Weekly Edition
Return to the Announcements page
| |||||||||||||
Aaron Seigo on the Future of KDE (Datamation)
Bruce Byfield talks
with Aaron Seigo about the future of KDE. "According to Seigo,
the large-scale changes that began two years ago with the release of KDE
4.0 are mostly complete now. "We've reached the stage with the 4.4 release
that happened in January where we've got this nice feature set on the
desktop and we have applications available for it and some nice refinements
in the look and feel. That's where we are. But where are we going? That's
always the difficult question. Once you've arrived at a place,what are you
going to aim for?" Seigo's answer to his own question is that KDE is
currently moving in three directions: adding functionality to the desktop
in both small features and within specific applications, extending the
concept of the social desktop, and the introduction of KDE on to every
possible hardware platform. Each is a small story in itself."
(Log in to post comments)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 16, 2010 21:47 UTC (Fri) by mgb (guest, #3226) [Link]
As someone who has tried and failed KDE 4 a couple of times, my only question is - when can my users start relying on KDE 4 to do what they're now doing in KDE 3?
P.S. A big big thank you to Debian for supporting KDE 3 in Lenny.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 16, 2010 22:18 UTC (Fri) by dlang (✭ supporter ✭, #313) [Link]
that depends on what your users are doing with KDE3.
I have been happily using KDE4 since 4.1/4.2 timeframe without running into the 'crippling problems' that seem to be so common in those versions.
the one feature that I know hasn't made it into KD4 yet is the ability to have a script generate your wallpaper (no world clock wallpaper background).
Has KDE 4 caught up with KDE 3 yet? Posted Apr 16, 2010 22:36 UTC (Fri) by horen (subscriber, #2514) [Link]
For starters, how about fixing the addressbook in KDE4, so that it DOES use (and DOES NOT strip-out) the group information associated with an entry! Also, the addressbook UI in KDE4 looks like crap!!
Has KDE 4 caught up with KDE 3 yet? Posted Apr 16, 2010 22:37 UTC (Fri) by Sho (subscriber, #8956) [Link]
> the one feature that I know hasn't made it into KD4 yet is the ability to have a script generate your wallpaper (no world clock wallpaper background).
Plasma (KDE 4's desktop shell, i.e. the process responsible for rendering the desktop surface and the panels, and the library backing it is also used in a number of other places) is a highly modular animal, with most of its elements being plugins. A basic is unit is nicknamed "Plasmoid", the UI calls them widget. A Plasmoid is for example the task bar or the clock. A special kind of Plasmoid is the Containment, which is a Plasmoid that can house other Plasmoids, e.g. the desktop or the panels. Some are both: Folder Views, a file manager-like graphical view of a file system, can both live on the desktop and on panels, but also serve as desktop surface itself (while also housing additional Plasmoids on it).
Wallpaper providers are also plugins, to make it possible to share them between different desktop containment types rather than duplicating that code. Bundled wallpaper types are your basic image or a slideshow, but there are also more involved ones in the package, like a fancy Mandelbrot generator or a nice map renderer that uses KDE's Marble geodata display widget (interestingly, since data engines and visualizations for their data - the visualizations being the Plasmoids - are separate plugins in Plasma, the Marble wallpaper plugin can use the same geo location data engine plugin that Plasmoids like the Weather display use to show your current location on a map as your wallpaper).
Since Plasma loves scripting, and has bindings to write the various types of plugins in various scripting languages like Python and ECMAScript, it should also be possible to write your own wallpaper provider in a scripting language and throw it into the mix.
If I remember correctly, the KDE 3 feature you're referring to was a hard-coded facility that allowed one to specify a command that needed to return a PNG image and would be run at a specified interval. Writing a wallpaper plugin that does that, and embeds the same config UI into the wallpaper settings dialog (where you choose the desired wallpaper type and then configure the options of the underlying plugin) wouldn't be too hard.
(Yeah, I totally abused this opportunity for a mini-sermon on Plasma's architecture. :-)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 16, 2010 23:06 UTC (Fri) by nix (subscriber, #2304) [Link]
I tried switching to KDE4 a couple of weeks ago, and ran into a crippling hang bug in konqueror (e.g. KDE bug 233341, which I bear complete responsibility for not helping track down yet; that's a job for this weekend).
Also I ran into WM problems. Effectively KDE4 is unusable with fvwm2, because *all* the plasma windows and dialogs have the same window class; so either the plasma panels all get title bars and borders (ew) or the dialogs end up with no title bars or borders and are thus immovable. Great stuff, guys :/ apparently we need to enhance fvwm to match on some EWMH2 'role' attribute before it'll work properly.
(I tried to use kwin, but despite much improvement since the 4.1 days it is still incapable of things I can do with a couple of lines in fvwm2, which I absolutely rely on. I'll open a thread on the kde list to see if anyone else knows how to satisfy my rather bizarre requirements before I give up, though, because I really want to use kwin, mostly because its desktop effects are absolutely gorgeous. But I need functionality too...)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 7:32 UTC (Sat) by wstephenson (subscriber, #14795) [Link]
KWin not fulfilling your needs as well as fvwm2 is pretty ironic, as KWin trades on being the functional WM with added pretty effects, as opposed to Compiz, the pretty effects with added functionality.
Out of interest, can you point me to the mail to the kde list or describe the missing functions here?
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 12:17 UTC (Sat) by nix (subscriber, #2304) [Link]
The missing functions, in brief:
- Starting with the minor stuff, I'm used to being able to run programs on the root window, or at least display their output there. Can't do that in KDE4. Thanks to a comment here yesterday I now know I have to do some plasma hacking to fix this, which should be fun. :)
- In fvwm+KDE3, I have bound virtually everything to keys (so maximize vertically and horizontally are on different keys). I have several hundred keybindings at this point, but I don't have several hundred keys. So I use key sequences heavily. These appear to have disappeared from KDE4, with no replacement.
- For that matter, khotkeys in KDE4 apparently only allows access to wm functions available over dbus. Virtually nothing related to specific windows appears to be available over dbus. Maybe I just haven't found it, but I'd have expected binding 'maximize this window' to a key to be obvious!
- In fvwm, I use layering and matching to customized window classes to have a set of borderless, titlebarless procmeters displaying a dozen critical stats on different machines down the right-hand-side of the screen. The rightmost of all fills most of the RHS, is at the same layer as everything else, and is the procmeter for the local machine: as it is on the same layer, maximizing windows butts up against this procmeter and does not occlude it. The others are at a higher/lower layer (switchable with a hotkey, hyper-=/hyper-+, where hyper is PrtSc and flips on the mod5 bit). The effect is that with one keystroke I can cause all but the local-system procmeters (and the xdaliclock and pager) to vanish beneath the active window, or reappear again.
(Screenshot, with procmeters in raised position: <http://www.esperi.org.uk/~nix/temporary/procmeter-configu...>)
I've been using procmeters arrayed like this for nearly fifteen years now: if they're not running, I feel almost like the machines have suddenly gone opaque on me. With them there, I can feel the pulse of the machine: because procmeter doesn't normally fork, I can even spot it and guess what's wrong when fork() has started to fail!
In fvwm2 this is about ten lines, no matter how many procmeters happen to be present, plus a -name parameter to give the rightmost procmeter a class other than 'ProcMeter3':
# Flip procmeters to the background
# Flip them back into the foreground again
Key KP_Subtract A 5 BegoneProcmeter
Key KP_Begin A S5 Maximize 100 100
In KWin, it appears to be impossible to do *any* of this. There's no layer management, no binding of raise/lower to keys, no *nothing*. Plasmoids plainly can't do it because they're nailed to the desktop background, while these procmeters spend most of their time raised *above* the active window.
What am I missing? Is there some secret scripting engine I'd missed? 'cos this is really not a lot to demand of a window manager (fvwm 1 could do it way back in 1995, although there were no layers back then so I had to use raise/lower), and if KWin can't even do this I suspect it's not for me.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 23:04 UTC (Sat) by nybble41 (subscriber, #55106) [Link]
I'm not sure how one would go about maximizing a window under KWin via DBUS, but you can do it easily under a variety of EWMH-compliant window-managers (including KWin and FVWM) with wmctrl <http://tripie.sweb.cz/utils/wmctrl/>:
wmctrl -i -r 0x04000016 -b add,maximized_vert,maximized_horz
In addition to the window ID (-i), as shown above, you can also select the target window by WM_CLASS (-x) or title (default). You can also control other flags, such as 'sticky', 'above', 'shaded' and 'hidden', move and resize windows, switch virtual desktops, and change the focus.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 19, 2010 20:31 UTC (Mon) by rfunk (subscriber, #4054) [Link]
Regarding your ProcMeter issue, I use the System Load Viewer plasmoid to keep tabs on the pulse of the machine. It's not as detailed as your Procmeter setup, but I'm pretty sure there are other plasmoids that give more detail.
In KDE4 you just have to start thinking in high-level terms of plasmoids and widgets, rather than in low-level terms of borderless windows.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 19, 2010 21:38 UTC (Mon) by nix (subscriber, #2304) [Link]
Well, yeah, but plasmoids can't give you info on remote machines (you can't ssh to another machine and run a plasmoid there), and as I mentioned, plasmoids are nailed to the desktop background, which is completely useless if you want them to float above your windows most of the time. :/
Has KDE 4 caught up with KDE 3 yet? Posted Apr 19, 2010 21:52 UTC (Mon) by rfunk (subscriber, #4054) [Link]
A plasmoid can do anything any other program can do. Your plasmoid can be written to ask another machine for its stats.
And in fact plasmoids now seem to get cross network capability ("share this widget on the network") automatically. I haven't tried it and don't know the details of it though.
And they're only nailed to the desktop background if that's where you put them. They can also be on a transparent panel if you like.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 19, 2010 22:10 UTC (Mon) by nix (subscriber, #2304) [Link]
Aha! So all I need is a way to have a triple-width vertical panel with three plasmoids on it, and, er, it still doesn't work because they can't adjust their depth independently because they're nested on a single entity at the side of the screen. (And it's not as if I can use one panel on each side of the screen, because the whole point of this is that stuff overlapping your window on the right hand side is often not obscuring anything important. On the left, it's quite different. Also, they should be next to each other for visual-comparison's sake.)
Gah.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 16:50 UTC (Fri) by jospoortvliet (subscriber, #33164) [Link]
While I'm unsure plasmoids can do all you want, they CAN run in a separate window if you like. Furthermore, KWin has a pretty powerful custom configuration. It does not work on non-window items like plasmoids on the desktop but it DOES work on plasmoids in a window - though I'm unsure if they are identified properly for you to do what you want.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 22:54 UTC (Fri) by nix (subscriber, #2304) [Link]
It looks like plasmoids in a window and the ability to position them precisely and hotkey-control their raising/lowering would do the trick. Unfortunately the latter still seems undoable, and if all the plasmoids have the same window class and name (as it appears they do), I don't see how KWin can distinguish between them to nail them in the right places. (But maybe it doesn't need to. I'll give it a try this weekend.)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 24, 2010 8:48 UTC (Sat) by jospoortvliet (subscriber, #33164) [Link]
You might be completely right. However, fixing the window class and name issue might be pretty doable, and adding the needed hotkey capabilities to kwin might also be possible.
Another thing, as I'm behind a plasma desktop now I can check out a few things, so let me do that.
I don't know what you need exactly, but under the global keyboard shortcuts (kwin) I see a lot of shortcuts. you can also give individual windows a shortcut and configure their options under 'right mouseclick on decoration > advanced'. Combined, does that bring you what you need?
Has KDE 4 caught up with KDE 3 yet? Posted Apr 24, 2010 9:42 UTC (Sat) by nix (subscriber, #2304) [Link]
Oh, I'm sure it's possible. It's just not *there*. :/
Ah well, the kde devs are not telepaths and I'd never mentioned any of this before. I'll repost it on the kde@ list shortly.
Combined and window shortcuts would let me switch focus to the window on a keystroke. These windows contain nothing but procmeters that don't respond to keystrokes, so giving them the focus is a waste of time; and it raises the window but does not lower it. also, the 'maximize does not cover the outermost procmeter' still isn't satisfied :/
gah. this took *five minutes* with fvwm2.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 24, 2010 10:06 UTC (Sat) by jospoortvliet (subscriber, #33164) [Link]
Well, I'm assuming you've seen all the capabilities of Kwin by now (I did find shortcuts to lower an active window but that's not what you need either is it?) so yes, I guess this simply isn't there. I wish you luck finding a dev interested in writing these features (or figuring out how to do it yourself).
Cheers and enjoy the weather :D
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 16:20 UTC (Fri) by jschrod (subscriber, #1646) [Link]
AFAIU, all plasmoids run in the same process. The "low-level term" of borderless windows has the BIG advantage that one of them crashing won't bring down my whole desktop. Processes were invented for a reason...
While Web browser rightily start to use more processes to shield windows or tabs from each other, KDE starts to integrate everything and its darling into one big giantic mess, able to overwrite memory of each other's plasmoids at will.
And this is called a Good Design(tm) nowadays. I'm getting old.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 16:52 UTC (Fri) by jospoortvliet (subscriber, #33164) [Link]
on the other hand, having 20+ processes (a number of plasmoids which is conservative on many setups) will eat memory like crazy. I'd rather have them in one multi-threaded process. Plasmoids should become more and more scripted (ECMA, python, ruby, etc) in the future and only the core plasmoids will be C++. And should obviously be as much stabilized and tested as possible...
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 17:37 UTC (Fri) by dlang (✭ supporter ✭, #313) [Link]
if they are well written to begin with they should not eat significantly more memory when run as separate processes then when running on one giant process.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 24, 2010 8:44 UTC (Sat) by jospoortvliet (subscriber, #33164) [Link]
As I'm not a developer, I can't really argue this. The only thing I know is that the plasma devs told me that running each plasma widget (panels, background and every widget on them) in a separate process would eat memory like crazy and introduce a lot of overhead in synchronizing things. The latter has to do with the synchronizing of animations they do and saving power as well - events are gobbled up and fired at certain intervals to wake up the proc as little as possible, saving power on battery. This is hard if not impossible when running in separate processes, where all the separate plasmoids would wake up the proc at different points in time preventing it from sleeping or clocking down.
I greatly respect the core plasma hackers and I'm sure the situation might be more complex than you think...
Has KDE 4 caught up with KDE 3 yet? Posted May 3, 2010 19:21 UTC (Mon) by dlang (✭ supporter ✭, #313) [Link]
with the linux COW feature, they won't use much more memory.
synchronization would be an issue, but how close does the animation of different things need to be synced? and why? In anycase, this is what Futex are good for.
wakeups can be batched by the OS (there have been some articles on this recently)
I suspect that the core plasma hackers are focused so tightly on their use-case and world that they are missing the other advantages.
putting everything in one process is great for some things, but it makes things very fragile as there are a LOT of ways for things to go wrong.
Has KDE 4 caught up with KDE 3 yet? Posted May 3, 2010 20:22 UTC (Mon) by jospoortvliet (subscriber, #33164) [Link]
Hmmm... I'm unsure if the reasons behind the one-process thing have been documented somewhere, couldn't find something quickly. If you want you could ask on #plasma (freenode) and see if anyone knows - maybe there are indeed possibilities and if you can argue a good case I'm sure the plasma devs would listen... I suspect it wouldn't even be that hard to write a proof-of-concept, and for that I'm SURE the devs would support you...
Again, I can't really argue about this, but I certainly would appreciate a more stable Plasma so if you could find it in your hart to put in the effort to make it possible, my blessing is yours ;-)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 22:52 UTC (Fri) by nix (subscriber, #2304) [Link]
And in KDE4 *Konsole* does that as well. Unless you specifically demand otherwise *all your Konsoles run in the same process*, to facilitate dragging tabs between konsoles.
