Beryl

hellonorman · Post by **hellonorman** » Thu Feb 22, 2007 7:16 am

Is beryl really ahead of the curve if it's not stable enough to be included in Ubuntu?

Don't get me wrong I love beryl and can't wait to be able to use it without giving it another thought. But the fact is that I've tried it on 3 different machines and on 5 different distros and it always eventually becomes unstable and causes glitches to the point that I have to turn it off for good.

Whatever your feelings are about MS, aero glass is out and works and most people find it looks very nice. No matter whether beryl is better or not it's STILL unstable.

Oh well, I just felt a need to comment on it after hearing the argument on the Linux Action Show and then Dann agreeing with it on the last TLLTS.

CptnObvious999 · Post by **CptnObvious999** » Thu Feb 22, 2007 8:02 am

Maybe right now but like Bryan said from the Linux Action Show it has gone leaps and bounds, and by the time Feisty ships it should be pretty good.

chuck · Post by **chuck** » Thu Feb 22, 2007 11:03 am

CptnObvious999 wrote:Maybe right now but like Bryan said from the Linux Action Show it has gone leaps and bounds, and by the time Feisty ships it should be pretty good.

Yet when you're putting together a release at a particular time you can not just hope for the best. You have to take what you have and consider the track record for the project and what the likely hood is that everything will be done. And with a project as high profile as Ubuntu the front facing items must be solid.

snarkout · Post by **snarkout** » Thu Feb 22, 2007 11:45 am

I'm wondering exactly what the technical shortcomings of beryl are - I know there were recently several shots fired at beryl from the compiz team with their usual "It's Hackish, ugly and short" (or something to that effect), and with comments that beryl isn't "giving back." They also seemed pissed that beryl is stealing potential compiz devs because of the mindshare they have. It certainly seems to me that beryl has a lot more mindshare at this point, and has grabbed quite a bit of attention. I'm wondering why compiz devs didn't/don't have a devel release as well as a stable release. This could potentially deal with two issues at once - the code done up for their dvel release *could* and *would* be "given back," and they could capture the mindshare they need to push their project forward more quickly.

Of course, being largely a novel/gnome project that was initially developed in private probably hasn't earned compiz much trust/cred in some camps...

Personally, I don't know. I abhor hacks that are actually hackish and not clever workarounds (and I have no idea whetehr this describes beryl or not). On the other hand, berly is here and functional now, seems to do a lot of things compiz cannot, and may not be a lot buggier than compiz is - both have been very crash-prone IME.

Anyone have good, solid info on any of this? Thor?

hellonorman · Post by **hellonorman** » Sat Feb 24, 2007 5:56 am

Snarkout wrote:I'm wondering exactly what the technical shortcomings of beryl are

Yes it would be interesting to know. If they could put out a stable version with just the cube, the expose clone and a nice theme that would great. My feeling is the problems are more in the fundamental handling of the window manager stuff though.

Vogateer · Post by **Vogateer** » Sat Feb 24, 2007 11:31 am

My second nVidia based card just crapped out on me, so I can't even bother to test them both.

I wonder if there's something wrong with the cards or the AGP slot or something. Why have two cards wiped themselves out in my box? I only use compiz for 3D stuff, and I've only had this one for a few months.

allix · Post by **allix** » Sat Feb 24, 2007 11:36 am

Vogateer wrote:My second nVidia based card just crapped out on me, so I can't even bother to test them both.

I wonder if there's something wrong with the cards or the AGP slot or something. Why have two cards wiped themselves out in my box? I only use compiz for 3D stuff, and I've only had this one for a few months.

On a p3 box that still gets used has a nvidia tnt2 which works fine, i hope you do not fork out on a new card each time it brakes rather than taking it back and getting it replaced.
AFAIK software can not brake a video card, of course i could be wrong, but it would be well known if they was some virus in compiz.
thats sounds funny, but its a faulty video card and not a compiz virus.

Wally Balljacker · Post by **Wally Balljacker** » Sat Feb 24, 2007 12:45 pm

Not to start a flame war, but this is where Microsoft has the lead. Ultimately is doesn't matter if Linux has had XGL/AIGLX/Compiz/Beryl for over a year now, because it is really just a tech preview. It is no where near stable enough for mass adoption. Linux seems to always be playing catchup with Apple and Microsoft on the desktop, and right when we get competetive with "eye-candy", MS comes up with Aero, which people can actually use without hacking configuration files, and compromising system stability.

