This is just a transcription of a Google+ post. Is that really what ZDNet has come to? Can't even email the guy to get a quote first and pretend to be doing journalism?
FWIW: I'm using Gnome 3 and don't hate it. It's thrown lots of stuff out the window, but I find many of the new idioms pretty nice -- the vertically stacked automatically expanding virtual desktops are amazingly great, for example. And the things that don't work (the app launcher basically sucks, "app" switching instead of "window" switching with Alt-Tab is a total fail when you have a dozen console windows open, nautilus doesn't manage icons on the desktop...) are easy enough to configure around or ignore. The biggest specific complaint seems to be the very poor multiple monitor support, but I use a laptop screen full time and am not affected.
Yeah, it's started growing on me once I learned about what's possible via extensions. At some point as long as they expose the APIs to build the right thing it stops mattering to a degree how the defaults work. Witness things like Conkeror (http://conkeror.org) being built initially as a Firefox extension due to the fact that all the extensibility to build a better UI was there from the start.
Your observation is insightful - the article is a blow by blow sportscaster-style rehashing of the "action" in a blog post. I think you may have discovered a new literary form.
Gnome 3 has been shipping in Fedora for over a year now, and they just pushed the third such revision. I'd say they're running out of their matters of time and entering the realm of "long standing breakage" at this point.
Here's my biggest personal peeve: keyboard navigation in the overview screen has never worked. The gnome-shell process in that mode steals events from gnome-settings-daemon (which normally handles hotkeys). This means that if you like to do things like launch apps from the keyboard (I mean: who wants to launch an app when in the overview/app-launcher screen?!) it just doesn't work. Combine that with the fact that the overview screen pops up automatically when you close the last window on a screen (I mean: who wants to launch an app just after closing another one!?) and it drives me up the wall daily.
Three times now I've straight up decided to fix this, but it's a non-trivial codebase and I always give up before making it work. The core folks really need to look at this problem, but at this point I doubt they ever will. Consistent keyboard navigation is clearly not a priority.
"However, it's only a matter of time until these minor issues are fixed"
I think that is not true, And I'll share my reasoning as to why I believe that and then someone can help tell me why I'm wrong.
My argument is that 'minor issues' are a function of design change + implementation maturity. Which is to say that as a design evolves, the evolution requires changes in the base code, these changes have ripple effects on other things that had made assumptions about the implementation and tracking down the ripples gets as far as the 'big ones' and then it ships with the proviso that the smaller problems will get ironed out 'over time'. Except once you release the newer version it sparks new design changes and those changes create ripples of their own and so even when the original minor irritations are fixed, you cannot tell they are fixed because there are new minor irritations to contend with. I claim that this is an unavoidable consequence of the 'bazaar' aspect of FOSS development. People write code, they want to contribute, they change stuff, there is few barriers to good ideas getting into the project, so there is no governor on change so there is no time for the changes to stabilize. In the 'cathedral' mode some company stops development, and works only on stability for some period of time, so that the final product is 'done.' This is not to say that cathedral software doesn't have 'minor annoyances' it can, but often that version will continue to be stabilized with no new features over time. And that is externally enforced by a management heirarchy.
I think it would be safe to say that minor issues that are left in Gnome 2 will be fixed over time, as there are very few people putting in new features to 2 and slightly more people putting in bug fixes. But that occurs because its a 'dead' project from the perspective of 'new' stuff.
Even if it does not fit the Gnome agenda, there is a way to make everyone happy: just add an annex section in Gnome documentation describing how to configure all the tweaks asked by Linus. This "Linus section" would be useful for many people to understand how to configure gnome to their wish.
I couldn't agree more with Mr. Torvalds. I'm relieved to know I'm not the only one - I was thinking maybe I was too picky, or too dumb to set Gnome 3 up the way I wanted it. I spent a half hour searching and Googling to find where they had hidden certain settings only to discover they had removed them, as if to say "This setting you depend on everyday, we don't think you need it - too bad!" I don't mind tinkering with things that may or may not break when I'm at home, but I've got to work at work, so I back-peddled to Gnome 2.32.
I totally agree. You have basic set of issues such as not being able to create icons on the desktop, cannot drag icons from start menu onto panel.
I cannot believe that gnome 2.6 was working fine, then I made the mistake of upgrading from debian stable to testing and got new GNOME. I have lots of SMB shares that are mounted. they don't show up. Neither can I drag a folder onto the favourites. I have to go into a folder then add bookmark.
Running applications became a multi-step process: Move mouse over to the panel on the left (that doesn't exist unless you focus mouse pointer over it), select applications button (forget what's it's called), then hover over to the right (full list of apps) then search there.
There is ALT+F2, but it's completely useless. Instead of assisting the user with trying to at least complete a command (or do some dingbat, but cute activity such as show how many times that command) it's just a plain text box, every basic window manager even tries to resolve things for you.
While I'm not big on 'minimalist' UI such as fluxbox, it's far more functional in basic use such as create a launcher and run a program than new GNOME.
