Ubuntu 16.10: Convergence is in a holding pattern; consistency’s here instead

Still no Unity 8, but the latest Ubuntu iteration delivers reliability we wanted in 16.04.

Read the whole story

 

[rndmairs]

You don't have to have drastic changes to make Ubuntu a success:

Kramden Institute

edit: first attempt to add a link...

 

[Leiesoldat]

I'm pleased to see a MATE edition for Ubuntu. I run Linux Mint MATE on virtual machines both at the office and at home. I'll reconsider my flavor choice now that I know there is this flavor on the official Ubuntu side.

 

[gmerrick]

I've encountered issues with 16.04 that have kept us for the time being on 14.04.

One of my main irritants, not related above, is that you now need to manually set monitor configurations manually for every user. It used to be that xorg.conf would do it for everyone, but that apparently is not used any more. Similarly with /etc/network/interfaces. You could manually configure IP addresses and other information here, but that seems to be no longer parsed.

 

[UndeniablyRexer]

For anyone curious what the naming convention will be after 17.04, a (the?) discussion can be found here:
https://wiki.ubuntu.com/DevelopmentCodeNames#A17.10

 

[a_random_guiy]

The question no one seems to ask: why is "convergence" a useful goal?

If you plug your phone into a monitor and keyboard, there are actually three choices:

1. You continue with the "phone" interface. It's just bigger, and doesn't make optimal use of the display.

2. Your phone OS uses a different display framework, for example, to allow individual apps to run in individual windows on the larger display. No change for the individual apps, each of which still things it is running on a phone.

3. Convergence. Everything changes: your phone now works like a laptop. But this means that the app developers must write apps that can tolerate this paradigm change.

As far as I can see, #3 (convergence) is the worst choice. Most app developers will not want to support two interface paradigms in their apps. Those that do? That's called "feature bloat": building extra crap into your app that will be useful to a tiny fraction of your users a tiny fraction of the time.

 

[raxx7]

gmerrick[/url]":qfic7p6n]I've encountered issues with 16.04 that have kept us for the time being on 14.04.

One of my main irritants, not related above, is that you now need to manually set monitor configurations manually for every user. It used to be that xorg.conf would do it for everyone, but that apparently is not used any more. Similarly with /etc/network/interfaces. You could manually configure IP addresses and other information here, but that seems to be no longer parsed.

Check the NetworkManager.conf. IIRC, it should have this, so it doesn't mess with the configuration from /etc/network/interfaces

[main]
plugins=ifupdown

[ifupdown]
managed=false

 

[daarong]

Oh weird, I hadn't noticed daily non-optional popups coaxing me to upgrade. I guess I'll wait for it to download in the background and slip itself in with other miscellaneous upgrades... ahhhh... right, that's Windows, not Linux

Seriously though. This all looks good, I think I'll give it a try. My issues with 16.04 have been mostly resolved, with minor lingering oddities. Possibly 16.10 will resolve those.

 

[raxx7]

a_random_guiy[/url]":1bbt0chr]The question no one seems to ask: why is "convergence" a useful goal?

If you plug your phone into a monitor and keyboard, there are actually three choices:

1. You continue with the "phone" interface. It's just bigger, and doesn't make optimal use of the display.

2. Your phone OS uses a different display framework, for example, to allow individual apps to run in individual windows on the larger display. No change for the individual apps, each of which still things it is running on a phone.

3. Convergence. Everything changes: your phone now works like a laptop. But this means that the app developers must write apps that can tolerate this paradigm change.

As far as I can see, #3 (convergence) is the worst choice. Most app developers will not want to support two interface paradigms in their apps. Those that do? That's called "feature bloat": building extra crap into your app that will be useful to a tiny fraction of your users a tiny fraction of the time.

Consider Firefox or Opera browsers: they have versions for PCs (Windows, Mac, Linux) and phone/tablets (Android, iOS, Windows Phone).

