Guest post by “Now that Linux is the most popular free Unix-like operating system, it shouldn’t be a surprise that some projects have begun treating non-Linux operating systems as second-class citizens. This isn’t out of contempt for the BSDs or OpenSolaris, it’s just a matter of limited manpower: if almost all the users of the application have a Linux operating system and if all the core developers are using Linux themselves, it’s difficult to keep supporting other operating systems. But sometimes the choice to leave out support for other operating systems is explicitly made, e.g. when the developers want to implement some innovative features that require functionality that is (at least for now) only available in the Linux kernel.”
It is true that, in general, the BSD developers are a minority on most FOSS projects, however they do take a certain pride of doing things the right way (TM). In general this involves being conservative and letting others go through lengthy, and sometimes useless, discussions: we even have a term for them bikesheds.
Innovation.. well look who’s got ZFS and Dtrace while others are working on beta-quality alternatives.
When the technologies mature and prove themselves they are likely to be adopted by the BSDs, meanwhile we are OK with the linux community doing the beta testing.
…and the ‘them and us’ attitude helps the situation how?
It’s not really “them and us”.
It’s two different approaches, and I think it’s cool that way.
I use Debian (linux kernel) on most machines, but for something like, say, a music-performance computer, a production server, custom embedded device, I’ll be going BSD all the way.
Day-to-day, I want the latest stuff, and I’m fine with working out the kinks myself.
There are two very different thought processes behind both working-on and using BSD/Linuxen, and within the two of them, you get further divisions about what the best method of getting to that goal.
I have much love for the BSDs (though OpenBSD is still somewhat intimidating, and NetBSD will run on a toaster, but not my V880, those are my only complaints. Oh, and the NDA-related modules in FreeBSD, which I can’t see myself needing).
You can see a somewhat lesser form of this division within the Linux distros, and even just in Debian!
Sid = Bleeding edge
Testing = Cutting edge
Stable = Solid, and unmoving (and dull ~_^)
them and us is wrong. But set party putting head in sand over particular items been marked to go end of life and not working on a solution has lead to a lot of the current breakage.
There will always been difference. But the portability divide is only as large as it now due to the need to address problems.
program->hald->kernel and back. Remember context switches are slow.
Hald is legacy of Unix when it did not support dynamic libraries. So there is need for a common dynamic lib to replace hald in some areas. Not a common daemon any more.
dbus even if it was still doing all of Hald. Difference is dbus support per application secuirty. Hald does not.
Lot of tech designs have there time to die. Incompatibility between Linux and BSD has expanded because BSD is simply not ready.
Really it would have been stupid for Linux developers to go out alone and try to design a new common library for all posix platforms if those platforms would not cooperate. Best option is to go on with Linux only solutions and wait for the hell to catch up with BSD and the like.
systemd is based of apple launchd design and expanded on to take advantage of the features Linux kernel offers. Now why is not BSD doing the same.
BSD developers have sat in one place too long that is the problem. Does the BSD service system offer fast startup? Nop. Has linux developers been searching for faster ways to start the system yes. upstart and many other different startup systems to freebsd.
BSD arguement we better wait for the innovation to settle. Fine. We will settle. We will simply not design a system that suits BSD because BSD was not their providing there limits.
Wait for innovation to settle is basically say when you massively have something that now does not suit you stiff. Why innovation is underway design errors can be simply fixed.
To late once the system is design as settled and locked down to be saying lets change features now. That is were BSD is now. Lot of features are locked down based in udev design. Because bsd devd developers were not in the design teams to provide feedback on what they required.
Simple fact waiting for innovation to settle. Is to wait be left in a stuffed up position and become hard to remain compatible. Linux is currently the dominate market share. BSD have to accept this.
Just like when particular Unix’s were the dominate market share that if BSD did not take part in the standard processes they got railroaded. Has not the BSD guys learnt from history they got railroaded with posix standard before linux existed.
Uhmm..how old do you think HAL is? Older than shared libraries? Seriously?? This amazing ignorance pretty much disqualifies you from being taken even remotely serious on this topic.
Becase they dont want systemd? I havent ever heard anyone in BSD complain about systemd not being portable. Heck, BSD and Linux doesnt even use the same starup system now.
