble following the incremental steps I’ve written to perform small updates themselves, I’m going to assume that they would find it even harder to follow instructions on writing an individual partition.
It would be a tech support nightmare, worse than the initial official 0.4 → 0.5 upgrade patch. It was an amazing concept in theory, just not the most “regular” user friendly.
You’ve probably noticed that I’m also primarily using the classic /usr/local/bin and /usr/bin file locations. Most of the config information ends up in ~/.config, which would no doubt need to reside on the system partition. This would make overwriting the partitions wipe any custom settings.
On top of that, most apps that are in ~/apps/emulators also store their config files, save states, screenshots and save files in this directory, and would also be in the system partition. Essentially, I’d also need to make a custom script to back up all of these files, taking into account every file extension that could possibly exist, and then returning them to the correct place.
Just backing up the entire ~/apps/emulators directory would be an easy approach, however that’s where the majority of my updates end up going. Essentially the process would involve a lot of user manual intervention, which would result in “regular” users making mistakes - something I’d need to personally spend even more time trying to answer people getting frustrated.
It’s funny! I actually wanted to do this back in my DEOT v1+ image!
So you’re definitely right on the money for what I want. It’s just a matter of delivering it in the easiest way for a non Linux user to be able to execute. I was going to go the two release path, ie one image with a partition and one without, but doing upkeep on that would be a lot of work. It’s already enough using my current workflow setup.
I’d need to multiply that by two, which is going to be a lot of work. Ideally it would be good to have a team who could work on things together. That would actually be an interesting proposition, although if that were to be a thing, it would probably be better to use an image such as your archlinux as a base and make it an entirely community driven collaboration from scratch.
As much as I would love to have it based on the official image, I simply don’t know how to get in contact with the official devs who are in charge of official releases. Guu has been extremely helpful, but it does sound like he needs to answer to powers above him, following guidelines that they have in place.
Back on topic, another thought I had was to have ALL of the contents of non user contents tarballed up, and have users transfer a massive file over to their gameshell, and run a script to essentially manually replace the contents of ~/apps, ~/launcher and /usr. They’d lose the config files and customisation etc I mentioned above, but would keep all of the contents of ~/games and ~/music.
It would also take a ridiculously long time to do, and would be plagued with user errors breaking the system. The amount of trouble and time it takes would make if worthwhile to just have users backup their files. Unless they’ve got a Linux system, this takes an eternity over Ethernet or wifi.
Likewise if I have everything in a .deb file, I need to be able to trust that a user can follow instructions perfectly. Another thought I had was to simply make my own warehouse repository, with scripts people can run to incrementally build up their stock OS to include whatever features exist on the forum. Think a Debian/Ubuntu style GUI installer.
I’ll be honest. I get scared of commitment, and thrive on instant gratification of user happiness. I just feel that the amount of work to do any of this would be so much, for so little reward. I am 100% on board with what you’re saying and suggesting. I just don’t want to be reinventing the wheel, and fragmenting the community more than it already is.
If anything, I’m hoping this thread will bring people together, and get the attention of the official devs. If that means they release an official image with partition support, I would be on board. Of course if we do a good job of implementing it here, it could do exactly that, and they could use what we have done as a base. Eg, look at how Lima was fixed up in the DEOT v1 thread. My fear is that I’d spend the time doing it, then officially they release something that effectively does the same thing, but does it using a different method/file structure.
I’d also realty want to work with you, using your arch Linux OS as a base, and making it much more of a user friendly experience. It is so powerful and clean, that it would be easy to work with. The clockwork OS is kinda all over the place, almost held together with sticky tape. It’s hard for users to know how to easily edit things.
So! In short. I want to have a separate data partition SO badly! But I have many fears. 