Nobody's complaing about its 'speed' of installation, it'd be really depressing if the number of menus made it take more than 10 minutes to install a rig; but the design really puts people off. I didn't really see it (having gotten used to it after a couple of rounds) but when I showed my friends the system they all agreed that it was clearly made by someone who is either hazardously stupid or just hates people and wants them to die.
Here's how to install only a boot loader, folks:
Go into a custom setup, start partitioning. Don't change anything.
Press 'w' (this is the secret they don't want you to know)
When prompted, pretend you know what you're doing. For some reason there's a difference in partitioning between installation and configuration
Select the mode of boot loader you want, or none at all (in which case you wasted your time)
It will fail if anything is 'special' about your drive. Care to argue?
This HOWTO from the "in any other system it'd be a one-liner" department.
I think DragonFly's biggest problem is that most ports don't build yet, and keeping the ports and overrides tree in synch is a lot of work that they don't have the man power for. This will all go away when pkgsrc gets testing on DragonFly (if you look at the blog, it should bootstrap there already), but that also halves the number of packages available.
I'd probably be running DragonFly now if I could actually get anything but the base system compiled. Shame having to go back to running Gentoo Linux just to get all the software up without hassle; but I'll admit overall it's really not that bad. The moment DFly takes a firm stand in package management, I'll try it again:)
Why not speak out on the mailing lists and explain why it's all good functionality? Given some of the other things they've been doing lately, it's hardly unreasonable. Provide a good case and it could be done.
I for one quite like/stand/sysinstall. It gives new and exciting challenges in circular interface design, undocumented features, misleading procedures (X* distribution acting like base system, installing like package), and being completely incapable of sensible disklabeling... these things make the perfect installer for any masochist.
DragonFly's installer is basically a more abstract and flexible brother to NetBSD 2's installer, though I'm not sure if they had that in mind. But the partitioning (BSD partitioning, no real DOS partitioning yet) is almost identical, just less functional. It's definitely got plenty of potential and once they get all the features ready will [hopefully] replace the others.
Maybe someone will even wrap a Gentoo distribution it so that finally has a good installation procedure. I swear, whoever thinks that installing a smaller package at the start is better for resulting performance is an idiot.
Re:Errata 2 - the Rob Enderle links I posted...
on
NetBSD 2.0 Released
·
· Score: 1
Care to give one example of this 'beating'? Keepin mind that acting like you've won doesn't mean you've won.
How do you say '2.0.3' in roman numerals? I've never heard of them even supporting decimals, let alone meta-decimals (well, what else are we to call the dot after a dot?), and certainly no zero.
Re:Errata 2 - the Rob Enderle links I posted...
on
NetBSD 2.0 Released
·
· Score: 1
Actually, that makes sense. You prove social (even psychological) trends by looking at real people. Slashdot contains, believe it or not, real people, and ulib identified a real trend among some of them.
So the moron is the person who calls someone, someone who is doing things exactly the right way, a moron and a zealot... i.e. you. It was entertaining for the rest of us though, so don't feel too bad.
Well, BSDs were making claims about their superiority, GNU/Linux makes claims about everyone else's inferiority. It's the difference between a snob and an asshole, which is still a significant difference.
FUD = Fear, Uncertainty and Doubt. BSDs don't spread that. They usually stick up for their own technical strengths, even if they aren't significant in the real world. BSDs are like that: has to be better during development, not necessarily use. Of course they have awesome use, but sometimes things just don't work out. They're not irrelevant or dead though, far from. Just not for everyone. I've heard a lot of "unsuccess" stories with BSDs, much less with Linux, which is disturbing but sensible; Linux has a much larger user and developer base, so problems tend to get resolved one way or another. People don't care if a solution is a hack or not, so long as it works for them.
Re:xBSD is indeed D.E.A.D. - Clarification
on
NetBSD 2.0 Released
·
· Score: 1
I greatly support your post, except that OpenSSL is not a BSD project (at least not the OpenSSL we know today). Firstly, its code is measurably filthy, but more importantly, it's under an Apache-style license (which is pretty evil compared to a BSD license). OpenSSH IS BSD however.
Right now I'm very ticked off at reality. Every OS I've used has at least some huge flaw which makes me want to run another, which leads to yet another huge flaw. Yes, even NetBSD 2. Said flaws wouldn't even be hard to fix, but it's as if the world just won't allow any one system to be perfect, and toys with the minds of developers and corporations to prevent such a thing happening.
