Nikos Chantziaras <rea...@arcor.de> posted gm3dmh$99...@ger.gmane.org, excerpted below, on Sun, 01 Feb 2009 07:58:54 +0200:
> Duncan wrote: >> [...] >> I can't say that Linus was wrong switching away for awhile, > > Linus said he never tried 4.2. And if he was happy with 3.5, why did he > go Gnome? That doesn't make any sense. Just because 4.1 isn't up to > your standards doesn't make 3.5 worse? As VAH pointed out, he's using Fedora, which pretty much dumped kde3. Thus, that wasn't really a choice for him, at least it wasn't unless he took the time to learn a new distribution, but even then it'd be an investment of time for something that may yet be great for some time, but let's face it, isn't getting much but maintenance development and eventually won't be getting even that. So I think Linus saw GNOME as the best use of his limited time ATM, being he just wants userspace to work well enough that it's not in his way when he's trying to work on the kernel. (Which, BTW, was the problem he blew up at GNOME on before, that they were dumbing everything down to the point that for power users it really DID get in the way.) > That's not an argument at all. Are you telling me those things are > working in 3.5? Er, 3.5 doesn't have compositing at all! That's incorrect. KDE 3.5's kwin made good use of xorg's composite extension from 3.5.1 and xorg 7.0, with user adjustable window transparency (set separately for active, inactive and moving windows), drop-shadows, animated open/close/minimize/maximize scaling and activate/ deactivate/transparency-change, menu transparency, etc. In fact, I still have posted a now rather dated screenshot from February of 2006, with its description the reason I know it was kde 3.5.1 and xorg 7.0. I was playing around with the then new composite, and posted a screenshot as various folks were asking about it. (The link may wrap.) http://members.cox.net/pu61ic.1inux.dunc4n/pix/screenshots/ screen.33.256.jpeg So KDE3, or at least its kwin component, has had composite support for some time. It just hasn't been as into-the-guts-of-things as it is with KDE4, because that was a lot of legacy code to rewrite to get it into the guts, which, with KDE4, they've basically done. >> The problem is that while it disables the effects themselves, there's >> little if any hint in the GUI what depends on OpenGL. I can checkmark >> them on and only by actually trying to invoke them and "nothing >> happens" do I see (rather, am left to /guess/) that said features >> require OpenGL. > > That's a bug then. Or I don't get what you mean. If I disable desktop > effects, all the checkboxes here (Improved Window Management, Shadows, > etc) become gray and inactive. No, I want effects in general, so that's enabled, but since I'm using xrender instead of opengl, it should disable all the opengl effects, since they do nothing. Some of the effects work fine with xrender, some work, but are slow, and some only work with opengl. What I'm saying is that when it's toggled to xrender, the opengl effects should all be dimmed-out/disabled in the list. Either that, or there should be separate lists for each. Either way, if the cube desktop switch effect (for example) isn't possible unless opengl is selected, that should be made clear -- it shouldn't even be selectable if the render method is set to xrender. But box-switch, to use an example that works perfectly fine with xrender, should still be selectable, because as I said it works just fine and in fact I have it enabled. >> And the features that work... without good OpenGL, are still relatively >> slow. Sure, they can be turned off, but there goes much of the reason >> one might otherwise find kde 4.x better, to date at least. > > Are you sure you have EXA acceleration enabled in the radeon driver? > Default is XAA which is slow as molasses :P That's: > > Option "AccelMethod" "EXA" Yes. exa is definitely enabled. It makes a BIG difference on KDE 3.5's composite processing and I agree with your XAA is slow as molasses opinion, so yes, it's definitely enabled! But thanks for asking. I'd be asking too. It never hurts to make sure! >> Without OpenGL and with some of the composite effects turned off that >> actually work reasonably well in 3.5 because they're too slow in 4.2... > > Hmm, last time I checked KDE 3.5 only had some kind of buggy > compositing. At least here all hell is breaking out if I enable it > (crashes mostly with lots of graphics glitches). It's quite stable on kde 3.5 here -- as long as I have EXA enabled, at least! As long as I keep to transparency only, it's reasonable speed, too. I keep the animations off as they DEFINITELY drag things down, and I just don't see the big