On Thu, Jul 2, 2015 at 3:26 PM, Carlos E. R. <robin.listas@telefonica.net> wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
On 2015-07-02 20:18, Chris Murphy wrote:
On Thu, Jul 2, 2015 at 10:51 AM, Carlos E. R. <> wrote:
It can take days!
This can still be ugly in certain cases, but it's immensely better with Windows 8.1 than previously. The recovery partition reset is not to be casually discarded,
Having a recovery partition (which is an old trick) is cheating.
Why? It's how Windows has been installed itself for years now. Even if you aren't using the OEM utility, and use Microsoft's installer on today's UEFI computers, you get a recovery partition. It's not cheating, it's how the OS works by default.
You can not compare installing Linux to installing Windows if one of them is installed from scratch and the other from an image.
I can compare them. One is closer to a stateless system, the other isn't. It's a completely fair comparison when discussion how OS's should be installed, and reset.
And I have heard of recovery partitions in machines sold with SLED, from HP I believe.
All users should have this option, not just people buying HP computers.
nor the refresh feature of Windows 8.1 which returns the system to a near factory state while keeping user data and settings intact. That alone is vastly better than anything we currently have with any Linux desktop distro.
You have something similar, even more powerful, with btrfs snapshots.
It is certainly going in the right direction. It's not as simple as just having two options: full reset (obliterate everything except the OS), vs system refresh (clean up temporary files and most user and system settings, keep user data). But Windows NTFS has had shadow copy snapshots for quite a while also, so they've been doing system update rollbacks for a long time now.
The functional equivalent of reset is the same as on Android and iOS.
But you need a third party app, like AppMgr III, to then recover all your previously installed apps, which is what the OP wants.
It is a matter of granularity that needs improvement. At least on my Cyanogenmod phone, apps I've installed are separate from the OS, so I can to major OS upgrades and rollbacks by changing that OS image, without having to reinstall apps or user data. But there isn't sufficient granularity between non-OS apps I've downloaded and my data.
If you have a branded system (Dell, HP, ASUS), it's a matter of plugging in a service tag into their web site, and you get all the recommended drivers for that system to download.
You do not need that in Linux, as all the drivers are distributed in the kernel. But if you did, why don't those brands do the same with Linux?
I don't understand the question.
I've done this and the whole thing takes maybe an hour or two. Not days.
And will it install for you the video driver, the printer driver, nero, the TomTom drive, java, flash, firefox, thunderbird, libreoffice, etc? Sure? With a single reboot?
Windows, no that's probably 8 reboots. OS X yes, and if its in the app store then you also get all of those things updated automatically. Again, GNOME Software makes this quite a bit better. The one thing I don't like are offline updates, where by it reboots to do the update and then reboots again after the update. Even Windows and OS X on OS updates close out the current user session, perform the update, then reboot once. So this could be better right now. Going forward: Btrfs snapshots + containers to get an atomic "offline" update with a user defined reboot when it's convenient for them. This would look like: 1. Create container 2. Launch updates daemon 3. Updater snapshots the file system and downloads updates. 4. Applies updates to the snapshot and cleans up. 5. User is notified there are system updates and to reboot when convenient to take advantage. 6. On reboot, the subvolumes containing the updated system is booted. So this way the software update is completely atomic rather than updating a live in-use system and libraries. So in use things are just yanked out from under running programs. The update can completely implode spectacularly badly and yet, the current system is totally unaffected, it could even crash the container and things would be OK. To clean up, simply automatically delete the bad snapshots. And very nicely one reboot only. And at the users discretion. We're not too far from this. -- Chris Murphy -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org