How the heck does one add applets to the panel such as a sound mixer which disappeared for some reason, yet it can be found from applications.
"I’m really tired of the f*cking old “just use the keyboard shortcuts” crap."
Amen. This is the same bullshit you hear from Apple apologists whenever they're confronted with some glaring functionality omission in the Mac UI (or keyboard, as in the case of Apple's missing Delete keys).
I've never understood who exactly is the target audience of Gnome 3 (and, to a lesser extent, Unity). It seems like these things would be great on an iPad or something like that, but what Linux users are really running Fedora or Debian on a touch device?
Not only that, I think many of these designs don't make sense to novice users either. I recently let me wife try a few live discs to see which one she liked best (she is very non-technical but was sick of Windows running like maple syrup). She ended up liking Mint's LXDE remix the most because it was simple and very fast. Unity's unifed menus and Gnome's app-not-windows design just confused her. Granted, she's just one person, but if it's not for me (the nerd) and it's not for her (the non-techie), who is supposed to be using Gnome 3?
What confuses me is that Gnome is community driven, and somehow a consensus of intelligent developers decided that this is the direction they want to take the project, despite Gnome's poor track record on mobile platforms and that the design paradigms don't accurately reflect the hardware on which the software is run. You'd think someone might have raised a hand and said "um, this design makes more sense on cell phones, which people are not running our product on."
I doubt i'm the target audience, but strangely the only decision that's really stuck in my craw since gnome 2 is the lack of a delete option in the file chooser :)
> It seems like these things would be great on an iPad or something like that, but what Linux users are really running Fedora or Debian on a touch device?
What you just described, sir, is a classic chicken/egg problem. Why would they design Gnome or Unity for multitouch users when there are none? But why would anyone use Gnome or Ubuntu on a multitouch device when the previous UIs were so mis-suited for that environment?
This is exactly the reason they shouldn't even be trying to get into that market. Competing with Apple, Microsoft and Google, why?
There will always be workstations. It would have been a great opportunity for Gnome to make the best damn workstation UI that ever existed (and they where on the right track with Gnome 2) but instead they jump on the touch screen band wagon in the hope of being adopted by device manufactures while in the process of doing so they're alienating their existing user base.
Honestly, I hope that GNOME fork (based on 2) panned out. I didn't follow it closely, but after having Linux Mint for over 3 months now running "GNOME Classic", I don't think I can go back to GNOME 3 proper, every now and then I'd switch to Cinnamon, but the Compiz effects (or maybe just GNOME 3 in general) screw with my ATI video drivers.
My major concern with gnome/unity/xfce is that the default font sizes look too big and really ugly. I guess choosing the right font sizes is not so trivial given various resolutions, but on fonts, every bit of effort will be paid off. And this can certainly be achieved. I am happy with the default system fonts in Mac and Windows nearly all the time.
I still can't believe Linus doesn't use Arch, or LFS or some sort of custom setup. The great thing about Linux is that you don't have to put up with the crap some distro gives you. Just remove Gnome and try something else.
15 years ago I'd have agreed with you. However, there comes a time in your life when playing with your tools becomes less fun and you want to just get down and use them.
I've gotten to the boring stage of my life when I am happy with Ubuntu LTS, because it means I can opt-out of a week of less productivity every six months when a new version comes out and breaks my workflow by forcing me to work around some "clever new idea" that won't last another couple of years.
I'm sure Linus is even busier than I am, so I'm pleasantly surprised that he takes the time to experiment with and evaluate new distros and environments as much as he does. That's probably not what he wants to spend his time doing though, and not where he's most productive.
41 comments
[ 3.7 ms ] story [ 86.1 ms ] thread(edit: fixed link:) https://plus.google.com/102150693225130002912/posts/UkoAaLDp...
FWIW: I'm using Gnome 3 and don't hate it. It's thrown lots of stuff out the window, but I find many of the new idioms pretty nice -- the vertically stacked automatically expanding virtual desktops are amazingly great, for example. And the things that don't work (the app launcher basically sucks, "app" switching instead of "window" switching with Alt-Tab is a total fail when you have a dozen console windows open, nautilus doesn't manage icons on the desktop...) are easy enough to configure around or ignore. The biggest specific complaint seems to be the very poor multiple monitor support, but I use a laptop screen full time and am not affected.
However, it's only a matter of time until these minor issues are fixed, and I happen to think that the general design is excellent.
Here's my biggest personal peeve: keyboard navigation in the overview screen has never worked. The gnome-shell process in that mode steals events from gnome-settings-daemon (which normally handles hotkeys). This means that if you like to do things like launch apps from the keyboard (I mean: who wants to launch an app when in the overview/app-launcher screen?!) it just doesn't work. Combine that with the fact that the overview screen pops up automatically when you close the last window on a screen (I mean: who wants to launch an app just after closing another one!?) and it drives me up the wall daily.