deal with shadows so I keep them off too. That's how I ultimately configured kde 4.2 as well, composite and transparency enabled, box-switch enabled (presentation works too; I'll have to play with it a bit more before I decide which I like best), drop- shadows and scaling animations disabled. It definitely helped, but it's not to 3.5.10 speed yet. Then again, neither was 3.5.1. I expect the 4.2.1 and 4.2.2 bugfixes and optimizations will help in the speed department, after which it might actually compare at least decently with 3.5.10 in that regard. With the sheer amount of screen real estate I have to draw and the age of my card, I'm not kidding myself on speed. But, if 3.5.x can hit a reasonable speed with composite "bolted on", 4.2.x in a similar composite features configuration should be able to hit at least the same speed with composite designed into the recode from the beginning. It doesn't yet. But I do expect it to get there... it'll just take time and hard work. > That's not a good attitude. You're not given a product. You're given a > project and people are expecting constructive feedback rather than "I > demand you fix the bugs or else!" :P Which is one reason I'm appreciative of the fact that I didn't pay for it. =:^) I'm just a bit disappointed in what's there for an X.2 release, is all, and I've never been one to prioritize political correctness over honesty. If I'm taking the time to post about it, you better believe I'm giving my honest opinion on it. But I *DO* believe it'll get there! Maybe I'm a foolish optimist in that regard, but I do still believe it. Which of course makes the disappointment that much worse, as if I didn't believe it, I'd simply no longer care, much as I no longer care how bad MS screwed up with Vista, because I've left them to do their thing and moved on to personally better choices. Actually, something similar happened with Mandrake to push me to Gentoo, as well. I adopted it, got quickly into Cooker, and might well have stayed on it, contributing and bug testing. But Mandrake was pushing the club at the time, and I wanted to contribute money as well. Only funny thing, as soon as I started thinking about it in terms of monetary value, I began realizing how poor a fit it actually was for my needs. What made it worse was that I switched to amd64 at about that time, and while Mandrake was "supporting" amd64, their "support" consisted of taking the latest x86 main distribution packages (and I had been on Cooker, remember?) and making that the amd64 Cooker. So instead of being several months ahead of x86 main as I had with x86 cooker, I was now several months behind x86 main. Worse still was that on amd64, they were primarily targeting servers. Finally, they made it basically impossible to actually buy something useful for me from them, as they didn't offer a freedomware only product to buy and on the amd64 side, the only thing they /did/ offer was a set of CDs, which would be of virtually zero use to me since they were even more outdated than the already months behind amd64 cooker! Now there's nothing really wrong with all that, after all, it's where the big corporate money was. They LIKE that sort of stability aka being months and months behind, but it's not what I was looking for. And that was the point. It was a very bad match for me, and the more I looked at actually putting my money where my interest and time were at the time, the more I saw just how bad a match it was. So I did the smart thing and started looking for a better match. Gentoo is what I found. It was and remains VASTLY better for my needs, and now I spend a lot of time contributing to it, instead of Mandrake as before, or MS before that. >> But it's still nothing the FLOSS community can be proud of > > I totally disagree. KDE 4.2 is a milestone in GUI design and everyone > can be proud of it. Milestone, yes. As 4.0-beta1, they could be VERY proud of it indeed! As 4.2, meh, not so much. The hard work they can certainly be proud of, I agree. And it'll get there, I agree/hope/believe, as well. But there's a lot in a name, or a number, and 4.2... just isn't appropriate, nor is this level of "finishing touch" something to be proud of in a product labeled version 4.2. The features are there and work. The finishing touches, the optimization, the little stuff like making sure a user can't select something that you already know won't work for him anyway because you've tested and won't let him enable the necessary features, that's what should be in a 4.2 release, as opposed to a 4.0 release, but simply isn't. > I still don't get what you're talking about. What effects? Hopefully I made it plain above, but in case I