Phoronix: Unity: Very Intrusive & A Nightmare To Maintain
Along with the discussion around a rolling-release version of Fedora Linux, having been discussed recently has been the possibility of providing Ubuntu's Unity desktop as an alternative desktop environment for Fedora. This is obviously a topic that gets some riled up...
Although, I'm not a Fedora packager/developer, I'm planning on porting Unity to Fedora. A few Arch Linux users (City-Busz, thn81, AUR packagers) and I have already successfully ported Unity to Arch Linux. I do agree that Unity is hard to maintain though; at the moment, 86 packages need to be compiled (on Arch Linux) for the complete experience
The most annoying part is not that GNOME packages need to be patched, it's that Xorg needs to be patched for uTouch/XInput 2.2, although that will change once Xorg 1.12 is released.
Last edited by chenxiaolong; 02-03-2012 at 02:10 PM.
Nice anti-unity piece, think this should be followed up with a new GNOME Shell is hated by everyone post too.My take - at the time I was doing it, we were actually fairly close, just got stalled on the nux package review and then I lacked time. I'm not aware of anything that would completely roadblock getting Unity in Fedora. There was no need for a variant GTK+, at least when I was looking at it (and to the best of my knowledge). The issue which required a patched glib was also resolved, I believe. The remaining area where non-upstreamed patches were 'required' wasn't really a requirement; these were patches to support Ubuntu's libindicator indicators. Unity will work without these, it's just not exactly as upstream intends the experience to be. As far as I'm aware, Canonical were reasonably good about proposing the libindicator patches for upstream inclusion, but many upstream projects - especially those that are part of GNOME - weren't exactly rushing to adopt the patches. I think Canonical did try to implement libindicator support as a plugin for apps with sufficently sophisticated plugin frameworks, which obviously helps. There may have been changes since the last time I looked at things, of course. Note that I only ever did this as a personal side project. Even though I'm @redhat.com it's not a Red Hat project, it's not supported or paid by RH. It also has no kind of 'official' standing within Fedora, it's not like I proposed it as a feature and got anyone else to sign off on it. I just upped and started packaging things, entirely on my own initiative. I don't consider my efforts to have any more status than anyone else's. http://lists.fedoraproject.org/piper...ry/161803.html
Last edited by pdffs; 02-03-2012 at 08:14 PM.
I don't want Ubuntu to die -- Ubuntu basically is Linux in the public's eye, and so Ubuntu's reputation becomes Linux's reputation -- but I think they need to give up on Unity, or completely rewrite it to use as much pre-existing infrastructure from other projects as possible. Canonical is more than welcome to build their own UI, but they need to work from existing components, and contribute to upstream projects when they want to make things more flexible or add new capabilities that they want to make available in their OS. If the upstream won't cooperate, then just fork their libraries/apps and make your fork so good that they eventually get merged, like egcs did.
Unity as-is will eventually eat Ubuntu whole, because it's a huge schism with the rest of Linux distro ecosystem, and Ubuntu isn't strong enough on its own to pull the entire app development community behind it.
I keep hoping they'll retire Unity, but it seems like Ubuntu is digging in deeper and deeper. The worst part of it is that they could probably replicated 90% of the "Unity experience" just by extending Gnome Shell. While I dislike them myself, the Linux Mint Gnome extensions are a good example of what can be done in a short time. Personally I think Gnome Shell + a few choice extensions creates an amazing desktop. Unity? Not so much.
[QUOTE=allquixotic;249414... or completely rewrite it to use as much pre-existing infrastructure from other projects as possible.[/QUOTE]
This is what was driving me crazy as I watched the OpenSUSE Unity project fail. OpenSUSE spent a Google Summer Of Code project taking their library that allows YaST to run as either a GUI or an NCurses interface and making it as distro-agnostic as possible so that other distros could use it for their installers. Benefit to OpenSUSE: zero. This is in addition to turning their OpenSUSE Build Service into Open Build Service so other distros could use it, making other tools available, making the theme of their conference that year cross-distro collaboration and inviting representatives from other distros to attend to see where they could work together, etc. I remember thinking... why couldn't Canonical devote a GSoC project to making Unity more distro-agnostic, or as you put it better, more compatible with upstream projects? I know I'm relatively new to the world of Linux and all, but it seems like some distros are giving away all of their "secret sauce", while others are doing quite the opposite. At least offer a "I'll clean up my code if you clean up yours" deal or something.... trade the Puppy people some distro-agnostic code for a better package manager in exchange for info on how they squeeze so much into such a little space, for example....