Each version requires a lot of platform specific work -- not only the different GUI paradigms, but vastly different underlying platforms.
Yet, these developers obviously saw the value in doing that work, because all of these versions exist.

Convergence is about reducing the amount of platform specific work such developers must do.

 

[kokovios]

I am still with the 16.04 version. I have to say I am completely satisfied with Unity 7 and I don't understand the nagging with Unity 8 / Mir. The current desktop looks and behaves great. I think that taking the time to be sure that the next desktop works reliably is more important than rushing to put out a half baked release. The only thing that is missing, is an official flat theme, but there exist many good ones by the community.

A tip for the author. This article is focusing too much on version numbers, to the level of individual components, something that is interesting only to hardcore linux users. I would suggest to review the OS more functional wise.

 

[fatpugsley]

gmerrick[/url]":g89vwzmz]I've encountered issues with 16.04 that have kept us for the time being on 14.04.[...] Similarly with /etc/network/interfaces. You could manually configure IP addresses and other information here, but that seems to be no longer parsed.

Oh the joys of the systemd migration and the completely dysfunctional NetworkManager.service!

This release also sees some significant updates to the GNOME components and applications that Unity depends on, bringing most of the GNOME stack up to version 3.20 (a couple appear to be at GNOME 3.22,which is downright cutting edge by Ubuntu's backporting standards).

Some say updates, others say regressions. For the last few years updating Ubuntu Desktop has turned to a game of "Show me where the GNOME 'designers' touched you". My favorite in 16.04 was the all-new "Videos" app that DOESN'T SUPPORT PLAYLISTS.

 

[Be Jammin]

I'm gonna have to guess at why Geuss is considering Yakkety Yak.

(The link under the first image doesn't link to an article)

 

[hizonner]

But the difference is that GNOME 3 isn't tied to Wayland and has been cranking out impressive releases for some time now while Unity 7.5 is feeling, well, a bit dated.

If Gnome 3 is good, then I want to be bad. Seriously, if you actually want to get some work done and make use of your large screen, then it's the steamiest steaming steamer ever to steam a steam. AND it's ugly. Not that Unity looks any better.

Also, 50 points off for calling a UI "dated". Function, not fashion.

 

[Derecho Taketh Away]

Does this version have the Radeon support lacking in 16.04?

Edit/observation: A downvote for a sincere question? (Somebody's Lucky Charms didn't have enough pink hearts in it this morning...)

 

[KonoWatakushi]

Anyone know if work is being done to extend HiDPI support beyond integer only scaling?

 

[Lily Liliaceae]

kokovios[/url]":cc984e64]I am still with the 16.04 version. I have to say I am completely satisfied with Unity 7 and I don't understand the nagging with Unity 8 / Mir. The current desktop looks and behaves great. I think that taking the time to be sure that the next desktop works reliably is more important than rushing to put out a half baked release. The only thing that is missing, is an official flat theme, but there exist many good ones by the community.

A tip for the author. This article is focusing too much on version numbers, to the level of individual components, something that is interesting only to hardcore linux users. I would suggest to review the OS more functional wise.

Nah, focusing too much on numbers would be actually giving the version numbers of stuff like Mesa, DRM, X Server, LLVM, Pulseaudio, SystemD, etc. all of which reveal the degree of functionality the distribution contains to a finer degree.

 

[Deleted member 1]

2017 - still not the year of the linux desktop

 

[pokrface]

I'm gonna have to guess at why Geuss is considering Yakkety Yak.

(The link under the first image doesn't link to an article)

Yeah, something really weird is going on with that link (there's a specific caption box for story intro images, and for some reason, that box is mangling URLs). Working on it with Jason and the Tech Argonauts right now.

(Link is supposed to go here, FWIW.)

edit - ok, we fixed it, I think. The problem was due to weird handling of (what else?!) double quotes.

This reminds me of a time I tried to write a bash alias for a command that included both single and double quotes. It was like staring Cthulhu in the face. I could feel madness eating my soul. I worked on it for probably fifteen minutes before realizing that I was better off just giving up and trying to do what I wanted to do a different way.