Yep but who cares? its a pointless, penis-measurement metric.
That’s not a BSD argument or attitude. Now you’re just making shit up.
Edited 2011-03-12 06:24 UTC
If lack of information is grounds for disqualification you just ruled yourself out.
Hald selection at the time was due to backward compadiblity with Unix’s and embed posix systems at the time that did not have shared libraries.
This baggage has to be discarded. Please go back and read the early debates on Hald and you will find a lot of people quoted a lot of older Unix service based designs as well.
Hald is Legacy of Unix is correct. Its the Legacy of the Unix way of doing things before shared libraries.
Not that it existed when all Unixs lacked shared libraries.
systemd is not portable yes. But Launchd where the Linux design for systemd starts has been around for quite awhile and BSD did not take it up or anything matching its speed either.
Simple fact of the issue BSD system is too old. Work is need asp to address many issues. Then more grounds to be talking about sharing can form.
Tech that does not meet todays requirements cannot just keep on being used.
In fact is not a pointless metric. On devices every second counts. Since that second of runtime is eating into the battery what is the total runtime you have. Slow starting is costly.
Calling it a pointless metric shows how far out of touch you are with todays device requirements. Also people want to flick the power-switch and have a device on almost instantly.
Poor bootup speed is not a fault that can be left.
Then explain why BSD personal have been missing from the shared forums of freedesktop.org. Some of the BSD world have put forwards the argument stated. Not all.
Any putting forward the argument that freedesktop is for Linux or wait for the innovation to settle need sorted out need to be hammered for it.
Edited 2011-03-12 09:07 UTC
What is with the recent flood of people who think they know nearly everything about computing, and can’t get basic facts down…
If lack of information is grounds for disqualification you just ruled yourself out.
Hald selection at the time was due to backward compadiblity with Unix’s and embed posix systems at the time that did not have shared libraries. [/q]
This statement is completely and utterly wrong. HAL is a daemon for two reasons:
1) To have a centralized hub for all hardware information.
2) The most important one: Because hald needs root level privileges in order to access hardware directly to get information. Direct hardware access is a security nightmare if given to normal user programs.
You’re hilarious. Unix has had shared libraries since around SunOS4.
Also, udevd is a daemon. Why is that? I thought “dynamic libraries” was the solution?
hal and udev are daemons because they need escalated privs, not because some obscure Unix variant that doesn’t support shared libraries.
BSD is not designed for those kind of personal devices.
tl;dr
Just write it in your native language. I’ve done translation work, and speak ~4 languages.
Even if I don’t speak your native tongue, I’m sure you can express yourself better in your native tongue, and that I have the skills and/or tools needed to translate it well enough for my own understanding.
As it stands 4 sentences gave me a headache.
I know I don’t do very well when trying to talk tech/computers in Japanese. I’ve tried, and failed pretty spectacularly.
(My comedy style suits Japanese/Japan better, though)
Edited 2011-03-12 06:49 UTC
Uhm, hald is only a couple of years old. And there were working hardware abstraction / detection layers before it (like devd on FreeBSD).
Because it’s not needed? And no one has stepped forward to make a case for it that stood up to even the most basic of questioning.
Bzzt, wrong. I can boot a FreeBSD system in under a minute. Of course, I can also boot a Windows XP system in under a minute. And I can configure a Kubuntu station in such a way that it boots in over 2 minutes.
What is everyone’s obsession with boot times? If you are rebooting your system so often that you notice a 30s savings … then you need to reevaluate your setup. Why are you rebooting that often?
Sure, a system boot should not take 15 minutes. But getting to a login screen in 5s, with everything still initialising in the background, isn’t anything to brag about either.
I’ve yet to see anything actually useful in upstart/systemd, other than making it impossible to know exactly how things are working, or to debug things.
There’s something to be said for deterministic behaviour (where things always start in the same order and always end up in the same state).
There’s something to be said for having a single, human readable *and* human editable text file managing the configuration.
Anyone who says that GRUB2’s configuration layout is better than GRUB1’s menu.lst seriously needs to get their head examined. upstart vs sysvinit (even with all its warts) is the same. systemd isn’t much better (Really? You can’t support a system where /usr is a separate filesystem from /? Really? And you don’t see a problem with that?)