Whoever thought this was a better idea than having each tab a separate process and the containing window another one was an absolute maniac. It is evident to everyone (other than, apparently, the Konsole authors) that you do not want *every shell you own* to implode *simultaneously* if Konsole has a bug...
Has KDE 4 caught up with KDE 3 yet? Posted Apr 25, 2010 14:22 UTC (Sun) by boudewijn (subscriber, #14185) [Link]
How is calling the konsole authors "absolute maniacs" "polite, respectful and informative"?
Has KDE 4 caught up with KDE 3 yet? Posted Apr 25, 2010 15:32 UTC (Sun) by nix (subscriber, #2304) [Link]
Perhaps it was somewhat hyperbolic :) still, it's a pretty accurate reflection of what everyone I have discussed this with has said; generally it starts with 'they did *what*? Why would anyone ever consider that a good idea?' and segues into a desire to not have anything to do with KDE ever again: I am generally able to deflect this by mentioning the existence of --no-fork and konsole replacements... Why is --no-fork not the default, anyway?
Has KDE 4 caught up with KDE 3 yet? Posted Apr 28, 2010 10:50 UTC (Wed) by daenzer (✭ supporter ✭, #7050) [Link]
> I'm used to being able to run programs on the root window, or at least
> display their output there. Can't do that in KDE4. Thanks to a comment > here yesterday I now know I have to do some plasma hacking to fix this, > which should be fun. :)
FWIW, you can't do that with compositing anyway, as the root window contents are clobbered by the compositing manager.
P.S. I couldn't live without my gkrellm any more than you could without your procmeters. :)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 28, 2010 12:00 UTC (Wed) by nix (subscriber, #2304) [Link]
I tried gkrellm years ago but it just looked... wrong, and at the time running lots of them on remote machines was tricky for reasons I forget. I should try again someday; it's much more flexible than procmeter.
(Now back to trying to figure out why X crashes and wrecks my console on exit in 2.6.33/4 but not in 2.6.32 :/ gah. Bugs, bugs, all around me bugs...)
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 0:21 UTC (Sat) by sfeam (subscriber, #2841) [Link] If I correctly understand the bug-tracking entries here and here , and here, printing from KDE4 is not likely to work for quite a while yet. The basic flaw has apparently been fixed upstream in Qt itself, but won't likely make into KDE before Qt4.7. Until then, basic stuff like non-A4 paper and duplex printing will remain broken.IMHO a better fix would be to bring back the KDE3 universal print utility kprinter. That way a work-around or bug-fix to this one application would benefit all desktop printing. The KDE4 approach of having individual apps call directly into the print libraries has turned out to be a bad idea. Even aside from outright buggy printing, the current crop of KDE4 apps fail to provide a sufficiently general UI for basic printing. E.g. okular 0.9.5 (KDE 4.3.5) offers no option to print an arbitrary set of non-contiguous pages. I have found no fully general alternative, although in extremis I use acroread to print from pdf docs. So anyhow, no, KDE4 has not yet caught up with KDE3. And I say that despite using it every day, continually cursing the print problems.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 1:06 UTC (Sat) by dlang (✭ supporter ✭, #313) [Link]
interesting, I am much more frustrated the printing on gnome much more than the printing on KDE, but it depends on what your pain points are.
in my case I have the printing dialog stall while it queries the printer to find out if it's online or not, this can take 10s of seconds in some cases, meanwhile the 'print' button is grayed out. I just want to queue the print job up and let it print in the background!
Has KDE 4 caught up with KDE 3 yet? Posted Apr 21, 2010 1:50 UTC (Wed) by cleary (guest, #41669) [Link]
OT, but you can comment this line in /etc/cups/cupsd.conf to resolve this:
Listen /var/run/cups/cups.sock
On topic, the missing print manager is my one bugbear, and it's not critical to me. Personally I've found the kde 4 series from 4.2 on a pleasure to use.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 1:12 UTC (Sat) by dlang (✭ supporter ✭, #313) [Link]
reading through those links I am very puzzled. I have not had the problems that they describe (or if I do have them I haven't noticed them), one thing that I do differently is that I have separate print queues setup for duplex vs simplex printing. I want that so that I don't have to load printer drivers on every desktop, I just load postscript printer drivers and have them print to the appropriate queue.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 12:10 UTC (Sat) by cowsandmilk (guest, #55475) [Link]
I don't know what this is about. I regularly print duplex to letter paper from okular with no difficulties. Maybe OpenSUSE has patched this with a fix in their KDE, but I'm quite sure it would then have quickly gone to upstream.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 23:30 UTC (Sat) by sfeam (subscriber, #2841) [Link] The issue reported in the bug tracking, consistent with what I see here, is that a default of A4 is hard-coded. Yes, you can print to US Letter and set simplex or duplex as you like. But you have to choose these options afresh every time; you can't get them to stick as a default setting.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 18, 2010 0:27 UTC (Sun) by k8to (subscriber, #15413) [Link]
Reverse cultural imperialism! Europeans have been dealing with this the other way for so long. Ironic.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 18, 2010 7:15 UTC (Sun) by niner (subscriber, #26151) [Link]
Yeah! I've lost count of how many times I stood at a printer deleting a job because Firefox or anything else try to force US letter on me every time...