If you want to reinvent the wheel to be object-oriented, go for it. But first write a new language in which to do it. C++ has way too many shortcomings to write the foundations of a whole system in.
The reason I prefer procedural languages to object-oriented ones is because computing itself is procedural. Object orientation feels like a hack. It's a weird way I look at it, yes, but it takes me personally a lot more effort to implement something in an object-oriented way than the same functionality in a procedural way, and the procedural way ends up cleaner and more efficient in the end. That's just me.
Hey, I wouldn't mind seeing a whole revolution in software design for FOSS systems, but the fact of the matter is that it would break a lot of familiarity and compatibility, and if it isn't good in the end (which is very possible) it will have all been for nothing. Not changing anything for decades is a good thing in this respect - there's no way to go 'wrong'.
NetBSD import some of OpenBSD too; they got EHCI interrupt routing from OpenBSD, meaning you can use USB2.0 hubs and the devices beyond them. FreeBSD still needs this (or working EHCI support to begin with; I've never heard a success story). Seen on mailing lists that multiple channel support on emuxki is going to be ported too, or at least looked at.
If you could store more than a few bytes of data in your head at once, you'd notice I never said anything about microkernels or such designs. Nevertheless, the BSDs are much more engineered than Linux is: they see a problem, set a goal for a solution, and work towards it. It gets done over a notably long time but it's done properly. Linux usually has lots of people trying their hand at solving a problem (how many new IO and CPU schedulers are in testing now?) and the most likable solution 'wins' and becomes the default or only option. It's the difference between monarchy and anarchy; something is grown to the top in BSD, or it fights to the top in Linux.
What's your point about not making progress for 20 years? What kind of progress are you after? They've managed to make kernels that perform well on modern systems, userlands that do what modern systems should do (even if they include nvi instead of vim), and documentation that is up-to-date. Technically Linux only provides the former of those, providing no definitive userland and virtually no useful documentation. If Linux itself was a project that involved attention to code quality and consistency, documentation as the code is written, and at least some attempt at release engineering more thorough than 'tar -cf - linux-x.y.z | bzip2 -9 > linux-x.y.z.tar.bz2', it would move slower but more people who need software integrity would take it seriously.
How about the SCSI system being completely changed between two MINOR releases of Linux 2.6, with practically no warning? The change wasn't even backward compatible. That's not the work of developers that believe in release engineering. Major versions are where big changes come in, not minor ones. Especially if said changes have no real reason (Linus just "didn't like it") and are very disruptive (if cdrecord couldn't do plain ATAPI control things would really be bad).
Security record of Linux isn't that great either. It needs one big bad code audit, and those 'hardened' and 'security enhanced' patch sets have to be merged in wherever things don't break standards. And some effort should be made to randomize kernel data; of all FOSS OSs, Linux is the easiest to predict even by nmap.
Ack, I had URL: in the first copy, but it didn't fit in so I had to remove other bits of the sig; now it should work./. puts way too short a limit on sig length.
It's about time we make a single comment including references to all of those articles AND those that include benches, technical reports, and other useful things. Someone people manage to consider those articles 'not good enough'.
Check that, next boot into a new kernel (which included ACPI support and so on) made the problem go away. Now sound is to hardware maximum, about as good as in Windows (which, you'll admit, continues to have a better sound system than ALSA, even if only because drivers are written by people who know the hardware). Sitting with ALSA for months made me forget what this card could really do.
Are you a Nazi? "Why waste valuable oxygen on Jews and gays when Germans have so much more... consonants in their language". There's a blessing to variety. BSDs have something Linux doesn't; engineering. Momentum isn't everything. I could run for miles and just die on the spot afterwards. Or I could walk forever and get a lot of things done on the way. You think about it and post again with what you think momentum is good for.
Really? That stinks. I have a Realtek that's handling the packets needed to post this, and it's running gloriously on NetBSD 2. But then it's not as simple as that I guess.
My one problem is that auich still has weird calibration so my laptop's sound card plays everything slower than it should be.
Shame really, great system but with some driver bugs. I guess sometimes the simplest possible driver code isn't enough.