Innovation is good. Code churn for the sake of code churn is not.
Perhaps if Linux devs put more than 30 minutes of thought into something, actually *designed* something before coding it, sub-systems would last more than 12 months, and downstream projects would actually have time to incorporate all the feature before having to start re-writing things for “new shiny sub-system”.
Why would i not turn of my computer when i im not using it. Power costs money and resources why use more than i need.
using the computer that way fast bootup is nice.
you know not every computer is a server that needs to be up 24/7
Ever heard of hibernate? Suspend?
Do you leave it plugged in? If you do, you are wasting resources, and it’s costing you money.
Ever tried “suspend”, “sleep”, or “hibernate”? No need to power off and suffer through a full boot process.
Sure, but there’s also no reason to go through a full shutdown/power off and power on/boot process everytime you leave the keyboard.
Wait, are you saying that Linux does not have reliable suspend/hibernate?
You have a strange definition of ‘fast’, BeOS booted in ~15second (not counting BIOS initialisation) *and* the desktop was responsive even at the beginning.
So in my book, Linux,*BSD and Windows are all booting *slowly*, but it doesn’t matter because now you can resume even desktop..
When I saw HAL for the first time I thought “this is garbage” but, somehow it got included on almost every linux distro. Now they are dumping it.
The same thing I thought about pulseaudio and systemd. Specifically on systemd, I really prefer the old method, even if it is a little slower to boot. At least is easy to locate POF and correct them, as you properly said, in very predictable/repetitive steps.
My point is, developers that want to collaborate on very basic subsystems should do a favor to all us and try to do it right from beginning. They should read about what others are doing/using and what are the shortcomings of their implementations. They should try to respect conventions. And they should break compatibility only for a very good reason. All this have been put apart on these “experiments” to fix problematic subsystems on linux.
As is very known on educated circles, “it is better/less frustrating/time saving/wiser to learn from someone else mistakes/success”.
Let me slap Linuxists in face…..
What can you guys say about your audio system?
Everyone is trying to make something new about it, none have made anything useful.
We BSD users still use good old OSS, and it works just great.
P.S. sorry I can’t find link to graph with all the Audio Mess on Linux
>We BSD users still use good old OSS, and it works just great.
Eww, ‘good old OSS’. It’s a new architecture, multichannel, etc. pp. Just don’t undererstimate it. Alsa needs a plethora of layers, just to get usable at all
you are a retard, and so is that adobe “person” that made that graph..
Linux uses alsa.. it changed from oss.. not a million others..
adobe made a graph showing a billion different interconnects between various THIRD PARTY SOFTWARE LIBRARIES that just so happens to be able to use one another. nobody has EVER configured them in such a way, but even if they did, how is this linux’s fault?
I have personally created 2500 audio libraries that works together, and compiles only on freebsd. WOW LOOK AT THIS MESS! freebsd audio must suck donkey balls…
stop being such a retard and learn a little about what you’re talking about
ALSA + PulseAudio works better than any other system I have ever used.
Really, it’s incredible how well it works, I can plug my Logitech USB Headset and audio will be redirected to this device on the fly, if I unplug said device the audio gets redirected to the master audio device (PC speakers). And everything just works, on the fly, I don’t have to install any drivers or configure anything, it all just works.
Per application volume control is also great and it all just works.
Edited 2011-03-14 00:20 UTC
The BSDs have had launchd for a while .. in the ports tree. It may just be that for most purposes the traditional BSD system just works OK, and it’s not difficult to install something different (like launchd) for those that need it.
FreeBSD has softupdates: some Linux developers were interested but none had the skills to do it. What’s wrong about having different interests?
The BSDs will not disappear because linux evolves faster into the unknown. some developers actually *like* stable APIs. The linux community has to accept this.
Soft updates are kludgy and complex. For this reason, soft updates will be removed in NetBSD 6.0:
http://article.gmane.org/gmane.os.netbsd.announce/399
Instead, NetBSD will provide FFS with journaling, very much like ext3.
Well, FUD isn’t the answer. Softupdates aren’t that complex and it’s easy to get them checked with background fsck. In FBSD 9 there are even journaling softupdates, the best of both worlds.