 

[ILikeTastyFood]

In Canonical's defense, the competing display server project, Wayland, hasn't exactly taken the world by storm.

Except that all LG television sets use Wayland since years. That alone is already an install base of millions in ordinary people’s homes (see http://developer.lge.com/webOSTV/sdk/lg ... e/emulator and search for "wayland" for proof)

The compatibility layer with Android applications in ChromeOS also uses Wayland: http://www.phoronix.com/scan.php?page=n ... s-ChromeOS

Last I checked Chromebooks sell rather well.

 

[Ainamacar]

Speaking of virtual machines, Ubuntu 16.10 is noticeably faster in VMs than previous releases. This is primarily due to the new Unity Low Graphics Mode, which tones down the fade effects and transparency in favor of faster rendering times.

Fortunately this has already been added to 16.04, and it did make a big difference on my Hyper-V VMs. The transition wasn't so smooth for me, though: blank desktop syndrome on my updated VMs that the usual tricks didn't seem to fix. Fresh installs worked alright, so I ended up migrating. To the devs' credit, I filed a bug and they resolved the problem on the original VMs in August.

 

[deviltry]

bothered[/url]":1yjxap39]2017 - still not the year of the linux desktop

Year of the linux desktop is current_year + 1, so you're wrong - 2017 will be the year of the linux desktop

 

[Goofball_Jones]

I dabble in Linux. Currently been using it as my daily OS for a month now, but using Arch with the Cinnamon desktop. So far it's been great. Even play many of my Steam games on it. Nice rolling release with no problems (so far).

EDIT: Oh, the reason I made this post is to say that yes, it IS the year of LInux on the desktop, if you really want it to be. Actually, been that way for a few years now. The only thing that doesn't run are specific applications like the Adobe stuff.

 

[K0DE]

I was able to get it working pretty well in a virtual machine, but on the hardware site it didn't work.

Maybe "the hardware side"?

 

[lewax00]

There's plenty in Ubuntu 16.10 that makes it worth the upgrade

Nope. Not on this laptop. I've learned, upgrading (K)Ubuntu is always a bad idea. It's always followed by a day of trying to figure out why the hell it won't boot properly, and how to get it working again (usually every update finds a new and exciting way to break the graphics system...).

I'm sticking to 16.04 until the next LTS at minimum.

EDIT: This kind of sounds like my usual upgrade pains:

neither one of these very modern, well-specced pieces of hardware can successfully boot to Unity 8.

 

[Deleted member 1]

kokovios[/url]":1lz95c3o]I am still with the 16.04 version. I have to say I am completely satisfied with Unity 7 and I don't understand the nagging with Unity 8 / Mir. The current desktop looks and behaves great. I think that taking the time to be sure that the next desktop works reliably is more important than rushing to put out a half baked release. The only thing that is missing, is an official flat theme, but there exist many good ones by the community.

A tip for the author. This article is focusing too much on version numbers, to the level of individual components, something that is interesting only to hardcore linux users. I would suggest to review the OS more functional wise.

I'm on 16.04 too but I have several issues with Unity 7.

For one, the graphical style seems to be all over the place. Windows use dark borders but light themes for subtabs and content (like the file manager etc). Even in the terminal which is dark itself, so the new rounded tabs look really ugly. Even the light square ones from 14.04 were better.

Personally I think they should have gone with the dark Gnome theme instead of the light one considering the entire interface is based on dark colours.

Also the icon bar is still fixed to the left like it was on the Netbook Remix originally (where admittedly it did make sense with the small widescreen netbooks when you needed every vertical pixel you could get!) and still can't be placed at the bottom of the screen. On a big 24" screen there is no need to have it at the side. Edit: Sorry, I just realised that since 16.04 this is now possible with a shell command. It's much easier this way with multi-screen setups because the bar doesn't go through windows overlapping both screens then.

Not sure if Unity 8 fixed these issues but really there's many points to improve.

