On 10 February 2018 at 21:54, Felix Miata <mrmazda@earthlink.net> wrote:
https://lists.opensuse.org/opensuse-factory/2017-01/msg00367.html
That looks a lot like the way DNF works in practice, which IME is awful, filling up filesystems with rpms, then complaining they cannot be installed; often impossible to fix if interrupted.
Then you have fundamentally misunderstood the benefit of transactional-update Any issue, any failure, any problem with the upgrade leads to the snapshot being discarded. This can be detected during the update, or after the update on next boot. No data stored on disk. zero space in use. zero changes made to the running system. Nothing to fix, because nothing changed. Very different from the typically snapshot arrangement we have on 'traditional' *SUSE systems (yes, I am now referring to btrfs rootfs as 'traditional' - because it's old compared to this stuff). In that case we have to take snapshots before and after any update to take into account other activities on the rootfs, post install scripts, other services running, other user processes, and the like and that approach eats up space, and has more things that can go wrong with it. Transactional-update just uses 1 snapshot at a time, and has no need for historical ones (unless you really want them for some reason), so it's infinitely more efficient.
Welcome to the 21st century, we have new technologies, built to address the problems with your primitive systems from the dark times.
We think you'll find it interesting here if you adopt our modern ways ;)
Adopting requires adapting, which requires time. I don't much care for allocating time to fixing what ain't broke, like many who have been around more than 6 decades, and value what's left of our time more than we used to. I have more than enough excess of todos over round tuits. Many things work as well as they did days, weeks, months, years, decades or centuries ago, at least as well as new, ostensibly improved versions complete with their brand new bugs, and even when not as well, well enough to stick with, with or without reasons that have anything to do with the technology involved. My car, newly acquired 28 months ago, is 2 decades old. My house, which dad built when I was 13, is 53 years old. CD players had been around at least 5 years before I stopped buying vinyl. DVD player/recorders had been around more than a decade before I bought one. I still have no mobile phone or pay TV, and use daily the 12' satellite dish I bought 30 years ago. OS/2 was over a decade old before I switched to it (instead of Win98 or NT) from DesqView. I test drove SUSE roughly 5 years before using it to replace OS/2. EXT4 was mainstream probably a decade before I started using it instead of EXT3, and still I usually don't use it except for / filesystems. Installation targets and /homes here remain on rotating rust exclusively. Is the tail wagging the dog?
BTRFS here would require magnitudes of changes in the way I do things, not the least of which are disk space allocation and backup/restore procedures. All of my installations are on multiboot PCs. Most of my installations have freespace somewhere in the 0-25% range on the / filesystem and have been installed without recommends, or separate filesystems for any but /home and /usr/local. I cannot see snapshotting on a single storage device in my foreseeable future.
See my post titled 'Ancient History' and share your thoughts there, I think that's a better topic for such philosophical discussions. I think it's fair to say I find your mindset hard to grasp and think that you've had plenty of time to start getting to grips with this new stuff. I mean, it's not like you don't have much spare time - the amount you post on this mailinglist suggests you have an abundance of it ;) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org