So why not make it a checkbox? This is something that screams give users an easy option and choice. One that is supported and tested. Many users like the gnome/mac style, many users like the windows/cde/etc. style and it is really hard to adjust to the other style.

It only takes a minute to leave. I think you have too much confidence that they'll come back. Maybe some will, I didn't, I doubt I will.

There seems to be the point of view that having good defaults means you can't have options. These are not mutually exclusive ideas. I might like the default font, but changing the font size is not some crazy esoteric need. By all means give me a desktop that needs no tweaking, but give me the option of changing some settings if I want. And extensions don't count, especially not for preferences and settings. Like don't suspend on lid close, font size, alt-tab behavior, panel placement, and panels, etc.

Having tried KDE 4.1, 4.2, etc. that was not the problem, at least for me. Panels losing options, Kasbar disappeared and I never did figure out what the "semantic desktop" was. But hey, plasmoids everywhere and configuration was no longer easy. I'm still looking for the spiritual successor to KDE 3.5.

Kasbar went away. Panel options like user hiding went away. KDE made design choices that negatively impacted my use (and now lack of use) with KDE. Wasn't that it was buggy, it became something that I didn't want, full stop.

True. But I had more faith in falling sales fixing the interface than I ever do of GNOME fixing its interface; or KDE bringing Kasbar back.

"Intractable" is a word. Two or three successful forks (depending on how you count Unity) and the main employer of most GNOME developers ships their pay OS with classic mode enabled over shell. GNOME limits options (settings kill kittens) to make life easier. Except we have gone from two dominant desktops w/ Gnome 2 and KDE 3, to one dominant desktop with the move to KDE 4 ,to no dominant desktop with GNOME 3. Now to use Linux I have to consider: Unity, Cinnamon, MATE, XFCE, LXDE and E. Two more if you are new to Linux and haven't experienced GNOME and KDE. How many GNOME devs have linked to the tired saw about how "Linux isn't about choice." Apparently it is all about choice now, starting with your DE.

It is all academic at this point. GNOME could make a 180 reversal but I don't see users coming back. I haven't tried KDE since 4.4 or so; at some point you just permanently lose most of your former users. You could make life easier for your users who do muddle though with extension by adding a couple of dozen options, and it isn't hard to see what those options would be. But even if I could convince you, Day, McCann, Bassi, etc. would never go for it.

It really is an epic fail. GNOME is suppose to be for inexperienced users but ZDNET today suggests migrating from XP to Cinnamon. http://www.zdnet.com/why-linux... At one point I helped an uncle install KDE 3 on his computer. Now no one in the family runs Linux. Linux wasn't even a fleeting thought a couple of years ago when my grandmothers computer died. Windows 8 was out and we paid extra to put her on Windows 7. Maybe MATE or XFCE could fill the gap, but GNOME would be out of the question for her. I doubt she is alone, is your family running GNOME?

I guess I just don't get the mentality. I loved KDE 3 because it was easy to customize and had great choices. Kasbar was the killer app (especially combined with configuring panel behavior). I could control the behavior of windows with a quick right click on the appropriate task. I dislike KDE because of the lack of choice. I can't easily remove what I don't want, I don't have the options I do want. When GNOME devs hold up KDE as too customizable I am staggered because in the ways I care about, it isn't customizable enough.

I've been thinking more about this and have to say, one of your weaker arguments. Your argument seems to be that extensions suck, but might suck less in the future. It sounds like you agree that extensions are inferior to checkboxes for the user when handling preferences but that the trade off for developer ease is worth it. The issue being code complexity and testing. But you then write about how short the extensions are and that they are code reviewed. So there is a trivial amount of complexity, honestly most of the code is already there it is just difficult to set the preference. And your users want the testing. Extensions suck because they aren't tested, they interfere with each other, they break; but then you say we should try them if we don't like stock.

So as I'm still having my coffee this is less lucid then I would like; but which is it? Is this a tiny bit of code that Gnome developers just can't bother adding or is it truly complex, in which case we can expect extensions to be a mess and make the entire system a tottering mess?

Please tell me that your last line about users abandoning Gnome if you add preferences was a joke or a throw away line. Specious argument, and while easy to deflate I'd rather not take the time. If you have your "perfect defaults" no one would even have to see that you had preferences anyway. And actually maybe not specious, I don't think it is even remotely plausible, everyone has been screaming for preferences.

http://www.ioccc.org/

But yes, the attack vector is the least of my worries. You write: "Putting in prefs and checkboxes also increases code complexity." True, having the extension increases the code complexity by at least the same amount. But no testing, no planning, no updates, no review. Just an unorganized mess of hundreds upon hundreds of extensions that conflict.

Gnome has the reputation, and for very good reason, of not acknowledging when users need a choice of behavior. Look at your own plus.google post about the negative feedback on Nautilus. Users want those options, you and I both know it. The response: 1) We don't acknowledge that people don't like the changes. 2) People who don't like the change are elitist. 3) Yes, it is worse but it works better for touch.

Think about leaving the laptop on when the lid closes. No one has argued that that should be the default. Many gave reasons why it should be a preference. Users were not given an option but you can easily read the snarky comments from developers: "you can run this line of code and keep both halves if your system breaks" on blog posts. And self congratulation for doing the tough work while users fume or leave. Same story with power off as the only option under the name. I read that bug report with my mouth hanging open. I honestly claim that 50% of users would want the other option and Gnome developers would not give their users the check box. Moreover, there is a good chance most would come across as arrogant asses while discussing it. You and Emily do a fantastic job, but I read what others write and I don't want to even give Gnome a fair shot; not that it much matters.

Kasbar back? I'm mad at KDE because it LACKS the features I found useful back in KDE 3. I don't mind if people also fault it for Anakondi and Nepomuk. Semantic desktop crap.

Also forgot the "wack-a-mole" effect. 46 pages of extensions, many that do the same thing. Have fun finding and picking the right extension. Again, clearly inferior to system settings and checkboxes.

Because:

• -They are Beta software
• -They are not typically upgraded when Gnome upgrades, if at all
• -They are an attack vector
• -They can and do conflict causing stability and speed problems
• -Only can be installed when online

But mostly because just about every extension is really something that should be a preference and is every way inferior to a checkbox.

Amen, wish I had mod points.

And stopped serving alcohol and limited the jukebox yo Justin Bieber.

Hey Ebassi! Now we have the GNOME folks on.