But then, do you really consider those ext-hacks an alternative? Once in a while they have to be fcked due steady inconsistencies. This I call progress … it’s a hack as I said. And if FUD is really your base, just use gjournal in FreeBSD.
Wow you went from defending FreeBSD to just trolling with this comment.
Lack of knowledge on your side suggests a troll on the other side. As I said, FUD … in case you don’t know: fear, fear, uncertainty and doubt. I don’t know where do you see the behaviour of a Troll, but ignorance is for some people obviously an argument.
But I have something for you to support your stereotyped thinking: most Linux users don’t have a clue about *BSD. That’s a fact, why should they bother? But most BSD users really do know Linux, because usually they started with Linux. However the migration from Windows to *BSD is a rare event.
When you assume something about me with out knowing then you look like a bigger troll. I have used FreeBSD in the past 3.x to 4.7 ish, I ran an image and mp3 http/ftp server on FreeBSD in 2001-2004 for friends and DJ’s in my area. I loved using FreeBSD back then I still dabble in it to this day. I also setup NetBSD, OpenBSD and other OS’es because I love to learn and play with and make mistakes dabbling in hobbyist computing. Why troll on Linux by way of the EXT-X’s is lame when the EXT series of file-systems are the most stable and bug free out of the any file system native to the Linux kernel?
Oliver, this article counters your arguments pretty well:
https://lwn.net/Articles/339337/
Even within BSD projects, it is generally acknowledged that softupdates are complex (and again, NetBSD dropped it for that reason, too hard to debug).
Maybe you could make an actual case what is wrong with journaling as in ext3? It is a simple and elegant solution to a problem: asynchronous metadata writes make a filesystem faster, but are also more dangerous. So, you keep a log of updates synchronously, that you can replay in case of a crash. It is really a sane solution that is also used by nearly all databases.
A viable alternative are log filesystems or COW filesystems, but they are a new generation.
Hrm, just because NetBSD devs can’t figure out how to make SU work correctly, doesn’t make it a kludge.
Afterall, the FreeBSD devs have added journalling to the SoftUpdates code without affecting compatibility with UFS2. Thus, you can now mount a UFS2 filesystem without SU, with SU, or with SU+J.
For those who don’t want to muddle around with SU+J, there’s always GEOM Journal that adds journalling to any block device (aka, journalling for every filesystem).
I agree, I wasn’t trying to say that there is a them and us situation, more that people in both camps are often too quick to jump to a parochial response to issues like this.
Maybe if there was less of this attitude there would be more willingness from devs in both camps to jump the fence now and again and we wouldn’t be forced to choose between “portability and innovation” quite so often.
We try different approaches and eventually we learn from each other. In the BSDs we call it “crosspolination”. Adopting a restrictive license like the GPL doesn’t help though.
+1
Personally I could never understand the BSD licence. It’s asking for abuse. Maybe that is why this devil-mascot is so popular in BSD?
Uh? Given that both ZFS and Dtrace were created by Sun, I wouldn’t count those as *BSD innovation..
Plus Sun used a license volontarily incompatible with the GPL, so *of course* the *BSD can use them but Linux can’t..
I think you missed the point.
There is a video online with Lennart Poettering speech where he said these things (I didn’t read the article).
Do you program ?
His talk was about systemd.
When you code, supporting different operating systems takes a lot of extra time and code.
Obviously he likes using Linux so that is why he choose to code for just that operating system.
He also choose to use the system specific API’s because it makes the code shorter, simpler and takes less time to write because things are handled by the kernel.
That was basically all he said. He didn’t say BSDs are bad or something like that.
That would be this video if I remember correctly:
http://video.fosdem.org/2011/maintracks/systemd.xvid.avi
Not that I ‘approve’, I think system specific API’s should just be used for improving performance, etc.
Edited 2011-03-12 22:08 UTC
He is the enfant terrible even in the kernel community, similar to this libc-guy Ulrich Drepper.
“What I actually suggested in that interview was not so much that the BSDs should adopt the Linux APIs, but instead that people should just forget about the BSDs. Full stop.”
http://lwn.net/Articles/430699/
So that is exactly the opinion of this guy.
