As already mentioned, UniPup runs in the initramfs, does not do a switch_root and does not have a layer f.s. on '/'. This is a very simple configuration, so I have rewritten the boot and shutdown scripts accordingly. The scripts are very simple.
I have implemented session-saving, just something primitive for now. I have been tossing ideas around in my head about the best way to save sessions, and I'm thinking that when the user installs packages, if the initrd.gz file is in a writable partition then offer to rebuild initrd.gz at shutdown with the new packages -- that way they will be compressed. Personal data, like stuff in /root and /etc, can still be saved to a normal save-file.
...yes, I might have a go at implementing that later today.
I'll do that, plus fix a couple of bugs, then upload the demo. I find UniPup quite exciting, as it is a move back to something simpler. The current live-CD is 73MB (Xvesa only, somewhat cutdown modules, no LZMA compression just gzip) which works fine in a PC with 256MB of RAM.
Comments:Posted on 7 Jul 2008, 14:18 by Jesse
UniPup Sounds Cool!
This UniPup idea sounds very cool!
With the simpler approach it lends to a system more able for dynamic changes, swapping .sfs files, user save files, and potentially UniPup kernels/ramdisks as well with the kexec kernel function.
I'm looking forward to an evaluation testing edition!
Posted on 7 Jul 2008, 20:49 by urban soul
This IS exiting. Particularly the hybrid model unionfs/rebuild-initramfs.
One idea I had in mind when testing SliTaz was to avoid rebuilding initramfs to be compressed - just because I am annoyed to wait at boot. SliTaz gives you the choice if/or how compression will take place.
BTW I sent you the files some hours ago.