Not everyone gets corporate sponsorship based on media hype. Linux gets all the easy breaks. Does this make it a good system? Not really. Just means that it goes in every direction with limited quality.
Re:Printed documentation (diff NET/FREE BSD)
on
NetBSD 2.0 Released
·
· Score: 1
Good to hear you like it, but why did you expect rubbish? BSDs are known for good documentation. Personally I didn't even know there was a PDF available, but hey that's the way it is.
Even if it doesn't work out for you, it's always a good learning experience.
Of course. Didn't you know that all open source software initially starts in the Linux source tree and eventually grows out, but still has major contributions by Linus and all the corporate sponsors?
That's how media hype around "software for Linux" makes it sound, anyway. It's like everything open is immediately for Linux and somehow spawned by the same community.
Anyway, using any GPL-licensed software as a success story for Linux is ridiculous. They share a license so they're the same project?
Simple answer: It doesn't. I don't think one person notice OpenBSD, for instance, scaled badly in their use. It takes literally thousands of iterations before the problems become really apparent (except some really flaky cases, let's not mention those) and getting under that kind of load is pretty rare for what most people need. I hate when people say "my 2-user file server will be much faster under Linux, look at those benchmarks!" when the non-scalable solution is faster for low load than the scalable one (programmers, think how much it takes to fill the first ten spaces in an array, versus the same data into a binary tree).
But nevertheless, I was impressed by Linux 2.6 having consistently fast and scalable results for those tests; while that doesn't impress me enough to want to use it for anything serious, it's nice to know they're wasting time on that instead of getting a code cleanup and audit done. Gotta have someone do 'software engineering' the Wrong Way, right?
You possibly misconfigured your X server's refresh rates. These things happen; best bet is to Google for your model and try everything.
I wish to celebrate now: XOrg supports my laptop's video card in-tree with the nv driver (previously it didn't), so I can run NetBSD fine. No acceleration or anything, but then I never used it in the first place.
Here's someone who has NetBSD on a range of machines with no Linux anywhere. A happy, happy admin.
Slashdot Mirror
Imagine a Slashdot Effect on a physical location like a single store. Good thing walking places isn't as convenient as clicking hyperlinks.
"Damnit, Bob's Hardware imploded from the gravity off too many Slashdot users visiting at once."
Nobody's complaing about its 'speed' of installation, it'd be really depressing if the number of menus made it take more than 10 minutes to install a rig; but the design really puts people off. I didn't really see it (having gotten used to it after a couple of rounds) but when I showed my friends the system they all agreed that it was clearly made by someone who is either hazardously stupid or just hates people and wants them to die.
Here's how to install only a boot loader, folks:
Go into a custom setup, start partitioning. Don't change anything.
Press 'w' (this is the secret they don't want you to know)
When prompted, pretend you know what you're doing. For some reason there's a difference in partitioning between installation and configuration
Select the mode of boot loader you want, or none at all (in which case you wasted your time)
It will fail if anything is 'special' about your drive. Care to argue?
This HOWTO from the "in any other system it'd be a one-liner" department.
I think DragonFly's biggest problem is that most ports don't build yet, and keeping the ports and overrides tree in synch is a lot of work that they don't have the man power for. This will all go away when pkgsrc gets testing on DragonFly (if you look at the blog, it should bootstrap there already), but that also halves the number of packages available.
:)
I'd probably be running DragonFly now if I could actually get anything but the base system compiled. Shame having to go back to running Gentoo Linux just to get all the software up without hassle; but I'll admit overall it's really not that bad. The moment DFly takes a firm stand in package management, I'll try it again
Why not speak out on the mailing lists and explain why it's all good functionality? Given some of the other things they've been doing lately, it's hardly unreasonable. Provide a good case and it could be done.
I for one quite like /stand/sysinstall. It gives new and exciting challenges in circular interface design, undocumented features, misleading procedures (X* distribution acting like base system, installing like package), and being completely incapable of sensible disklabeling... these things make the perfect installer for any masochist.
DragonFly's installer is basically a more abstract and flexible brother to NetBSD 2's installer, though I'm not sure if they had that in mind. But the partitioning (BSD partitioning, no real DOS partitioning yet) is almost identical, just less functional. It's definitely got plenty of potential and once they get all the features ready will [hopefully] replace the others.
