Proposal For Gnome To Become Linux-Only
Moderator writes "Could Gnome drop support for non-Linux operating systems? That was a recent proposal on the Gnome mailing list, although there were significant objections in response. Quoting: 'It is harmful to pretend that you are writing the OS core to work on any number of different kernels...the time has come for GNOME to embrace Linux a bit more boldly.'"
I support this because it can only help to make Gnome more irrelevant.
It's open source. If there are people who want it on other platforms, they can just fork it. Right?
This is a dumb idea for software architecture reasons, too. I'll explain.
When writing a Windows application, you must recognize that the interface between your application and TODAY'S version of Windows must remain fluid such that you can support changes delivered by patch or by OS release. This is known formally as "decoupling" and it is necessary to isolate big systems that need to communicate. Decoupling is important for unix applications as well, because kernels change over time and APIs vary slightly between unixes.
If you truly believe your application gains anything by eliminating a decoupling library/layer, you have missed the point of the past few DECADES of object-oriented programming.
If I were a non-Linux dev who'd contributed to Gnome, I'd be seriously considering a fork no matter what the outcome of this is. If there's one thing I've learned from working on open-source projects, it's that once the Linux Zealots' radical proposals start gaining real traction it's time to bail.
And KDE is a ugly mess of a turd as well right now.
Why cant we get a nice polished and WORKING desktop UI group of programs that has a full click and drool interface that is complete and working? Why is it that every time they get close, KDE or Gnome, they decide that a radical change is needed that utterly breaks everything?
Gnome3 is complete fail because you have to revert to 1998 and editing text files to configure it. WTF is that?
KDE on the next release is looking to be the same. Hello? release it as early beta and do NOT call it a release until all the tools that are used to make it useful to noobs are ported to the changes.
Instead we get a mature interface that is abandonded and will not support it anymore because they all moved to the new shiny.
The days that desktop environments are only GUIs and only consisted of a bunch of windows that paint stuff on the screen are long over. These days desktop environment handle a lot more lower-level stuff, and users rightfully expect them to do so. Think for example user interfaces for managing hardware, system settings (user accounts, security, firewall, wired and wireless network), etc. GNOME depends on various background daemons that must be started at boot. All of these things have system-dependent mechanisms. Configuring the wireless network is completely different between FreeBSD, Solaris and Linux. All 3 of those OSes have a completely different init system, completely different firewall system, etc.
1. Resize Konsole on Nvidia. Kernel panic. /var/tmp/kdecache
2. Change network config, watch kded consume 100% CPU.
3. Unplug, go to power saving profile, watch panel lose transparency, have to clean out
4. Connect to an SMB share with Dolphin and watch it crash every single time.
I found those in the first half-day with KDE4.6. And that's my best track record with KDE4.
It's interesting: KDE4 looks fundamentally strong, but seriously lacks polish and has serious usability trouble. GNOME gives the opposite impression: a kludge of technologies underneath the hood, but polished mercilessly. The two projects could really teach each other a a thing or two, and a combination of the best of each would give MacOS and Windows serious discomfort.
--srj/mmv
Gnome went from being the most usable, stable, "just works" DE for unix-like systems, to a steaming pile of crap, IMHO. I'm still in shock that they took a stable, functional foundation that was Gnome 2, and just literally threw it all away. I tried to give Gnome 3 a chance, but it's like a damned cell-phone UI.
You are doing it wrong.
1. If you are an adminstrator of any worth you can do it without X via command line.
2. Almost all enterprise Applications that are fairly new are Web Based
3. There are other just as affordable or more affordable remote access "thin client" solutions available.
X11 is an aged and out of date protocol. It had its use, today it is a dinosaur. Just because you work on badly managed enterprise or aged model, it doesn't mean everyone else does.
If something is so important that you feel the need to post it on the internet... It probably isn't that important.
Too many layers of abstraction creates problems when you need to repair bugs or troubleshoot issues.
Because the abstractions hide what's really going on, the more abstractions you add, the less you know, and too many layers ultimately makes the computer more opaque to both developer and user.
And ultimately results in a buggy mess, when you get bugs and it takes years to trace due to the massive number of wrappers you may have to pass through to ultimately figure out where the bug might be.