 

[raxx7]

KonoWatakushi[/url]":3lh0wagp]Anyone know if work is being done to extend HiDPI support beyond integer only scaling?

Short answer:
"nay".

Longer answer:
Even before HiDPI was common, there has been a lot of work in making the GUIs DPI independent.
Eg, there's a text-scaling-factor setting (or whatever the name is now), which can be an arbitrary value, and most of Gnome's desktop and applications will honor it and scale accordingly.

But, for example, they won't scale pixmaps.
So, you end up with things like a very nicely scaled Evolution with some tiny icons (because they're pixmaps).
Or a nicely scaled image viewer, but the image (at 1:1) is tiny because it wasn't scaled.

HiDPI support on the other hand, can scale the pixmaps so things look good but it only does it in integer ratios.
Non-integer ratios are a bit of a can of worms, AFAIK.
Note than even OS X only supports 1.0, 1.5 and 2.0, not arbitrary scaling.

 

[David T.]

Xubuntu FTW. XFCE is so simple.

 

[theoilman]

A convergence OS is something I really want, but I've long since given up thinking Canonnical will be the ones to bring it to market. One because it's already two years late with no real indication it'll be ready soon, and two because I've come to the conclusion it really needs to be a mobile first OS bringing in desktop functionality rather than the other way around. Hypothetically you could do it from either direction, but it's just easier to do right coming from mobile.

 

[David T.]

theoilman[/url]":14kymt74]A convergence OS is something I really want, but I've long since given up thinking Canonnical will be the ones to bring it to market. One because it's already two years late with no real indication it'll be ready soon, and two because I've come to the conclusion it really needs to be a mobile first OS bringing in desktop functionality rather than the other way around. Hypothetically you could do it from either direction, but it's just easier to do right coming from mobile.

Can you say Chrome OS/Andromeda?

 

[theoilman]

David T.[/url]":1b4ril28]

theoilman[/url]":1b4ril28]A convergence OS is something I really want, but I've long since given up thinking Canonnical will be the ones to bring it to market. One because it's already two years late with no real indication it'll be ready soon, and two because I've come to the conclusion it really needs to be a mobile first OS bringing in desktop functionality rather than the other way around. Hypothetically you could do it from either direction, but it's just easier to do right coming from mobile.

Can you say Chrome OS/Andromeda?

Andromeda sounds interesting but so far it's just rumors. I don't doubt Google has been working on it, but whether they actually decide to keep developing and launch is a whole other issue. They could easily have already decided to scrap it.

 

[pjcamp]

Unity is still brain damaged. I'll stick with Mint.

 

[brokkr]

Proper systemd user sessions are - weird as it sounds - pretty tempting. Now that all the brouhaha has died down, I find I actually like doing things the systemd way.

I have an unsightly collection og user specific services spread out over various .config/autostarts, systemwide systemd services that drop privileges and even various desktop specific options, like openbox's autostart script. Properly implemented, this would make things a lot cleaner and easier. Hopefully this is across desktops? I believe the Lubuntu team sometimes opts out of stuff just to be different :-/

 

[jasonridesabike]

How long has it been with Mir now? 4 years? I've entirely given up hope in seeing it this side of 2020, but it seems a yearly humiliation to constantly promise 'next release' and miss the target. It's shaken my confidence in the abilities of Canonical.

 

[pokrface]

Proper systemd user sessions are - weird as it sounds - pretty tempting. Now that all the brouhaha has died down, I find I actually like doing things the systemd way.

I have an unsightly collection og user specific services spread out over various .config/autostarts, systemwide systemd services that drop privileges and even various desktop specific options, like openbox's autostart script. Properly implemented, this would make things a lot cleaner and easier. Hopefully this is across desktops? I believe the Lubuntu team sometimes opts out of stuff just to be different :-/

Starting out in the systemd world seems pretty nice. But, after upgrading a couple of servers that started out as 12.04 boxes and that I upgraded a couple of years ago to 14.04, I can say that all my shit was broken as hell on that first post-16.04 upgrade reboot. It took a couple of weeks of evening work to rip out all the cruft and un-fuck everything, and I'm still not sure everything's really working right (I'm having batshit crazy problems getting services to start in LXC containers, for example).