“And on Linux, if we ever want to catch up with MacOS or Windows we probably even need a much faster pace. ”
A fine answer to this bewildering saying …
“Brute forcing software design through several revisions of code or several forked projects detracts from the collaboration effort and I would question whether or not true innovation is being made, or if it’s just code churn.
How often does OSX, Windows, etc dramatically change system interfaces or how components interact with one another, and how do end-users receive the changes? I know based on my experience that people absolutely loathed the Windows 2000 -> XP and XP -> Vista/7 transitions and those only happened every 3-5 or so years. Freedesktop and other associated projects (Gnome, KDE, LXDE, XFCE) pulls this stuff every release, which is either biannually or annually!”
http://lwn.net/Articles/432914/
And he is entirely entitled to that opinion. Given that Linux runs on many very many platforms these days (at least as many as NetBSD), includes a lot of security technology (no SEBSD in OpenBSD), and has also shown to be performant (the days where FreeBSD was the king of the hill have long passed by), it isn’t hard to come to the conclusion that BSD (outside OS X) is obsolete.
As much as I dislike the GPL, even most device vendors do not seem to have problems with using Linux/busybox/…, despite their use of the GPL. They have proven that the license advantage of BSD isn’t as big as we once assumed.
There is something wrong: Linux doesn’t actually run on all those platforms, it is possible in theory. NetBSD is ready to run on those, just download the proper image for the chosen architecture.
FYI: I was [email protected]. NetBSD ‘supported’ platforms for which distributions were built, but where nobody checked wether they would actually boot. Simply because there was no one on the relevant port lists with working hardware. In that manner the list is somewhat deceptive. And even if a machine does boot, it is pretty worthless without drivers. And Linux has an edge on drivers on nearly every widely used platform (ARM, i386, x86_64, and PPC).
Edited 2011-03-13 19:59 UTC
OS X alone will guarantee that FreeBSD stays around.
How much does Apple still take from FreeBSD? I don’t think Darwin would be threatened in anyway if FreeBSD disappeared from the map tomorrow.
Uhm… Dtrace and ZFS didn’t come from the BSD community.
You could argue that SunOS was indeed a fork of BSD UNIX.
But the authorship doesn’t really matter too much anyways in free software … it’s innovation and now that everyone discards OpenSolaris as being dead it’s pretty much BSD technology.
You could state the fact, with no controversy or reasonable argument, that the people who wrote ZFS and Dtrace were employed by SUN Microsystems, rather than a member or former members of any previous or existing *BSD community.
Porting code from one *Nix platform to another does not count as innovation by any sane person’s standards.
And being conservative isn’t necessarily a bad thing – Linux does go through an unfortunate amount of technology churn before getting things right.
But by dismissing that process as ‘bikeshedding’, the BSD’s also remove themselves from any influence over the outcome. UDev and friends are the result of Linux-based developers actually producing something, and if that ‘something’ doesn’t suit BSD – that’s what happens when you don’t play a part in the design process.
Well if you are the minority in a community it doesn’t make much sense to participate in a discussion about something you consider a broken design from the start.
OTOH, most standards come from processes where you don’t have a voice or a vote anyways. It’s ultimately the use the API gets that defines if the proposed standard is successful or not.
Please learn how freedesktop is meant to work. Remember everything at freedesktop is draft standards. So is perfectly permit-table to have 2 standards doing exactly the same thing.
If the design is considered by you as broken is your job to put forward another standard. Freedesktop will take standards written by anyone.
Issue do nothing is not a defense. Process to voice dispute exists. If its not resolvable the solution is 2 standards and let the best win or both lose and a new one made from both. This has happened before.
There is no such thing as no voice in the freedeskop model. There are only people who choose not to speakup. They always have and had voice.
devicekit process is a freedesktop process. Basically show me the draft standards were BSD’s guys have done stuff. No draft standards exactly that is why they have no voice. Because they have not spoken up properly. Down right speakup people working at Freedeskop are not mind readers and take no messages as approval to keep going forwards.
Freedesktop design is designed to protect minorities if they speak-up with standards.
The BSDs didn’t participate in the development of udev … because they already had working hardware frameworks and device notification frameworks … with a single human readable config file (http://man.freebsd.org/devd.conf).
Why should BSD devs participate in the development of an inferior product, when they already have a working product?
