DNF/RPM Copy-On-Write Eyed For Fedora 34 To Speed Up Package Installation

[phoronix]

Phoronix: DNF/RPM Copy-On-Write Eyed For Fedora 34 To Speed Up Package Installation

Fedora 34 is shaping up to be another exciting Fedora Linux release on the feature front. Among the material to look forward to in this spring 2021 Linux distribution release is routing all audio through PipeWire by default, enabling systemd-oomd by default, an independent XWayland package, and more. The latest proposal involves making use of DNF/RPM copy-on-write support atop Btrfs with Fedora 34...

http://www.phoronix.com/scan.php?page=news_item&px=DNF-RPM-CoW-Fedora-34

[Britoid]

This will likely work on xfs too since it implements reflinks.

[phoronix_anon]

Why not just promote SilverBlue if that's what Fedora is converging on.

[skeevy420]

Why not just promote SilverBlue if that's what Fedora is converging on.

Have you ever used it? I have and it isn't ready for entry level users...maybe not even ready for intermediate level users either. If something doesn't work from a Flat or a container and you need it then you have to terminal it out to manage layers which make OS version upgrades a nightmare. I found it more tedious than anything and nothing that I'd want to subject a new user to Linux with. It's a great concept but it needs more fleshing out before it's ready for prime time.

[You-]

With a bit more focus I think they can get Silverblue ready. However I can no longer find a central place to track development, so I dont know how much focus is being given to it.

1. The update story needs some improvement. Currently you have to use the command line to see the ostree updates.
2. Codecs need some work. The base system doesnt have all the codecs, so you need the flathub flatpaks to use codecs. It would be good if somehow non flatpacked gstreamer could also search the locations that flatpak uses to store plugins and (after sandboxing), use them for the system gstreamer aswell.
3. Flatpaks are awesome. All the major browsers are also available via flatpak.
4. There is a bulletproof install there that will be very hard to bork for newcomers. When tinkering I have often managed to get to a stage of mangling with normal fedora where I needed to put some effort in to restore close to the default state. This is harder to do and for OEMS etc it should be something easier to support.

[skeevy420]

With a bit more focus I think they can get Silverblue ready. However I can no longer find a central place to track development, so I dont know how much focus is being given to it.

1. The update story needs some improvement. Currently you have to use the command line to see the ostree updates.
2. Codecs need some work. The base system doesnt have all the codecs, so you need the flathub flatpaks to use codecs. It would be good if somehow non flatpacked gstreamer could also search the locations that flatpak uses to store plugins and (after sandboxing), use them for the system gstreamer aswell.
3. Flatpaks are awesome. All the major browsers are also available via flatpak.
4. There is a bulletproof install there that will be very hard to bork for newcomers. When tinkering I have often managed to get to a stage of mangling with normal fedora where I needed to put some effort in to restore close to the default state. This is harder to do and for OEMS etc it should be something easier to support.

2 & 3, both combined and separate, were my biggest issues and the primary reason I needed to use layers. Combined, no matter what codecs I installed in Flat, Firefox would not work with Netflix, Hulu, Prime, etc so I always ended up having to layer on some codecs to get Firefox to fully work.

3, separate, was needed printing support and I couldn't get LibreOffice from Flat to print so I had to layer it on.

The problems with 1, for me, are directly caused by 2 & 3 and a bit of 4. If Flats just worked and I didn't have to putz with codecs or if I could have more easily created a custom base image like OEMs would want to do then I wouldn't have had to deal with the update issues.

[Anvil]

Why not just promote SilverBlue if that's what Fedora is converging on.

Silverblue is pretty much Dead, i dont think it ever really took off.

[pWe00Iri3e7Z9lHOX2Qx]

Silverblue is pretty much Dead, i dont think it ever really took off.

Any references for this or is it just assumption?

[Anvil]

Any references for this or is it just assumption?

an assumption . i dont think its Dead as such, it just never really took off the way Redhat/Fedora an possibly IBM wanted it to take off. you'd probably be better off to ask Matthew Miller more about silverblue, personally i think Fedora are doing to many Editions when they dont really have the resources to do it all. you'd be better off to get stats on how many people actually use an or download Silverblue.

[finalzone]

i dont think its Dead as such, it just never really took off the way Redhat/Fedora an possibly IBM wanted it to take off. you'd probably be better off to ask Matthew Miller more about silverblue, personally i think Fedora are doing to many Editions when they dont really have the resources to do it all.

Silverblue although usable is still under intense testing to solve some quirks like conversion from RPM packages to flatpak (still in progress) for desktop applications and switch to btrfs filename done on Fedora 33 release among them.
It will take several releases to declare it ready to replace the current workstation.