Vogateer · Post by **Vogateer** » Sat Feb 24, 2007 1:18 pm

I've been using Compiz, and I can't remember when I compiled it myself, but it's been pretty stable. The problem is that it doesn't revert back to Metacity when things go south. Also the nVidia drivers broke with the last kernel update, and the nv driver was fine for usual stuff, but couldn't manage compiz, obviously. This sort of thing would destroy confidence for a new user.

I just know I wouldn't install it on my sister's laptop, because if something happened, she'd be out of luck until I could drive up there and fix it. I think Linux is far less behind in this manner than we have been in other areas in the past, and the effects (according to this guy I know at Dell who's using Vista and saw my own compiz desktop) we do have even with Compiz are better, but I would disagree with Bryan from the Linux Action Show that they're stable and ready to be put out there for the average user. Once again, when you're changing someone over to another environment, no matter how much better, it's already a daunting task for that new user just to grow accustomed to the little changes, and the littlest thing can have a serious effect on someone's perception of that new thing. I don't want someone's first impression of Linux to be that we prefer splashy effects over stability.

[EDIT: Oh yeah, I've already sent an RMA to the company for my broken nVidia based card, and plan on sending it back. I think I bought the card last August. I doubt that compiz could cause problems, but I do wonder whether a bad motherboard could be the root cause.]

snarkout · Post by **snarkout** » Sat Feb 24, 2007 1:38 pm

Wally Balljacker wrote:Not to start a flame war, but this is where Microsoft has the lead. Ultimately is doesn't matter if Linux has had XGL/AIGLX/Compiz/Beryl for over a year now, because it is really just a tech preview. It is no where near stable enough for mass adoption. Linux seems to always be playing catchup with Apple and Microsoft on the desktop, and right when we get competetive with "eye-candy", MS comes up with Aero, which people can actually use without hacking configuration files, and compromising system stability.

Hell, apple has had quartz/luna forever. MS just caught up with them. Personally, even though I'm a dyed in the wool pessimist, I'm always amazed at just how awesome LOTD is. Mplayer and xine pwn. Amarok pwns. Yakuake pwns. Hell, most of my favorite apps are either linux-centric or at least run on linux. But, yeah, I guess a little more eye-candy couldn't hurt (even though I'm the kind of d00d who still uses gkrellm instead of *desklets), and of course it's foolish to think that all beryl/compiz are is "eye candy."

Vogateer · Post by **Vogateer** » Sat Feb 24, 2007 1:47 pm

I'm still surprised by how much I miss compiz when it's gone. Working on OS X without the wobbly window makes it feel so artificial. I can obviously live without it, but greatly prefer having it.

I still think that even the cube effect should have obvious usability enhancements. Before cube effect, my sister just sees her windows disappear, and thinks they crashed and vows to never use that feature and wants it off her machine entirely. With the cube effect, she sees the windows merely moved out of the way, and now that functionality is not remotely scary, but cool AND useful. The Alt+Tab functionality and expose like effect are both brilliantly done as well. If you don't see the functionality enhancements, I believe you're simply not trying to think like a regular user.

allix · Post by **allix** » Sat Feb 24, 2007 1:52 pm

Wally Balljacker wrote:Not to start a flame war, but this is where Microsoft has the lead. Ultimately is doesn't matter if Linux has had XGL/AIGLX/Compiz/Beryl for over a year now, because it is really just a tech preview. It is no where near stable enough for mass adoption. Linux seems to always be playing catchup with Apple and Microsoft on the desktop, and right when we get competetive with "eye-candy", MS comes up with Aero, which people can actually use without hacking configuration files, and compromising system stability.

Had XGL been open source a long time ago, thats right, XGL was closed source for a while, developers would have had more time to test it, work on it etc.
Ms and apple have research labs where they throw perhaps 100s of people at really experimental projects that more than half properly never make it into a product.

Open source on the other hand throws everything into the mix and makes sure everything is stable and included where possible.

Linux and open source as you rightly say is behind what they do, but we look at it differently, after all i do not want to see linux/bsd turning into another mac osx/windows.
Reactos is a windows clone for those who are interested not linux/bsd .

The open source movement has not until recently attracted artist to make a pretty desktop, we have been from the start full of console warriors.

Tsuroerusu · Post by **Tsuroerusu** » Sun Feb 25, 2007 4:30 pm

CptnObvious999 wrote:Maybe right now but like Bryan said from the Linux Action Show it has gone leaps and bounds, and by the time Feisty ships it should be pretty good.

You're basing that guesses, the Japanese have yet to invent the time machine, so you can't tell for sure. A year ago, I thought Compiz would be stable in a year, but guess what, it's not!
To throw Compiz onto regular users' desktops, you need to be able to say "this is stable, we know it's stable, because it's proven stable!". Now how do you "prove" that something is stable? You use a "tool" that has been here since the creation of the universe: Time!
Why is Linux so much more secure than Windows? Because time have proved it! Why are UNIX-like OSes good platforms today? Because time has fixed it! The original AT&T UNIX sucked. Today, free software UNIX-like OSes, written by volunteers I might add, are more stable, secure and the whole kitchensink, than the OSes made by a certain Northwest, multi-billion dollar american company.

chuck wrote:
CptnObvious999 wrote:Maybe right now but like Bryan said from the Linux Action Show it has gone leaps and bounds, and by the time Feisty ships it should be pretty good.
Yet when you're putting together a release at a particular time you can not just hope for the best. You have to take what you have and consider the track record for the project and what the likely hood is that everything will be done. And with a project as high profile as Ubuntu the front facing items must be solid.

Completely agree! Unless you have a monopoly, you can't just throw what you have out there, and hope that it will work for people and that people will use it, if you're aiming at expanding your userbase. When you're fighting a monopoly, you have to put stuff out that works, and works well, or people won't use it.
To step aside from software monopolies for a moment, look at AMD's battle against Intel. Especially with the release of Intel's Core 2 Duo chips, AMD just can't made a 10 billion GHz CPU and hope that it's gonna work for them, because a such CPU wouldn't work, it would meltdown faster than Chernobyl!
With Pentium 4, Intel made a buggy, very inefficient and expensive processor (Which is why Core 2 Duo are based on the Pentium 3 design, through the Pentium M work done by Intel's research center in Israel), and guess what! It cost them marketshare! But because they have a monopoly, it didn't hurt them that much in the end.
Right now, AMD's upcoming quad-core CPU sounds amazing, it really sounds like it's gonna kick-ass, but do I actually know that it's gonna be all that? NOPE! Because like I said, the Japanese have yet to invent the time machine, they're working on female cyborgs in the mean time ...

Snarkout wrote:I'm wondering exactly what the technical shortcomings of beryl are - I know there were recently several shots fired at beryl from the compiz team with their usual "It's Hackish, ugly and short" (or something to that effect), and with comments that beryl isn't "giving back." They also seemed pissed that beryl is stealing potential compiz devs because of the mindshare they have. It certainly seems to me that beryl has a lot more mindshare at this point, and has grabbed quite a bit of attention. I'm wondering why compiz devs didn't/don't have a devel release as well as a stable release. This could potentially deal with two issues at once - the code done up for their dvel release *could* and *would* be "given back," and they could capture the mindshare they need to push their project forward more quickly.

Of course, being largely a novel/gnome project that was initially developed in private probably hasn't earned compiz much trust/cred in some camps...

Personally, I don't know. I abhor hacks that are actually hackish and not clever workarounds (and I have no idea whetehr this describes beryl or not). On the other hand, berly is here and functional now, seems to do a lot of things compiz cannot, and may not be a lot buggier than compiz is - both have been very crash-prone IME.

Anyone have good, solid info on any of this? Thor?

THOU SHALL NOT DISHONOR THE HAMMER OF THOR!!!

......
Sorry, excuse me! Having a moment over here of honor over here.

The last time I tried Beryl, it really screwed me up, big time! My X server locked up and froze my keyboard, so I couldn't even access a virtual console and reboot the system, I had to use the power-button.
I have, personally, never used Compiz or Beryl in a real sense, because I can accept crashes or overall instability, and sadly, that's one of the things Compiz and Beryl do really really well in my opinion.
Also, since I have an NVIDIA graphics card (I bought it in 2003 when I was still to discover GNU/Linux as a desktop OS, I knew of it as a server OS), I can't get hardware accelerated 3D out of it, yet, without using a binary blob, and I'm not gonna sacrifice my software freedom (Which RMS shortens to just "freedom") for "bling", that don't really do much for me but provide instability.

To the Compizists and Berylans in here, I have not tried either one recently, so I'm sure they have improved a lot, but as I mentioned earlier, stability and security are proven, not claimed (In the words of Steve Gibson).

I do believe that things like the fade, cube and exposé-like effects are improvements to usability, with the cube it's really really easy to explain the concepts of virtual desktops to a new user. When I was in about 5th grade I was at a youth club (Where kids could stay after school and hang-out, and do all sorts of stuff), one of the adults there said "You can have like 7 desktops in Leenuux" (Yes, we danes are really awful at pronouncing stuff at times, sometimes you'll see people trying to use Danish gramma in English) and I was like "What?" and then I thought about it and kind of saw what it might look like, but it was not exactly what it is. I think usability can be improved in various ways by actually using cleaver ways to explain things or concepts to the users. I'm not talking about giving them a manual and say "RTFM!", but actually doing things that explain stuff to them in their mind. Keep in mind this is not the solution for everything, but it could be useful in a bunch of situations.

You mention technical shortcomings. The main one, is absolutely stability and reliability. People mention "fallbacks" to Metacity, and yes that is somewhat of a "solution", but it's a solution that shouldn't be needed. In the two years and eight months that I have been using GNU/Linux, and lately BSD, I have never had KWin (The KDE window manager) crash, never ever has it crashed and left my apps without a window manager and stuck to the "background". So in this case, a fallback is not really needed. I am sure that KWin has crashed for some people, but for me it hasn't. No matter what free OS that I have run KDE on, be it GNU/Linux, FreeBSD, OpenBSD etc., I have never experienced a key part of KDE crashing, sure I have had applications crash (Kaffeine, grrr!), but not like KWin, kdesktop. I have had kicker crash, because the code is like sausage (Just ask it's maintainer, mr. Aaron Seigo), and I have had weird behavior because I was using some bleeding-edge stuff, but not critical crashes.
I personally turn my PC on and shut it down, or merely reboot, two weeks later. So to me, stability and reliability is extremely important, and a window manager crash, is an immediate turn-off.

Also, something that I have heard, is that some legacy UNIX applications has problems running under Compiz and Beryl, which is a concern to companies, not us, but I'm just mentioning it since you asked.

In terms of contributing back, this is probably both a semi-political issue and a licensing issue. Compiz is under the X11 license (Also known as the MIT license) and Beryl is under the GPL. This is an issue which originated back in the 80' when RMS initially created the GPL, and the Berkeley hackers released BSD under the BSD license. Just like you can take permissively licensed code and make it proprietary, you can take software under a permissive free software/open source license and GPL it. However, you can't take GPLed code and put it under a different license, except for maybe a newer version of the GPL, which is the case with stuff that uses the "GPLv2 or later" licensing.

The last thing I want to add, is that Beryl really needs a more user-friendly configuration dialog. I think they are working on this, but I don't know for sure, but their current configuration program really really really sucks.

In my opinion, Beryl has really been doing what the OpenBSD people have accused the Linux kernel developers of doing (Since I don't know, I can't confirm or disconfirm this in the case of the Linux kernel) for a long time, of just shoving everything possible into the code and hoping it will work. If they had decided on a certain feature set about 8 months ago, and worked on just fixing bugs and improving stability, it could be better than it is today (Not sure though).

hellonorman wrote:
Snarkout wrote:I'm wondering exactly what the technical shortcomings of beryl are
Yes it would be interesting to know. If they could put out a stable version with just the cube, the expose clone and a nice theme that would great. My feeling is the problems are more in the fundamental handling of the window manager stuff though.

This would actually be quite easy to do, since Compiz is completely plugin-based, you could just strip out the plugins you didn't need or were buggy, and build a hellonorman version of Compiz.

Wally Balljacker wrote:Not to start a flame war, but this is where Microsoft has the lead. Ultimately is doesn't matter if Linux has had XGL/AIGLX/Compiz/Beryl for over a year now, because it is really just a tech preview. It is no where near stable enough for mass adoption. Linux seems to always be playing catchup with Apple and Microsoft on the desktop, and right when we get competetive with "eye-candy", MS comes up with Aero, which people can actually use without hacking configuration files, and compromising system stability.

True, Aero Glass may be extremely hardware demanding, but it works.
The thing is, Xgl is a sucky "solution", it adds complexity where it shouldn't be added, yes it works with the piece-of-crap ATi fglrx proprietary drivers, but that doesn't change the fact that as a non-research project "solution" stinks. AIGLX is the way to go, in my opinion, and it's basically finished and working!
And yes I agree (Surprise), that Beryl needs to be easier to get running! We talk about all the time about GNU/Linux having all sorts of cool feature compared to Vista, however if it virtually doesn't exists, because non-technical users can't get it running, then it's no better than suspending most laptops!

allix wrote:
Wally Balljacker wrote:Not to start a flame war, but this is where Microsoft has the lead. Ultimately is doesn't matter if Linux has had XGL/AIGLX/Compiz/Beryl for over a year now, because it is really just a tech preview. It is no where near stable enough for mass adoption. Linux seems to always be playing catchup with Apple and Microsoft on the desktop, and right when we get competetive with "eye-candy", MS comes up with Aero, which people can actually use without hacking configuration files, and compromising system stability.
Had XGL been open source a long time ago, thats right, XGL was closed source for a while, developers would have had more time to test it, work on it etc.
Ms and apple have research labs where they throw perhaps 100s of people at really experimental projects that more than half properly never make it into a product.

Open source on the other hand throws everything into the mix and makes sure everything is stable and included where possible.

Linux and open source as you rightly say is behind what they do, but we look at it differently, after all i do not want to see linux/bsd turning into another mac osx/windows.
Reactos is a windows clone for those who are interested not linux/bsd .

The open source movement has not until recently attracted artist to make a pretty desktop, we have been from the start full of console warriors.

XGL has actually never been proprietary, it was developed behind closed doors for some time, but that doesn't make it proprietary. As far as I know, Ximian/Novell Evolution was developed behind closed doors until the day it was announced.

Judland · Post by **Judland** » Sun Feb 25, 2007 6:50 pm

Tsuroerusu wrote:The last time I tried Beryl, it really screwed me up, big time! My X server locked up and froze my keyboard, so I couldn't even access a virtual console and reboot the system, I had to use the power-button.
I have, personally, never used Compiz or Beryl in a real sense, because I can accept crashes or overall instability, and sadly, that's one of the things Compiz and Beryl do really really well in my opinion..

Been using Compiz on my Mandriva 2007 system at the office since its release in October. I've never had it crash on me. This is using it nine hours a day, five days a week for five months now.

Can't say how other distros use/implement Compiz, but on Mandriva 2007 it's proven to be stable... at least to me.

Tsuroerusu · Post by **Tsuroerusu** » Mon Feb 26, 2007 6:02 am

Judland wrote:
Tsuroerusu wrote:The last time I tried Beryl, it really screwed me up, big time! My X server locked up and froze my keyboard, so I couldn't even access a virtual console and reboot the system, I had to use the power-button.
I have, personally, never used Compiz or Beryl in a real sense, because I can accept crashes or overall instability, and sadly, that's one of the things Compiz and Beryl do really really well in my opinion..
Been using Compiz on my Mandriva 2007 system at the office since its release in October. I've never had it crash on me. This is using it nine hours a day, five days a week for five months now.

Can't say how other distros use/implement Compiz, but on Mandriva 2007 it's proven to be stable... at least to me.

See this is the thing that concerns me. Compiz apparently does work quite well for you on your machine, but there's so many hardware combinations out there, and unlike Apple, we can't just test against two or three different combos. There's hardware combinations where Compiz and Beryl, apparently, works great, but there are also hardware combinations where it doesn't work at all, or causes system instability.