Maybe someone will even wrap a Gentoo distribution it so that finally has a good installation procedure. I swear, whoever thinks that installing a smaller package at the start is better for resulting performance is an idiot.
Care to give one example of this 'beating'? Keepin mind that acting like you've won doesn't mean you've won.
How do you say '2.0.3' in roman numerals? I've never heard of them even supporting decimals, let alone meta-decimals (well, what else are we to call the dot after a dot?), and certainly no zero.
Actually, that makes sense. You prove social (even psychological) trends by looking at real people. Slashdot contains, believe it or not, real people, and ulib identified a real trend among some of them.
So the moron is the person who calls someone, someone who is doing things exactly the right way, a moron and a zealot... i.e. you. It was entertaining for the rest of us though, so don't feel too bad.
Or for starting "The War on Teabags"
Well, BSDs were making claims about their superiority, GNU/Linux makes claims about everyone else's inferiority. It's the difference between a snob and an asshole, which is still a significant difference.
FUD = Fear, Uncertainty and Doubt. BSDs don't spread that. They usually stick up for their own technical strengths, even if they aren't significant in the real world. BSDs are like that: has to be better during development, not necessarily use. Of course they have awesome use, but sometimes things just don't work out. They're not irrelevant or dead though, far from. Just not for everyone. I've heard a lot of "unsuccess" stories with BSDs, much less with Linux, which is disturbing but sensible; Linux has a much larger user and developer base, so problems tend to get resolved one way or another. People don't care if a solution is a hack or not, so long as it works for them.
I greatly support your post, except that OpenSSL is not a BSD project (at least not the OpenSSL we know today). Firstly, its code is measurably filthy, but more importantly, it's under an Apache-style license (which is pretty evil compared to a BSD license). OpenSSH IS BSD however.
Right now I'm very ticked off at reality. Every OS I've used has at least some huge flaw which makes me want to run another, which leads to yet another huge flaw. Yes, even NetBSD 2. Said flaws wouldn't even be hard to fix, but it's as if the world just won't allow any one system to be perfect, and toys with the minds of developers and corporations to prevent such a thing happening.
If you want to reinvent the wheel to be object-oriented, go for it. But first write a new language in which to do it. C++ has way too many shortcomings to write the foundations of a whole system in.
The reason I prefer procedural languages to object-oriented ones is because computing itself is procedural. Object orientation feels like a hack. It's a weird way I look at it, yes, but it takes me personally a lot more effort to implement something in an object-oriented way than the same functionality in a procedural way, and the procedural way ends up cleaner and more efficient in the end. That's just me.
Hey, I wouldn't mind seeing a whole revolution in software design for FOSS systems, but the fact of the matter is that it would break a lot of familiarity and compatibility, and if it isn't good in the end (which is very possible) it will have all been for nothing. Not changing anything for decades is a good thing in this respect - there's no way to go 'wrong'.
NetBSD import some of OpenBSD too; they got EHCI interrupt routing from OpenBSD, meaning you can use USB2.0 hubs and the devices beyond them. FreeBSD still needs this (or working EHCI support to begin with; I've never heard a success story). Seen on mailing lists that multiple channel support on emuxki is going to be ported too, or at least looked at.
If you could store more than a few bytes of data in your head at once, you'd notice I never said anything about microkernels or such designs. Nevertheless, the BSDs are much more engineered than Linux is: they see a problem, set a goal for a solution, and work towards it. It gets done over a notably long time but it's done properly. Linux usually has lots of people trying their hand at solving a problem (how many new IO and CPU schedulers are in testing now?) and the most likable solution 'wins' and becomes the default or only option. It's the difference between monarchy and anarchy; something is grown to the top in BSD, or it fights to the top in Linux.
What's your point about not making progress for 20 years? What kind of progress are you after? They've managed to make kernels that perform well on modern systems, userlands that do what modern systems should do (even if they include nvi instead of vim), and documentation that is up-to-date. Technically Linux only provides the former of those, providing no definitive userland and virtually no useful documentation. If Linux itself was a project that involved attention to code quality and consistency, documentation as the code is written, and at least some attempt at release engineering more thorough than 'tar -cf - linux-x.y.z | bzip2 -9 > linux-x.y.z.tar.bz2', it would move slower but more people who need software integrity would take it seriously.