In ~6 years of using ubuntu for my servers, this was far and away the most painful upgrade ever. But, hey, on the upside, I got a crash course in how systemd works, so I guess that's good.

 

[petertirrell]

Mention of kernel improvements so that Skylake is better....how does that affect Kaby Lake? I haven't run a linux desktop in awhile but am approaching being the market for a new laptop. One of the considerations I'm making is linux support since I'm starting to itch to get back to developing on a linux platform again. If I get a newer laptop am I just in for more headaches?

 

[Bhairitu]

David T.[/url]":2iv1rk3e]Xubuntu FTW. XFCE is so simple.

That's why I run Ubuntu Studio. Unity was just not an interface for developing software.

 

[daneren2005]

Pokrface[/url]":1hx8bgwu]

Proper systemd user sessions are - weird as it sounds - pretty tempting. Now that all the brouhaha has died down, I find I actually like doing things the systemd way.

I have an unsightly collection og user specific services spread out over various .config/autostarts, systemwide systemd services that drop privileges and even various desktop specific options, like openbox's autostart script. Properly implemented, this would make things a lot cleaner and easier. Hopefully this is across desktops? I believe the Lubuntu team sometimes opts out of stuff just to be different :-/

Starting out in the systemd world seems pretty nice. But, after upgrading a couple of servers that started out as 12.04 boxes and that I upgraded a couple of years ago to 14.04, I can say that all my shit was broken as hell on that first post-16.04 upgrade reboot. It took a couple of weeks of evening work to rip out all the cruft and un-fuck everything, and I'm still not sure everything's really working right (I'm having batshit crazy problems getting services to start in LXC containers, for example).

In ~6 years of using ubuntu for my servers, this was far and away the most painful upgrade ever. But, hey, on the upside, I got a crash course in how systemd works, so I guess that's good.

I'm glad (sad?) to hear I'm not the only one who lost a crap ton of time on that update. Holy crap was 14.04 -> 16.04 painful on servers. It took me several hours to just get a successful boot. And I knew that systemd would be the new default, but I guess I just assumed they would have some sort of migration which would make old Upstart scripts still work instead of just nuking everything from orbit.

 

[brokkr]

Pokrface[/url]":xlwi0vz9]Starting out in the systemd world seems pretty nice. But, after upgrading a couple of servers that started out as 12.04 boxes and that I upgraded a couple of years ago to 14.04, I can say that all my shit was broken as hell on that first post-16.04 upgrade reboot. It took a couple of weeks of evening work to rip out all the cruft and un-fuck everything, and I'm still not sure everything's really working right (I'm having batshit crazy problems getting services to start in LXC containers, for example).

In ~6 years of using ubuntu for my servers, this was far and away the most painful upgrade ever. But, hey, on the upside, I got a crash course in how systemd works, so I guess that's good.

He, at least Ubuntu made the transition in a new release. I was on Arch at the time they switched and IIRC the announcement was made in an obscure mail thread along the lines of 'Does anybody object? No? Alright, then...' What I learned was to read the contents of upgrade messages. Well, after I learned how to boot a system without a working init, of course.

EDIT: 'system', not 'systemd'

 

[raxx7]

Arch is a rolling release... so that's how things roll.
Same thing for Debian unstable/testing or Gentoo.

Major changes just pop up.

 

[issor]

For years I've been Ubuntu for desktop Linux and CentOS for servers. I've maintained RHEL based systems for years. Just a few weeks ago I had the opportunity to reimage/refresh my home server operating systems (one router, one NAS/camera server) and I went Ubuntu 16.04. I did so because 1) they've now got ZFS out of the box, whereas I had been pulling RPMs and manually maintaining those, and 2) It seems easier to find a lot of software packages for Debian-base (e.g. unifi stuff). So far I've been pretty happy.