Solaris. Ok wait. BSD did not Innovate either of those. They copied.
Dtrace is out gunned by http://lttng.org/ lttng was designed to debug and track down performance issues so has full timing support. SystemTap and Kernel markers are based off the idea of Dtrace. But big thing they also extend on what Dtrace was able todo. There is no point of just coping a debuging system build for a different OS if you don’t extend functionally to support the differences of your OS.
http://zfsonlinux.org/ Not like ZFS is a issue to Linux if Linux users need it. There is a snage with ZFS you cannot do a transparent upgrade from ext file-systems to it. Btrfs does support in place upgrades from ext file-systems to Btrfs. Yes less file comping equals better. And that is not the only issue that ZFS has with trying to convert a large scale install. ZFS don’t always out perform ext filesystems either. Btrfs goal is to give the features of ZFS without the draw backs of ZFS in performance and migration.
Edited 2011-03-14 02:32 UTC
That might be the goal, but they are still a good 5 years away from even reaching feature parity with ZFSv28 (the last fully open-source version, as Oracle has not yet released the code to ZFSv31, which is in Solaris 11).
Btrfs, as of today, is really only a CoW filesystem with in-line snapshots and built-in RAID0/RAID1 support. There’s no RAID5+ support, no dedupe support, no fsck, no encryption, no support for read/write cache devices, etc, etc, etc.
Btrfs today is like ZFSv6 (except even that had working RAID6 support) from several years ago.
My problem with the idea of supporting Linux over other free OSes is that Linux is also missing features that I like in my present OS.
And it is no good saying they can be added to Linux, because if features like DataTypes for example are not there from day one then most software for Linux can not use it.
Linux is a technically excellent (or at least very good) system, and it serves as a strong contender against proprietary platforms.
Why should Linux folks care about BSD exactly? This is not meant to sound offensive (though it probably will set off some fires), but the question stands: what’s the point of more than one perfectly good system, except the usual argument of “choice is good, competition is good, what if I don’t *like* Linux, and what if Linux suddenly dies overnight”?
More importantly, why should we somehow let 40 years old o.s. design principles dictate how Linux should be designed?
Somebody enlighten me. I’m looking for a well-grounded argument for why we must be paranoid about not stepping on anybody’s toes and why we should increase the burden of development and hinder innovation by carrying all that legacy baggage.
Disclaimer: I am not a sysadmin and I am not a kernel developer.
My suggestion would be for you to look into all of the research done on OS design, and figure out a better one(including the kernel, system calls, and a number of new apis).
Next, figure out how much work it would/will be to get a booting system that’s somewhat usable. Next, you’ll need native-running development tools.
Now you can start to attract hobbyists.
It would be quite a bit of fun, but there’s quite a bit of history behind why Unix is the basis for most operating systems.
Do I have a basic design concept written for an OS?
Yup.
Do I think it’ll be through stage 1 in the next year?
Nope. I have too much to learn for that level to be reached.
Doesn’t it make you sick how linux installs all config files in /etc/ ?
Or doesn’t it make you crazy when you don’t know where to look for files that were installed by some package?
for example is it in /usr/bin or /usr/local/bin or /opt/bin or god know where?
dpkg -L {packagename}
Nah, it’s “pacman -Ql packagename”.
Sorry, couldn’t resist. It does bring up what a nightmare packaging is like in Linux land with a bazillion different formats though. *BSD’s got that one partially right, their package formats are pretty similar, but binary compatibility between them is no better than with different versions of Linux. They’re actually worse in that area, so that bit evens out I suppose.
edit: Corrected some very bad grammar (even native speakers of English can seriously fsck it up sometimes).
Edited 2011-03-12 14:11 UTC
>but binary compatibility between them is no better than with different versions of Linux.
You’re talking of *BSD forks, those are not distros. But apart from that OpenBSD is able to run Linux and FreeBSD binaries, a recent FreeBSD is able to run binaries down to version 5. I do call this way better than any practice in Linux.
Actually there is generally very good binary compatibility between the different GNU/Linux distributions. This is evident by the large number of binary-only releases for GNU/Linux distributions (particularly games and scientific applications). You can usually use a binary from one distribution on another distribution; but occasionally it requires the creation of some symlinks or other ugly hacks.