didn't... (from memory as I'm in 3.5.10 ATM): On the desktop control applet/control/whatever-it's-called-now, there are several tabs. On I believe it's the front one, there's several "basic" options, including enabling/disabling effects entirely, and if enabled, whether xrender or opengl is used. On a tab further to the right, there's the big list of effects. If xrender is selected on the first tab, then all the effects on the effects list tab that require opengl should be dimmed/disabled, so the only ones that can be selected are those compatible with the chosen render method, xrender. That would be a HUGE usability improvement (Note that disabled doesn't necessarily mean their selected status changes. Thus, if someone had working opengl and had set it how they wanted, then switched to xrender to see what it did, it'd disable the opengl effects of course since xrender couldn't handle them, but they'd still remain as selected, so once the user figured out how much less functionality he had with xrender, he could switch back to opengl, and the setting he had with opengl previously would be still there. It wouldn't have deselected all of them just because he played with xrender, just dimmed them out to reflect their disabled state (since with xrender they wouldn't work in any case), as long as xrender remained the choice on the front tab.) >> and the infrastructure is already >> there, so why are they still forcing trial and error to find out what >> works and what doesn't, especially by X.2 (I could see it for x.98 >> betas, feature complete but rough aroudn the edges)? That makes no >> sense at all! > > I'm not playing dumb or something. I *really* don't understand what > you're referring to :P "Infrastructure" here refers to the fact that when I try to turn on OpenGL and hit apply, they already detect that it won't work on the full desktop and thus disable it -- they won't let me enable it, IOW. Since they already detect that, they already know whether OpenGL works on my system or not. Since they won't allow it to be enabled, they should simply disable all the individual effects requiring it, thus making it much simpler to choose between the ones I can actually use, instead of having more than half of them able to be /apparently/ turned on, but do absolutely nothing at all when actually tried. >> Meanwhile, khotkeys sort of works now... but the "extra" key I had most >> of my app-launchers stacked on (two-key sequence invocation), XF86WWW, >> works fine in 3.5.x and has since before 3.5, and it's detected on 4.2, >> but the message when I try to set it is, "Qt 4.x doesn't support the >> selected key." or something to that effect. WTF!? It can SEE the key >> and knows I pressed it in ordered to give me that message. It worked >> perfectly fine in 3.5 (and before)! What's this about not supported?! > > That's the only key that doesn't work. I suspect you're just angry that > KDE 4.2 isn't a perfect clone of KDE 3 with just the theme being > different and now every little thing that doesn't work seems like the > biggest bug ever to you. No, a clone wouldn't be progress. I really do believe (and see) progress. It's just that this isn't anywhere near the finish an X>0 X.2 software release normally has. By the time it reaches X.4 or X.5, people will hopefully be forgetting how bad X.0 thru X.2 were, and yes I do believe KDE4 will ultimately get there, but that's yet some way in the future. Meanwhile, we have the not so pleasant present to deal with. But, as I said it's actually working well enough to use and test and file bugs on now (what is commonly a W.95 or so release) or an X.0-rcY release), and I do fully intend to do so. Hopefully they'll be fixed by 4.2.1 or 4.3.0... which of course after fixing the bugs in what amounts to the -rc, that would make 4.3.0 what should have been the 4.0 release. > Me, I looked at KDE 4 with a positive eye from the start. Actually, as I pointed out above, it's that I'm still positive about it in general that's the whole reason it's so frustrating. If I didn't believe in what they were doing, I'd no longer care. I'd simply switch to something else, and let them continue doing what they will, just as I switched to something else from MS, and then from Mandrake, and let them continue on their way, because it was very plain they were headed somewhere very different from where I was headed, and it no longer made sense to continue with them. That I've not yet reached that point with KDE means I still believe in them, or I'd not be spending all this time discussing it, I'd simply move on. But it's that belief in them that makes it so frustrating! -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman