For one thing, it is (of course) theme-able, so if you don't like the green, change it. Already an improvement over MS-explorer.
>>>>>>>>>>>>>
Sorry, no dice. There are a bunch of themers for Windows.
And then, the file icons aren't just improved,
there's also varying levels of detail about each file based on how far you "zoom" in or out. So, you can go from just the plane-jane name up to "this file has X items and was last
modified at 4:30."
>>>>>>>>>>>>>>>>>>
Tracker, Explorer, and just about every file manager ever written allows you to set the attributes you want to show.
I dare say better than what we've currently got
>>>>>>>>>>
That's because most of what you currently have is pretty bad.
Yea, but the app_server handles a bunch of other stuff as well. It handles the entire UI (not just the graphics display portion) it does font management, event handling, and everything else except audio, networking, and input. And aside from remote display, X really don't have many more features (pending the render extension of course) that justifies its high memory use. Besides, QNX's Photon is comparable to X in features, includes UI elements, can do remote display, and is still MUCH smaller than X.
Question. Why don't the KDE developers have the decency to keep the binary releases up to date? By the time flat.tgz packages get out, the next release of KDE is already available. Second, why aren't optimizations compiled in to begin with? Everybody and their mother already has a PII, and releasing versions optimized for the most common chips (PII/PII, Athlon, and straight-Pentium) would make much more sense. Seriously, i386 builds should be the unusual ones, and i686 (maybe i586) should be the default ones. I want to see KDE compiled for i686 by default, with an obsolete/ folder from which I can get i386 builds.
A agree. To tell the truth, Slack is probably the best way to go, because the newbie distros actually make things harder. For example, Mandrake 7.2 includes an asnine set of wrapper scripts over the normal SysV scripts. Since these things are only documented in the Mandrake docs (which are fairly sparse) it makes it more confusing for the new user that is reading from a HOWTO that assumes SysV. In comparison, Slack uses the BSD scripts, which are quite well documented. I found that Slack tends to have much fewer idiocies than other distros of Linux. Rarely as I used Slack did the HOWTO I was following differ from what was going on. When I used RedHat, I could never follow the HOWTOs because RH messed around with things and put them in different places. And I recently deleted the stupid linkes that Mandrake put in the/etc directory (to the rc.d directories) and the system would't boot anymore!
This doesn't really qualify as competition. First, it runs only on PPC and 68k. That pretty much limits it to competing with Apple. Second, it isn't even plausible competition. Its like saying that it is good that the socialist workers party is there to keep the Dems and Reps honest. Yea, there there, but are they really?
I was talking about the tests from 0.9-4, before Linux's Detonator 3 drivers. However, Linux DOES still lose in hi-res tests, but a good margin sometimes. For example, 37fps in 16x12 is marginal, while 46 is quite playable. However, its a moot point. Even at 640x480, Linux is SLOWER than Windows. That's simply should not be the case. (Especially since NT4 has even faster OpenGL performance than Win2K)
However, BeOS isn't POSIX certified. While BeOS does have hooks in it (from the software's point of view) to be multi-user, it always has one user. If you look at the file permissions and all that, there is a user called "Baron" and a user group. However, they don't really have any effect other than allowing POSIX software (such as CVS, which reads the username) to function properly. The OS itself isn't multi-user, it justs pretends that it is for POSIX software. However, there was some talk about making it multi user, and the API is already ready for it (for example the find_directory call has constants for the particular user folders and common folders) but it would be a lot of work, and I don't really see a benifet right now.
Actually, the GeForce2-class chips already blows away the Oxygen series in terms of performance. If you take a look at the Intense3D benchmarks you'll find that the WildCat (designed by Intense3D, acquired by Intergraph, then acquired by 3Dlabs) is the fastest midrange workstation card out there. However, take a look at the Elsa Gloria II scores. The Gloria II is almost exactly a GeForce running at 130Mhz. (10MHz overclock) with some anti-aliasing and other features enabled. Its already around 50-70% of the performance of the WildCat, and I wouldn't be surprised if a GeForce2 Ultra comes within 20% of the performance of a WildCat. At around 1/4 the cost. While 3DLabs might have some tricks up their sleeves, they better get those tricks out fast, because NVIDIA is not only taking over consumer space, but has a great chance at the workstation market as well. As for Matrox, you have to respect them. The G400 MAX was a little late, but had the best visual quality of any card (and still does) and was most of the speed of a TNT2 Ultra.
A) I get the 30% from the tests AnandTech and TomsHardware did of the 3D performance between Windows2K and Linux. And if you get the same performance with Win2K and Win98 for OpenGL, then there is something wrong with your setup (Win2K is significantly faster for most OpenGL apps)
B) I use Netscape6.
C) XF86Setup won't help anything. I have no problem using xf86config, it's a perfectly user-friendly program. Even the XF86Config file is pretty friendlY (nicely commented) What my problem is that there is no setting that allows you to set your refresh rate, and X-4 is limited to VESA modes, which means that getting anything above 75Hz on your 1152x864 monitor means getting modelines. Of course, modelines have a 30page HOWTO explaining them, and not everybody has a BeOS machine they can access and steal modlines from. (BeOS autodetects monitor modes) Its a moot point anyway. XFree86Config only exists because X is too stupid to detect anything for itself. About the only thing that should be there is fontpath.
truetype fonts work wonderfully
>>>>>>
Well, they work. And uh, I can load them. And umm, it doesn't crash when using them... BUT, they look aweful, all jaggedy. And its not just the lack of anti-aliasing, X fonts genuinely suck. But they DO work.
3d graphics are extremely fast
>>>>>>>>
No they are not. Even Windows 2000 (which isn't exactly the pinnacle of OS design) is 30% or so faster. And that's only because NVIDIA was nice enough to give X some pro-caliber drivers. (BTW, the NVIDIA drivers are around 20-30% faster on average than the stock XFree ones for 2D primitives)
stability is good
>>>>>>>>>>
I'll give you that.
X remote display works well even over a modem link
>>>>>>>>
True.
and even the notorious X bloat has been put on a diet
>>>>>>>>
Wow. Look mom! I lost 200lb and I now only weigh 350!
There are many well known defiencies, such as limited suport for drawing primitives, lack of alpha channel support, not to mention lack of dynamic resolution/ color depth changing
>>>>>>>
Which all other windowing systems have had forever
but they are all comming down the pipeline
>>>>>>>>>>
Real soon now!
Not to mention that it is still a bitch to configure and for something that has been worked on so much, it is quite underhwelming.
The BeOS GUI is independant of the kernel. I hope you don't think that the GUI is in the kernel, do you? And if you'd actually USE the BeOS instead of look at benchmarks, you'd agree with the BeOS users. The truth is, that benchmarks are rare for BeOS and are a pain to write. That's why you don't see them. The sheer fact that you can run a dozen AVIs on an 8proc machine (see BeNews) and can do video mixing, and audio mixing with dozens of channels is how we know BeOS is faster. Try some of the BeOS (particularly the audio apps) and see if you can find some that work as well on Linux.
What you miss, is that BeOS offers a level of user good feeling that approaches nirvana. BeOS makes you feel good, just like Linux makes you feel 'leet.
The fonts are monkey-ass ugly. It is hard to configure if you have good hardware. (VESA modes limited) It is slow, and takes up an insane amount of RAM. The 40MB that X takes up on my machine is more than the entire BeOS does. The simple fact that X contains 20+MB of binaries is hideos given that its equivilent in BeOS is contained within the 3.1 MB app_server.
A) Not multi-user, big loss. To tell the truth, I really don't miss multi-user. Some might, but I can guarentee you that the average desktop user won't.
B) Linux has crashed more often (devel kernels) than BeOS. What kind of stunt are you trying to pull. BeOS has been rock solid ever since DR8 (which was three or four years ago) Of course, it could be a problem with your hardware, in which case I suggest no buying a Packard Bell next time.
C) The filesystem is fast (somewhere between ext2 and Reiser, closer to ext2) reliable, and feature filled. It has features that are still in the developmental stages for ResierFS. What more do you want?
D) "emulated under Linux." That's the whole problem. The BeOS GUI isn't great because it looks spiffy, but it has a level of "Zen" unmatched by any Linux GUI. In Linux, it pains me to use the GUI sometimes. I end up just using xterm. But in BeOS, the GUI just feels so homey. (Of course, there is the terminal there too)
Yes excactly. That's what Linux is aimed at isn't it? GNOME and KDE are examples of "good" competition because they're trying to be best for the user? Not. It seems to me that a lot of "competition" in the Linux arena is just a pissing contest between developers. If KDE and GNOME were trying to be good for the user, they'd make a common binary API, and then duke it out over who could make the fastest/smallest/most feature filled desktop out there. Linux is turning into Windows as we speak. It already takes up as much RAM, and newbie type things are being installed that tries to trap growing users into using them. (One problem with both Linux and Windows is that they don't grow with the user. For example, instead of having scalable (to the user) set of initscripts, Mandrake 7.2 implements a hack that wraps a set of scripts over the standard SysV scripts. Ugly, inelegant, and totally confusing the the user who is trying to graduate from the simplfied ones to the real ones.) Every release (I'm talking about the upper stuff like Mozilla, X, GNOME, KDE, etc) is getting criminally bloated and features are added just for the hell of it. Linux NEEDS a competitive BeOS to keep it from getting fat and lazy just like MS.
Yes, that is EXACTLY what has happened with OpenBSD. OpenBSD is such a failiure. I mean, a project led by purists concerned with stability, good god, what WERE they thinking?
Give me a break. If BeOS was lead by a core team focused on keeping it BeOS-like, then it would kick serious ass.
I'd like to see an OSS BeOS too, with a tightly controlled development model like OpenBSD (to keep the small, fast, powerful, sexy focus) but like he said, it ain't going to happen. Even if Be wanted, a lot of BeOS is prorietory code (like the font rendering engine from BitStream, for example. BTW: Anybody hear what happened to the Font Fusion server BeOS was supposed to get?) On a morbid not, however, I would hope, that if Be ever goes under, they'd open the non-proprietory parts of the code so people could quickly fill in the gaps and keep BeOS going.
Slashdot Mirror
For one thing, it is (of course) theme-able, so if you don't like the green, change it. Already an improvement over MS-explorer.
>>>>>>>>>>>>>
Sorry, no dice. There are a bunch of themers for Windows.
And then, the file icons aren't just improved,
there's also varying levels of detail about each file based on how far you "zoom" in or out. So, you can go from just the plane-jane name up to "this file has X items and was last
modified at 4:30."
>>>>>>>>>>>>>>>>>>
Tracker, Explorer, and just about every file manager ever written allows you to set the attributes you want to show.
I dare say better than what we've currently got
>>>>>>>>>>
That's because most of what you currently have is pretty bad.
Yea, but the app_server handles a bunch of other stuff as well. It handles the entire UI (not just the graphics display portion) it does font management, event handling, and everything else except audio, networking, and input. And aside from remote display, X really don't have many more features (pending the render extension of course) that justifies its high memory use. Besides, QNX's Photon is comparable to X in features, includes UI elements, can do remote display, and is still MUCH smaller than X.
Question. Why don't the KDE developers have the decency to keep the binary releases up to date? By the time flat .tgz packages get out, the next release of KDE is already available. Second, why aren't optimizations compiled in to begin with? Everybody and their mother already has a PII, and releasing versions optimized for the most common chips (PII/PII, Athlon, and straight-Pentium) would make much more sense. Seriously, i386 builds should be the unusual ones, and i686 (maybe i586) should be the default ones. I want to see KDE compiled for i686 by default, with an obsolete/ folder from which I can get i386 builds.
A agree. To tell the truth, Slack is probably the best way to go, because the newbie distros actually make things harder. For example, Mandrake 7.2 includes an asnine set of wrapper scripts over the normal SysV scripts. Since these things are only documented in the Mandrake docs (which are fairly sparse) it makes it more confusing for the new user that is reading from a HOWTO that assumes SysV. In comparison, Slack uses the BSD scripts, which are quite well documented. I found that Slack tends to have much fewer idiocies than other distros of Linux. Rarely as I used Slack did the HOWTO I was following differ from what was going on. When I used RedHat, I could never follow the HOWTOs because RH messed around with things and put them in different places. And I recently deleted the stupid linkes that Mandrake put in the /etc directory (to the rc.d directories) and the system would't boot anymore!
A) How is it not the point? You said BeOS couldn't do remote display, I pointed out that it could.
B) Who cares if those capabilities are native? Technically, Linux doesn't natively have a GUI. Does that make Linux GUI's any less good?
This doesn't really qualify as competition. First, it runs only on PPC and 68k. That pretty much limits it to competing with Apple. Second, it isn't even plausible competition. Its like saying that it is good that the socialist workers party is there to keep the Dems and Reps honest. Yea, there there, but are they really?
I was talking about the tests from 0.9-4, before Linux's Detonator 3 drivers. However, Linux DOES still lose in hi-res tests, but a good margin sometimes. For example, 37fps in 16x12 is marginal, while 46 is quite playable. However, its a moot point. Even at 640x480, Linux is SLOWER than Windows. That's simply should not be the case. (Especially since NT4 has even faster OpenGL performance than Win2K)
Of course you can. BeOS has an X server. And then there is VNC and a BeOS-native program as well.
However, BeOS isn't POSIX certified. While BeOS does have hooks in it (from the software's point of view) to be multi-user, it always has one user. If you look at the file permissions and all that, there is a user called "Baron" and a user group. However, they don't really have any effect other than allowing POSIX software (such as CVS, which reads the username) to function properly. The OS itself isn't multi-user, it justs pretends that it is for POSIX software. However, there was some talk about making it multi user, and the API is already ready for it (for example the find_directory call has constants for the particular user folders and common folders) but it would be a lot of work, and I don't really see a benifet right now.
Are you sure? A context switch every 250 microseconds is pushing it.
Actually, the GeForce2-class chips already blows away the Oxygen series in terms of performance. If you take a look at the Intense3D benchmarks you'll find that the WildCat (designed by Intense3D, acquired by Intergraph, then acquired by 3Dlabs) is the fastest midrange workstation card out there. However, take a look at the Elsa Gloria II scores. The Gloria II is almost exactly a GeForce running at 130Mhz. (10MHz overclock) with some anti-aliasing and other features enabled. Its already around 50-70% of the performance of the WildCat, and I wouldn't be surprised if a GeForce2 Ultra comes within 20% of the performance of a WildCat. At around 1/4 the cost. While 3DLabs might have some tricks up their sleeves, they better get those tricks out fast, because NVIDIA is not only taking over consumer space, but has a great chance at the workstation market as well. As for Matrox, you have to respect them. The G400 MAX was a little late, but had the best visual quality of any card (and still does) and was most of the speed of a TNT2 Ultra.
A) I get the 30% from the tests AnandTech and TomsHardware did of the 3D performance between Windows2K and Linux. And if you get the same performance with Win2K and Win98 for OpenGL, then there is something wrong with your setup (Win2K is significantly faster for most OpenGL apps)
B) I use Netscape6.
C) XF86Setup won't help anything. I have no problem using xf86config, it's a perfectly user-friendly program. Even the XF86Config file is pretty friendlY (nicely commented) What my problem is that there is no setting that allows you to set your refresh rate, and X-4 is limited to VESA modes, which means that getting anything above 75Hz on your 1152x864 monitor means getting modelines. Of course, modelines have a 30page HOWTO explaining them, and not everybody has a BeOS machine they can access and steal modlines from. (BeOS autodetects monitor modes) Its a moot point anyway. XFree86Config only exists because X is too stupid to detect anything for itself. About the only thing that should be there is fontpath.
QNX Photon is small, fast, featureful, has great fonts, and has remote display to boot. In other words, it's every thing X should be, but isn't.
truetype fonts work wonderfully
>>>>>>
Well, they work. And uh, I can load them. And umm, it doesn't crash when using them... BUT, they look aweful, all jaggedy. And its not just the lack of anti-aliasing, X fonts genuinely suck. But they DO work.
3d graphics are extremely fast
>>>>>>>>
No they are not. Even Windows 2000 (which isn't exactly the pinnacle of OS design) is 30% or so faster. And that's only because NVIDIA was nice enough to give X some pro-caliber drivers. (BTW, the NVIDIA drivers are around 20-30% faster on average than the stock XFree ones for 2D primitives)
stability is good
>>>>>>>>>>
I'll give you that.
X remote display works well even over a modem link
>>>>>>>>
True.
and even the notorious X bloat has been put on a diet
>>>>>>>>
Wow. Look mom! I lost 200lb and I now only weigh 350!
There are many well known defiencies, such as limited suport for drawing primitives, lack of alpha channel support, not to mention lack of dynamic resolution/ color depth changing
>>>>>>>
Which all other windowing systems have had forever
but they are all comming down the pipeline
>>>>>>>>>>
Real soon now!
Not to mention that it is still a bitch to configure and for something that has been worked on so much, it is quite underhwelming.
20meg for a windowing server? That's criminal. That means that of the 40MB or so that Linux takes up on my system, half of that is X?
The BeOS GUI is independant of the kernel. I hope you don't think that the GUI is in the kernel, do you? And if you'd actually USE the BeOS instead of look at benchmarks, you'd agree with the BeOS users. The truth is, that benchmarks are rare for BeOS and are a pain to write. That's why you don't see them. The sheer fact that you can run a dozen AVIs on an 8proc machine (see BeNews) and can do video mixing, and audio mixing with dozens of channels is how we know BeOS is faster. Try some of the BeOS (particularly the audio apps) and see if you can find some that work as well on Linux.
What you miss, is that BeOS offers a level of user good feeling that approaches nirvana. BeOS makes you feel good, just like Linux makes you feel 'leet.
Actually, check out this page for a nifty logo that Be should adopt.
The fonts are monkey-ass ugly. It is hard to configure if you have good hardware. (VESA modes limited) It is slow, and takes up an insane amount of RAM. The 40MB that X takes up on my machine is more than the entire BeOS does. The simple fact that X contains 20+MB of binaries is hideos given that its equivilent in BeOS is contained within the 3.1 MB app_server.
A) Not multi-user, big loss. To tell the truth, I really don't miss multi-user. Some might, but I can guarentee you that the average desktop user won't.
B) Linux has crashed more often (devel kernels) than BeOS. What kind of stunt are you trying to pull. BeOS has been rock solid ever since DR8 (which was three or four years ago) Of course, it could be a problem with your hardware, in which case I suggest no buying a Packard Bell next time.
C) The filesystem is fast (somewhere between ext2 and Reiser, closer to ext2) reliable, and feature filled. It has features that are still in the developmental stages for ResierFS. What more do you want?
D) "emulated under Linux." That's the whole problem. The BeOS GUI isn't great because it looks spiffy, but it has a level of "Zen" unmatched by any Linux GUI. In Linux, it pains me to use the GUI sometimes. I end up just using xterm. But in BeOS, the GUI just feels so homey. (Of course, there is the terminal there too)
It wouldn't make much of a difference. The BeOS microkernel and HURD are as different as fast and slow.
Yes excactly. That's what Linux is aimed at isn't it? GNOME and KDE are examples of "good" competition because they're trying to be best for the user? Not. It seems to me that a lot of "competition" in the Linux arena is just a pissing contest between developers. If KDE and GNOME were trying to be good for the user, they'd make a common binary API, and then duke it out over who could make the fastest/smallest/most feature filled desktop out there. Linux is turning into Windows as we speak. It already takes up as much RAM, and newbie type things are being installed that tries to trap growing users into using them. (One problem with both Linux and Windows is that they don't grow with the user. For example, instead of having scalable (to the user) set of initscripts, Mandrake 7.2 implements a hack that wraps a set of scripts over the standard SysV scripts. Ugly, inelegant, and totally confusing the the user who is trying to graduate from the simplfied ones to the real ones.) Every release (I'm talking about the upper stuff like Mozilla, X, GNOME, KDE, etc) is getting criminally bloated and features are added just for the hell of it. Linux NEEDS a competitive BeOS to keep it from getting fat and lazy just like MS.
1) Be would rule the world.
2) It's not.
3) Slackware all the way.
Yes, that is EXACTLY what has happened with OpenBSD. OpenBSD is such a failiure. I mean, a project led by purists concerned with stability, good god, what WERE they thinking?
Give me a break. If BeOS was lead by a core team focused on keeping it BeOS-like, then it would kick serious ass.
I'd like to see an OSS BeOS too, with a tightly controlled development model like OpenBSD (to keep the small, fast, powerful, sexy focus) but like he said, it ain't going to happen. Even if Be wanted, a lot of BeOS is prorietory code (like the font rendering engine from BitStream, for example. BTW: Anybody hear what happened to the Font Fusion server BeOS was supposed to get?) On a morbid not, however, I would hope, that if Be ever goes under, they'd open the non-proprietory parts of the code so people could quickly fill in the gaps and keep BeOS going.