Has KDE 4 caught up with KDE 3 yet? Posted Apr 17, 2010 14:07 UTC (Sat) by stumbles (guest, #8796) [Link] That's odd, no problems with KDE+printing here. I guess its a distro thing.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 19, 2010 10:22 UTC (Mon) by buchanmilne (guest, #42315) [Link] The basic flaw has apparently been fixed upstream in Qt itself, but won't likely make into KDE before Qt4.7 So, this is a Qt4 bug, and if KDE3 had simple been a port to Qt4, then this bug would still have been present? If kprinter had been retained, this bug would also have been exposed in kprinter? According to Bug 6471, this is fixed in the Qt 4.6.3 branch. As far as I can tell, there is no reason why any distro shipping KDE 4.3 or 4.4 should not be able to provide this patch in a bugfix patch for Qt 4.6. But, again, this doesn't look like a bug in KDE4 itself.
Has KDE 4 caught up with KDE 3 yet? Posted Apr 19, 2010 16:46 UTC (Mon) by sfeam (subscriber, #2841) [Link] But, again, this doesn't look like a bug in KDE4 itself.Sure. I said it was Qt bug. That doesn't make it any less problematic for someone switching to KDE4. And the lack of an adequate printer interface in okular combined with the loss of kprinter are design flaws in KDE4 itself. Anyone know if there is an improved print menu for okular in the works?
Has KDE 4 caught up with KDE 3 yet? Posted Apr 23, 2010 9:27 UTC (Fri) by jnsc (subscriber, #35668) [Link]
I love kde, but apart for printing, proxy support is also broken https://bugs.kde.org/show_bug.cgi?id=155707 https://bugs.kde.org/show_bug.cgi?id=189019, SSL support is incomplete https://bugs.kde.org/show_bug.cgi?id=162485, and some have been reported in 4.0. So I would say no, KDE 4 has not caught up with KDE 3, but it also has some nice new features. For personal use it's ok, but I would not deploy it in a company.
Aaron Seigo on the Future of KDE (Datamation) Posted Apr 17, 2010 4:13 UTC (Sat) by leiz (subscriber, #46265) [Link]
There's still lots of regressions in KDE4. I'm sad we're on KDE 4.4 and many are still not fixed. For instance, one of the most important apps for many Linux users is Konsole. I'm annoyed that Konsole lost its setting for the terminal size from KDE3 -> KDE4. Telling KWin Konsole should be WxH pixels is not intuitive at all.
Aaron Seigo on the Future of KDE (Datamation) Posted Apr 17, 2010 7:16 UTC (Sat) by bcebul (guest, #41527) [Link]
Aaron Seigo is sadly deluded if he thinks that after KDE 4.4 he has "arrived". He did not mention fixing the regressions and numerous bugs which still exist in the latest 4.4.2 offering. The bleeding users have been screaming about these for years. The best public relations, Aaron, is to make the program work. Less talk, more coding. I, the bleeding user, have dutifully put in my bug reports.
Aaron Seigo on the Future of KDE (Datamation) Posted Apr 17, 2010 20:30 UTC (Sat) by kragil (subscriber, #34373) [Link]
The beautiful face of entitlement. Love it.
Entitlement? Posted Apr 17, 2010 23:50 UTC (Sat) by pboddie (subscriber, #50784) [Link]
Keep wheeling out the "entitlement" argument. This isn't specifically a KDE thing, but I had the recent misfortune to try and get Kubuntu Hardy (the not really "long term" support version of Ubuntu plus KDE) to work properly with a home WLAN, and spent quite a few hours tracking down issues with the general WLAN infrastructure as well as KNetworkManager - stuff which others had reported and which were marked as "invalid" seemingly either because no-one really knows what features such software provides, or because they don't care about that particular version of KDE any more.
What is depressing is that much of the same software will be floating around in the "supported" versions of KDE/Kubuntu and will have the same deficiencies - this is the CADT thing JWZ referred to once upon a time - and even if the infrastructure isn't the same, it's quite possible to imagine that the same deficiencies are inherent in the "usability" analysis. For example, it appears to be much more exciting to have the Avahi stuff getting in the way than it does to have a reliable, basic network configuration that just works in the common case.
I spent quite some time peeling back the layers of shiny configuration tools and "helpful" components like Avahi in order to have a properly working network connection from boot time, and even then various unresolved issues remained with reconnection/reassociation. If the developers think people like me, someone who actually seems to end up doing the quality assurance that someone else is quite probably paid to do in various cases, sinking numerous hours into working through problems and reporting the detail to get them fixed, only to see these problems marked as "invalid" or being brushed off, is complaining out of a sense of "entitlement", then these developers deserve all the criticism and complaints that come their way.
Network manager... Posted Apr 18, 2010 9:30 UTC (Sun) by aleXXX (subscriber, #2742) [Link]
I think what you describe is less a problem of KDE, but instead of problem of networkmanager:
See the reply from bille here: http://www.kdedevelopers.org/node/4129
For the users this ends up as a KDE problem, because KDE is the software which exposes this functionality on the desktop.
Alex
Network manager... Posted Apr 18, 2010 17:24 UTC (Sun) by pboddie (subscriber, #50784) [Link]
Actually, the problem was that I just wanted the home WLAN connection to come up automatically without anyone even needing to go and choose it from the list after logging in. This is, after all, a fairly desirable common case: you get a broadband connection, go to buy a router, and for many people the choice will fall on a wireless router because they aren't really any more expensive, and people probably don't want to run cables around their home. This connection should behave just like the wired LAN connection you'd have had instead a few years ago.
Now, had KNetworkManager been able to choose a specific network by default, I'd probably not thought any more of this, but after navigating the bizarre configuration settings dialogues for a bit, looking into KWallet and its odd interactions with KNetworkManager - you have to create a new wallet when deciding *not* to use KWallet to store credentials - I'd probably need to start hacking the code to get it to do something that should have been an obvious feature.
The alternative was to peel away those nasty layers of "helpful" functionality. And Avahi turns out to be one of those technologies made to deal with the baroque use-cases that emerge from usability brainstorming sessions: great if you're hopping from one Internet café to another, then on to an airport lounge before ending up at some Google office or other; not so great if you just want to take the DHCP address from your router without some stupid daemon popping up and deciding that you're on some special subnet that can't talk to anything.
In the end, after Avahi had been evicted from the system, the matter of configuring the network was three or four lines in the interfaces file. Quite usable, in contrast to the mountain of technologies and user interfaces that failed to provide the same solution.
Network manager... Posted Apr 18, 2010 20:03 UTC (Sun) by boudewijn (subscriber, #14185) [Link]
You know? Knetworkmanager, for all its faults, and most of those are by reason of the networkmanager backend, actually does select my home network automatically, or the hotel network if I'm travelling, or the t-mobile or kpn hotspot.
Yes, there are warts, not being able to connect to essid hidden wifi networks being one, but not automatically connecting to the home network is not one of them.
And your issues with kwallet are extremely bizarre. It doesn't do that on any of the laptops I've installed kde4 on. Sorry.
Network manager... Posted Apr 19, 2010 12:11 UTC (Mon) by pboddie (subscriber, #50784) [Link]
Well, if KNetworkManager and KWallet work for you in non-bizarre ways in KDE 4, then I'm pleased for you. All I'm doing is reporting what they do in the last version of KDE 3.x that Kubuntu shipped before pulling the rug on the support.
Of course, everyone wants KDE 3 users to upgrade to KDE 4 to experience the same software reimplemented or reworked and hopefully better than before. The march of progress is great; making users switch environments to hopefully get incremental improvements is not so great. But that's the story of KDE for the past few years.
Network manager... Posted Apr 19, 2010 23:19 UTC (Mon) by anselm (subscriber, #2796) [Link] It's not just kprinter. For example, in KDE 4 Konqueror apparently lost pretty much all of its SSL configuration capabilities, including the ability to deal with client certificates, which was really quite well presented in KDE 3.5. The KDE fanboys will now argue that (a) HTTPS is mostly a farce, anyway, and it makes more sense for the developers to spend their time on things that actually make a difference (like snazzy desktop effects in the window manager), and (b) if one desperately needs to use SSL client certificates then one can always switch to Firefox, which handles them very reasonably. This is KDE 4 for you. The developers always know best.
Network manager... Posted Apr 20, 2010 6:34 UTC (Tue) by kragil (subscriber, #34373) [Link]
It sure often seems like "developers know best", but I think Konqueror is a bad example. Most KDE devs don't really use it and there have been numerous blog posts and talks that it is holding KDE back (and it is IMHO).
Webkit seems to be the solution most people want. Project Silk comes to mind.
Konqueror is a zombie. It is alive and is slowly being developed although in reality it is dead. Just try any HTML5 site and compare with Webkit/Gecko to see how much of a future it has. None. I am sure it will be dropped from KDE5.
Network manager... Posted Apr 20, 2010 6:36 UTC (Tue) by kragil (subscriber, #34373) [Link]
Clarification: I am mostly talking about KHTML.
It is all about KHTML ... Posted Apr 20, 2010 9:15 UTC (Tue) by anselm (subscriber, #2796) [Link] ... which of course has no bearing whatsoever on whether and how Konqueror handles SSL. (This is what in the business we call a non sequitur.) HTTPS is about getting hold of resources. KHTML or WebKit is about what to do with the resources once you have got hold of them. In any sane browser design the two are completely independent. The SSL configuration bits of Konqueror, like kprinter, demonstrate how functionality that seemed to work perfectly fine and usefully in KDE 3.5 was thrown out of KDE 4 for no apparent reason whatsoever. This is not how one builds and maintains a loyal user base – I support people who happily use KDE 3.5 and who I can't upgrade to KDE 4.x simply because (among other reasons) they depend on features of kprinter that have no equivalent in KDE 4.x. This sucks. At the moment KDE 4.x is not a good desktop environment; it's just that all the obvious alternatives are even worse.
It is all about KHTML ... Posted Apr 21, 2010 12:34 UTC (Wed) by AndreE (subscriber, #60148) [Link]
I think the point is that there isn't a huge interest in maintaining konqueror at all, regardless of the bits (SSL, rendering engine)
It is all about KHTML ... Posted Apr 23, 2010 9:20 UTC (Fri) by stevan (subscriber, #4342) [Link]
>> I think the point is that there isn't a huge interest in maintaining
>>konqueror at all
Oh, I do hope that's not the case. Konqueror is one of the reasons along with digikam that I've gone back to kde.
Now if you're talking about the monstrosity that amarok has become and compare it with, say, exaile...
S
It is all about KHTML ... Posted Apr 30, 2010 13:54 UTC (Fri) by forthy (guest, #1525) [Link] SSL is not only for Konqueror - the same kio module is also used for KMail to access your IMAP account. And yes, you can do client-certificate based authentication on SMTPS and IMAPS, and it doesn't work with KDE4. WebKit doesn't help here at all, if the infrastructure below doesn't work. Any KDE-based WebKit browser will use the same broken kio module, and therefore not be able to do SSL properly. Of course, KHTML sucks compared to WebKit, so this is part of what needs to be done for KDE.
It is all about KHTML ... Posted Apr 23, 2010 17:13 UTC (Fri) by jospoortvliet (subscriber, #33164) [Link]
I understand your frustration. I do want to point out it wasn't "trown out" but nobody had time or interest to port/rewrite it. Unfortunate, but not the same. And delaying the release until everything was ported was no option - as for many things there simply was no developer interested in writing it. Yes, that sucks, no, little can be done about it unless you want to put a gun to the volunteers' head and force them to write it.
It is all about KHTML ... Posted Apr 23, 2010 17:41 UTC (Fri) by anselm (subscriber, #2796) [Link] With a project of the size and scope of KDE, it is natural that there are some tasks that seem dull and boring, like writing documentation and fixing old bugs, while others look sexy and get you developer cred. It's no wonder that there is a profusion of sexy-looking shallow stuff (much of it never quite finished) but the dull and boring tasks get neglected. Too bad if other people would really like those issues cleared up; let them learn C++ and Qt and do their own coding; KDE can always use more developers! At the end of the day it's a question of philosophy. If KDE wants to be a fun hacking project where everybody gets to do whatever they fancy whenever they like then fine, that's their privilege. However if KDE wants to provide a reliable desktop environment (where »reliable« means, among other things, being free of serious regressions that never seem to be addressed by anybody) then at some point someone will need to clench their teeth and do the unpopular stuff even if it isn't fun. The problem with KDE these days seems to be that they want to have it both ways. People who point out long-standing regressions like the SSL or kprinter issues get told that »nobody is interested in doing this« but at the same time the KDE folks wonder why most notable distributions nowadays seem to default to GNOME when KDE is so much better – if you keep your pink spectacles on, anyway.
Network manager... Posted Apr 19, 2010 9:55 UTC (Mon) by buchanmilne (guest, #42315) [Link]
This is one of the reasons Mandriva doesn't use NetworkManager by default.
As such, this is hardly a KDE problem.
You seem to have been upset by *one* Avahi feature, avahi-autoipd, which is there for the "Auto-configured IP" feature that is present on Windows (since Windows 95) and Mac OS X. Such features can be annoying when troubleshooting a DHCP problem, but once again, this is nothing to do with KDE. On Mandriva, such features aren't enabled by default either (but can be enabled quite easily). The rest of avahi does provide some nice features (such as server-less LAN-local IM with Avahi support in Kopete and Pidgin).
AFAICT, all of the problems you discussed here (related to networking issues) are distro-related, and would have affected you on any of the desktop environments provided by the distro.
Network manager... Posted Apr 19, 2010 12:23 UTC (Mon) by pboddie (subscriber, #50784) [Link]
It isn't so much a KDE problem - apart from KNetworkManager not being able to sanely choose a network and connect to it by default - but with regard to the underlying issues, I did write that it "isn't specifically a KDE thing".
As for whether it was one tiny (you could say "wafer-thin") feature that caused a great deal of disruption, it doesn't matter: if avahi-autoipd keeps setting an address, it undermines the credibility of the technology, especially in some situations where the technology may not be particularly useful anyway.
My response really addressed the issue of "entitlement": a convenient term used by developers to brush aside criticisms by implying that the lazy users do nothing but download the software, play with it, then whine, that they put in no effort to deliver anything or to improve the software. That testing or documentation is apparently of so little value to such developers comes hardly as a surprise given the experience frequently offered to the end-user.
Network manager... Posted Apr 29, 2010 16:08 UTC (Thu) by Epicanis (guest, #62805) [Link]
My impression of "NetworkManager" is that it's sort of the "PulseAudio" of network configuration tools - for the people for whom it doesn't cause problems it's great, while for everybody else is a Tool Of The Devil.
(I personally have a burning, admittedly even perhaps downright irrational hatred of it...)
I find WICD works just fine for me since at least about KDE 4.2.0 (perhaps earlier), so I haven't re-tried NetworkManager recently.
(As for PulseAudio, I have thus far completely avoided it by just using ALSA's dmix plugin, so I don't know if I would hate it if I tried to use it or not. Seems like it'd be "artsd" all over again...)
Aaron Seigo on the Future of KDE (Datamation) Posted Apr 23, 2010 17:09 UTC (Fri) by jospoortvliet (subscriber, #33164) [Link]
There will always be SOME pet peeve which no developer has been interested in yet. Aaron meant we're 99.99% there. Most ppl could actually work with 4.1 already, and with 4.4 there are just a few area's which still hit ppl because of their rough edges. Yes, it sucks if YOU are bitten by one of those, but that doesn't mean it doesn't work for pretty much everyone else...
Please make KDE lighter Posted Apr 17, 2010 13:17 UTC (Sat) by ejmarkow (guest, #56170) [Link]
KDE 4.4.2 is working fine for me. Please focus on making it a bit lighter? Thanks.
Please make KDE SC lighter Posted Apr 17, 2010 20:44 UTC (Sat) by kragil (subscriber, #34373) [Link]
Yeah, I agree. Being fast and responsive is getting more and more important and I don't think KDE SC is really there yet. I would love to see KDE SC on the Ipad for a comparison.
Sure, a lot of plumbing can help with faster application starts etc and offloading more work to the GPU will also help (GPU font rendering etc)
But in the end smoothness is very subjective. iPhoneOS caches application screenshots to be able animate something while the app still loads. I want something similar. Cache more stuff and preload libs I will likely load etc. KDE SC can be many times faster no doubt.
I really see no reason why my 3GB RAM desktop machine is idle the first minutes after startup. It should preload all the applications I usually start until the RAM is full so when I really start them they are just there.
Stuff like that makes a great desktop.
Would caching really help? Posted Apr 17, 2010 21:53 UTC (Sat) by tpo (subscriber, #25713) [Link]
If every bit of software would go down the caching path, then I recon that would mean heavy swap trashing. Like if you'd run some programm for a while and then CTRL-Tab, then you'd grow a beard before the other app would resurrect from swap. And, actually, the swap trashing is allready a reality on my 1GB RAM + 1GB swap Ubuntu/Gnome laptop when I dare to open some additional app on top of firefox. It's not a satisfying user experience.
The basic problem is I'd say, that even if there was a "cachealloc" call, that would give you memory that the OS would not bother to retrieve from slow media, then coding against such memory would be practically impossible from C since every access to that memory could return with a SIGSCRATCHEDBYNOW or similar....
Would caching really help? Posted Apr 17, 2010 22:12 UTC (Sat) by kragil (subscriber, #34373) [Link]
I was talking about the DE not every application. And I meant only real RAM. Of course it shouldn't touch swap and if I decide to run something else or new then the cached stuff which is most unlikely to be needed (e.g. apps that I only run from time to time) should be dropped if no real RAM is left (highly unlikely on my machine)
But honestly, what good is a DE if it doesn't even know which programs I like to run. KDE SC with stringi and nepomuk should be able to know and act accordingly.
Would caching really help? Posted Apr 17, 2010 23:44 UTC (Sat) by alankila (subscriber, #47141) [Link]
Sounds like what Windows 7 allegedly does. I think it's probably a great idea, because there's so much unused RAM right after bootup.
Would caching really help? Posted Apr 18, 2010 10:26 UTC (Sun) by nix (subscriber, #2304) [Link]
It's an awful idea. It means that the more RAM your machine has, the longer it's almost unusable right after bootup because it's too busy pounding the disk like crazy. (Of course, part of this is that Windows's I/O scheduling still sucks and its FS is slower than molasses, but still.)
Would caching really help? Posted Apr 18, 2010 13:14 UTC (Sun) by alankila (subscriber, #47141) [Link]
There's absolutely no reason why such preloading tasks couldn't have a low-priority and give up on the slightest hint of "real" system activity, and only resume when system is quiescent again.
Would caching really help? Posted Apr 18, 2010 18:37 UTC (Sun) by kragil (subscriber, #34373) [Link]
That would be a scheduling bug if the Linux desktop would behave that way.
Would caching really help? Posted Apr 21, 2010 14:46 UTC (Wed) by Wol (guest, #4433) [Link]
If you're getting swap thrashing, try upping swap from 1 to 2 Gb.
I know VM has been re-written "recently", but it isn't that long ago (ie postdates my current PC! that they *brought* *in* the rule "either no swap at all or at least twice ram". Anything in-between had efficiency problems, that were down to *design*, not *implementation*.
So, unless they've thrown out the *design* of the swap system (which is quite possible :-) since the 2.4 days, you might find increasing your swap to 2Gb will gain you a bit.
(For those who say "twice swap" belongs to the days of the dodo, don't forget early vanilla 2.4 kernels WOULD CRASH if you broke the "none or twice swap" rule ... and iirc that was a noughties kernel?)
Cheers,
Would caching really help? Posted Apr 21, 2010 17:11 UTC (Wed) by dlang (✭ supporter ✭, #313) [Link]
the design of the swap system (as part of vm) has very definantly changed since the 2.4 days.
it definantly no longer requires swap > ram or swap =2x ram (unless you are doing hibernation to your swap partition in which case you need swap > hibernation image + used swap)
Would caching really help? Posted Apr 21, 2010 17:15 UTC (Wed) by dlang (✭ supporter ✭, #313) [Link]
increasing swap will never solve a swap thrashing problem. a swap thrashing problem means that your working set is larger than ram and so you are pushing things into swap and then reading them back out again. having the space in swap be larger will make no difference.
Please make KDE SC lighter Posted Apr 18, 2010 8:45 UTC (Sun) by krake (subscriber, #55996) [Link] Cache more stuff and preload libs I will likely load...KDE actually does that, or more precisely kdeinit. It links against the most often used libraries so that they get loaded and their symbols resolved. When any main KDE application such as Konqueror starts, it is actually a fork of kdeinit with the programs specific core executed inside the new process.
Please make KDE SC lighter Posted Apr 18, 2010 10:29 UTC (Sun) by nix (subscriber, #2304) [Link]
If you prelink, it doesn't act this way. kdeinit (this aspect of it, anyway) is really a workaround for a lack of prelinking.
|
|||||||||||||