But why people would choose to rely on binary packages beats me.
>This is evident by the large number of binary-only releases for GNU/Linux distributions (particularly games and scientific applications).
This is no evidence at all, because software like Mathematica, Opera, etc.pp, avoids to use lots of dependencies or they’re coming with static libraries and so on. And in practice there are just a couple of distro, that are usable at all in this respect.
And the rest you get fresh via the package manager. So, I don’t see the advantage of maintaining backward compatibility with very old Linux versions.
Doesn’t it make you sick how two different distros can use completely different filesystem layouts … and still be FHS-compliant? That’s what boggles my mind.
Going from a RedHat to a Debian to an Arch system always leaves my head spinning, trying to figure out what is installed where. And they’re all FHS-compliant.
Give me a FreeBSD system anyday:
* / is where the OS lives, all the stuff needed to boot to single-user mode
* /usr is where the OS lives, all the stuff needed to run in multi-iser mode
* /usr/local is where you install stuff
* /etc is for the OS configuration files
* /usr/local/etc is for the user-installed configuration files
So nice and clean, and simple, and nicely layered. Considering how often Linux devs scream about “rampant layering violations”, you’d think they’d have a nice layered filesystem. But now, even the systemd dev doesn’t believe you should separate / and /usr.
Le Sigh! 
Ok. Seriously. systemd does not work if you have a separate /usr partition? I mean, really, you gotta be making this up. Right?
If only.
http://www.daemonforums.org/showthread.php?t=5772
That’s hilarious although to be fair it’s not his fault that Linux is borked when it comes to separate /usr,
And more info on it:
http://lwn.net/Articles/430887/
I know right. Why does people bother with anything but Windows?
Do folks here still bother to analyze comments or just prefer to mod down anything that questions the majority’s view?
I think my comment made it pretty clear that we need something to stand up to Windows/Mac as an open platform. But the corollary to that was “Why do we need more than one of those [open platforms] to do so?”
But then I got foolish and asked “Why should Linux necessarily try to bend over to please the BSDs?”
The replies along the lines of “this is how O.S.es have — and should always — been done” do not answer my question; the reason why we’re having this situation in the first place is because the Linux ecosystem develops some of its things in its own corner or tries to get rid of legacy stuff.
Is asking the hard questions not allowed here anymore?
I certainly hope you werent mod’ed down.
Why should Windows developers like Adobe bend over and please Linux users? To paraphrase Poettinger, these developers are already writing code for the most popular OS on the planet. Why should they bother?
It’s exactly the same argument.
Edited 2011-03-13 15:44 UTC
Nobody bent over. Adobe had its own reasons to support Linux. They didn’t do it for some ideological reasons. The effort at first was minimal but with recent developments in mobile space, they’re actually putting some weight behind Linux support because it makes economical sense for THEM.
In the areas where Linux is thriving, it isn’t thriving because it’s ideologically better. It is thriving by providing ample technical and economical benefits. It’s not about having whatever arguments.
The market and industry have changed to accommodate Linux because it was beneficial and not doing so meant competitive disadvantage but even then it didn’t happen by already established products or projects graciously helping adoption. Linux fought its way up.
So, there is no such argument that you’re referring to. There never has been and never will be. If *BSD want to be considered as primary target platform for system and desktop developments, it should generate its own weight and be able to justify and, basically, force the effort. Whining or arguing about it simply doesn’t matter.
Hald project accounted over 7 years ago it was going to be deprecated.
This is case of Moses and the flood. Linux guys have built a boat out of the historic valley and the BSD guys have been saying the flood will not come.
Also some BSD people have said freedesktop.org is about Linux. Sorry its about everyone. But if you don’t turn up and build your boats for the future its not freedesktop.org problem when you get left behind.
Ubuntu and Gnome is a another lot of stupidity. Ubuntu does not work with upstream well. About as bad as Gnome not working with their upstream standards body.
Disfunction and disfunction. No point complaining about being left behind when its not really invention but dealing with what was announced 7 years ago and should have come as no surprise.
Lack of innovation by particular parties have caused the lack of portability issue.
If you’re going to troll, do it after some proofreading.