How about the SCSI system being completely changed between two MINOR releases of Linux 2.6, with practically no warning? The change wasn't even backward compatible. That's not the work of developers that believe in release engineering. Major versions are where big changes come in, not minor ones. Especially if said changes have no real reason (Linus just "didn't like it") and are very disruptive (if cdrecord couldn't do plain ATAPI control things would really be bad).
Security record of Linux isn't that great either. It needs one big bad code audit, and those 'hardened' and 'security enhanced' patch sets have to be merged in wherever things don't break standards. And some effort should be made to randomize kernel data; of all FOSS OSs, Linux is the easiest to predict even by nmap.
Ack, I had URL: in the first copy, but it didn't fit in so I had to remove other bits of the sig; now it should work. /. puts way too short a limit on sig length.
It's about time we make a single comment including references to all of those articles AND those that include benches, technical reports, and other useful things. Someone people manage to consider those articles 'not good enough'.
Check that, next boot into a new kernel (which included ACPI support and so on) made the problem go away. Now sound is to hardware maximum, about as good as in Windows (which, you'll admit, continues to have a better sound system than ALSA, even if only because drivers are written by people who know the hardware). Sitting with ALSA for months made me forget what this card could really do.
Are you a Nazi? "Why waste valuable oxygen on Jews and gays when Germans have so much more... consonants in their language". There's a blessing to variety. BSDs have something Linux doesn't; engineering. Momentum isn't everything. I could run for miles and just die on the spot afterwards. Or I could walk forever and get a lot of things done on the way. You think about it and post again with what you think momentum is good for.
Really? That stinks. I have a Realtek that's handling the packets needed to post this, and it's running gloriously on NetBSD 2. But then it's not as simple as that I guess.
My one problem is that auich still has weird calibration so my laptop's sound card plays everything slower than it should be.
Shame really, great system but with some driver bugs. I guess sometimes the simplest possible driver code isn't enough.
Not everyone gets corporate sponsorship based on media hype. Linux gets all the easy breaks. Does this make it a good system? Not really. Just means that it goes in every direction with limited quality.
Good to hear you like it, but why did you expect rubbish? BSDs are known for good documentation. Personally I didn't even know there was a PDF available, but hey that's the way it is.
Even if it doesn't work out for you, it's always a good learning experience.
It's good to see you thwarting trolls, but a post as a grandson of the same post is probably overdoing it :)
Nevertheless you're doing BSD society a good favor and it's appreciated. Are you also the AC posting the same things? Or two separate entities?
You're a good man ulib, we need more like you.
Of course. Didn't you know that all open source software initially starts in the Linux source tree and eventually grows out, but still has major contributions by Linus and all the corporate sponsors?
That's how media hype around "software for Linux" makes it sound, anyway. It's like everything open is immediately for Linux and somehow spawned by the same community.
Anyway, using any GPL-licensed software as a success story for Linux is ridiculous. They share a license so they're the same project?
Simple answer: It doesn't. I don't think one person notice OpenBSD, for instance, scaled badly in their use. It takes literally thousands of iterations before the problems become really apparent (except some really flaky cases, let's not mention those) and getting under that kind of load is pretty rare for what most people need. I hate when people say "my 2-user file server will be much faster under Linux, look at those benchmarks!" when the non-scalable solution is faster for low load than the scalable one (programmers, think how much it takes to fill the first ten spaces in an array, versus the same data into a binary tree).
But nevertheless, I was impressed by Linux 2.6 having consistently fast and scalable results for those tests; while that doesn't impress me enough to want to use it for anything serious, it's nice to know they're wasting time on that instead of getting a code cleanup and audit done. Gotta have someone do 'software engineering' the Wrong Way, right?
Like I keep saying, we shouldn't allow anonymous posting. Maybe people would think before they posted if they had a name by them.
You make no sense and you make no point either. You also manage to pretend important things are trivial, just because they don't directly affect you.
Curl up and die at the soonest possible opportunity, people who know you will be better off.
You possibly misconfigured your X server's refresh rates. These things happen; best bet is to Google for your model and try everything.
I wish to celebrate now: XOrg supports my laptop's video card in-tree with the nv driver (previously it didn't), so I can run NetBSD fine. No acceleration or anything, but then I never used it in the first place.
Here's someone who has NetBSD on a range of machines with no Linux anywhere. A happy, happy admin.