Three times now I've straight up decided to fix this, but it's a non-trivial codebase and I always give up before making it work. The core folks really need to look at this problem, but at this point I doubt they ever will. Consistent keyboard navigation is clearly not a priority.
(edit: this is the issue in question: https://bugzilla.gnome.org/show_bug.cgi?id=643111)
I think that is not true, And I'll share my reasoning as to why I believe that and then someone can help tell me why I'm wrong.
My argument is that 'minor issues' are a function of design change + implementation maturity. Which is to say that as a design evolves, the evolution requires changes in the base code, these changes have ripple effects on other things that had made assumptions about the implementation and tracking down the ripples gets as far as the 'big ones' and then it ships with the proviso that the smaller problems will get ironed out 'over time'. Except once you release the newer version it sparks new design changes and those changes create ripples of their own and so even when the original minor irritations are fixed, you cannot tell they are fixed because there are new minor irritations to contend with. I claim that this is an unavoidable consequence of the 'bazaar' aspect of FOSS development. People write code, they want to contribute, they change stuff, there is few barriers to good ideas getting into the project, so there is no governor on change so there is no time for the changes to stabilize. In the 'cathedral' mode some company stops development, and works only on stability for some period of time, so that the final product is 'done.' This is not to say that cathedral software doesn't have 'minor annoyances' it can, but often that version will continue to be stabilized with no new features over time. And that is externally enforced by a management heirarchy.
I think it would be safe to say that minor issues that are left in Gnome 2 will be fixed over time, as there are very few people putting in new features to 2 and slightly more people putting in bug fixes. But that occurs because its a 'dead' project from the perspective of 'new' stuff.
"Give me an API flexible enough and a repl in which to hack it, and I shall move the world."
thanks
And I can virtually guarantee they won't... extensions.gnome.org is full of obsolete, incompatible extensions.
I cannot believe that gnome 2.6 was working fine, then I made the mistake of upgrading from debian stable to testing and got new GNOME. I have lots of SMB shares that are mounted. they don't show up. Neither can I drag a folder onto the favourites. I have to go into a folder then add bookmark.
Running applications became a multi-step process: Move mouse over to the panel on the left (that doesn't exist unless you focus mouse pointer over it), select applications button (forget what's it's called), then hover over to the right (full list of apps) then search there.
There is ALT+F2, but it's completely useless. Instead of assisting the user with trying to at least complete a command (or do some dingbat, but cute activity such as show how many times that command) it's just a plain text box, every basic window manager even tries to resolve things for you.
While I'm not big on 'minimalist' UI such as fluxbox, it's far more functional in basic use such as create a launcher and run a program than new GNOME.
How the heck does one add applets to the panel such as a sound mixer which disappeared for some reason, yet it can be found from applications.
Keyboard shortcuts?
But at least it's pretty.
"And for all the people wasting everybodys time with "Why don't you use Unity/KDE/xfce/xyz" - I've tried them. They are even worse"
https://plus.google.com/102150693225130002912/posts/UkoAaLDp...
So Gnome 3 is officially the best Linux desktop.
Amen. This is the same bullshit you hear from Apple apologists whenever they're confronted with some glaring functionality omission in the Mac UI (or keyboard, as in the case of Apple's missing Delete keys).
And cmd+arrow keys for home and end.
Not only that, I think many of these designs don't make sense to novice users either. I recently let me wife try a few live discs to see which one she liked best (she is very non-technical but was sick of Windows running like maple syrup). She ended up liking Mint's LXDE remix the most because it was simple and very fast. Unity's unifed menus and Gnome's app-not-windows design just confused her. Granted, she's just one person, but if it's not for me (the nerd) and it's not for her (the non-techie), who is supposed to be using Gnome 3?
What confuses me is that Gnome is community driven, and somehow a consensus of intelligent developers decided that this is the direction they want to take the project, despite Gnome's poor track record on mobile platforms and that the design paradigms don't accurately reflect the hardware on which the software is run. You'd think someone might have raised a hand and said "um, this design makes more sense on cell phones, which people are not running our product on."
What you just described, sir, is a classic chicken/egg problem. Why would they design Gnome or Unity for multitouch users when there are none? But why would anyone use Gnome or Ubuntu on a multitouch device when the previous UIs were so mis-suited for that environment?
There will always be workstations. It would have been a great opportunity for Gnome to make the best damn workstation UI that ever existed (and they where on the right track with Gnome 2) but instead they jump on the touch screen band wagon in the hope of being adopted by device manufactures while in the process of doing so they're alienating their existing user base.
In the end, everyone loses.
I've gotten to the boring stage of my life when I am happy with Ubuntu LTS, because it means I can opt-out of a week of less productivity every six months when a new version comes out and breaks my workflow by forcing me to work around some "clever new idea" that won't last another couple of years.
I'm sure Linus is even busier than I am, so I'm pleasantly surprised that he takes the time to experiment with and evaluate new distros and environments as much as he does. That's probably not what he wants to spend his time doing though, and not where he's most productive.