No one will be offended if they can’t understand you.
it might be that he didn’t used proper english, god knows i don’t. english is my fourth language
but what he said is true.
if you don’t believe him or me, read this from the person who started gnome/kde/canonical fiasco
http://blogs.gnome.org/bolsh/2011/03/11/lessons-learned/
and this from ex-canonical employee
http://aseigo.blogspot.com/2011/03/collaborations-demise.html?showC…
http://aseigo.blogspot.com/2011/03/collaborations-demise.html?showC…
and this is exactly same nonsense, linux developers should not develop for linux since it won’t work on os they don’t use
Your English is quite good, actually.
I honestly couldn’t read all of anything I’ve seen by that user. Just headache inducing if I try to make sense of it.
If I just pretended it was surrealist poetry, it’s awesome, but that doesn’t help the discussion in any way.
TheGZeas, from your tech comments you seem like a smart guy.
…why are you being such a prick though? It’s an international site, open for all to comment. Most users don’t have english as their first language(myself included).
I love OSAlert and it’s diverse culture. Apologies for not contributing meaningfully to the discussion.
That’s why I’m suggesting that people just post in a language they _do_ understand.
Also, what little I was able to discern from this poster was either trolling or utter stupidity, so I don’t care about their feelings, really.
What a terrible idea.
Gibberish few people can understand, or non-gibberish few people can understand, but has the chance of being translated?
Which of these sounds worse?
Seriously, google translate can do better than some of these people…
I put quite a bit of effort in to anything I write, so that what I mean is clear. If I’m posting something in, say, Japanese, I take a good deal of time to make sure I’m using words that pertain to the subject at hand, and that my grammar isn’t borked beyond understanding.
It’s plain rude to expect people to sift through garbage to find what you mean. I wouldn’t do it to anyone else, nor will I tolerate it.
If one refuses to post anything intelligible, be it in English or otherwise, then they shouldn’t post anything whatsoever.
Hell, have someone translate it for you.
If someone came up to you and started angrily shouting gibberish at you, you’d tell them to fuck off.
If they were _polite_ and were speaking gibberish, and you spoke 4 languages, you’d ask what their first language was, so you’d have a chance of understanding what they were trying to get across.
This humanoid is _trolling_ and making zero valid points, and I’m still offering to hear them out.
actually, in 99.99% you only need to go a bit away and reread the statement. translating always requires interaction
obviously you never saw how google translates my native language into english. compared to that translation, i’m willing to bet that OP can qualify for perfect english spoken person
there are first class citizens in google translate and there are second class, my country is in second
you mean you put a lot in grammar. there is still yet to be seen one sensible post that makes solid point.
lol, you should really listen to your self more
he didn’t troll (rough translation of his post at the end), in fact if i compare his and your statements. he did make a valid point, while you were successfully spouting nonsense in perfect english. truth be told, i had to bother my self actually trying to understand his statement, while yours cannot be understood even after rereading them. (taking OP thread only in this account)
p.s. note to Thom /sorry, couldn’t resist being sarcastic/
if you ever decide to post articles in your native language, at least make one last warning in english so i can remove my bookmark
p.p.s. here is rough translation of post that OP made (sorry, if i change some things and even more sorry if it doesn’t fit your english criteria, i can make same post in my language and use google translate, but i doubt you could make sense of anything, i can’t)
bsd doesn’t invent, they don’t even try to include something unless it is set in stone
linux on the other hand is all about innovation and new things
beside both being free (and even there they differ as one propose freedom for user and ones for software), they don’t have one thing in common as they work on completely different development approach. sure, it started with posix and so on… trouble is, paths they choose how to achieve their goal are completely different
now, if bsd has not enough users… maybe they should rethink their approach, either focus on servers where stability is preferred or adapt to the fact that desktops have new needs every day and stagnation doesn’t really work. it is a pure nonsense to demand world stagnation just because it doesn’t fit your view.
in fact… isn’t this just the same as natural selection where one species simply goes dodo? and more adaptable species survives? news for you, standing your ground while demanding support and recognition is not adapting. especially if only mediocre number of users actually uses your product
also, OP mentioned how this relates to gnome/kde/canonical fiasco from few days. yes, it is exactly like that. goals and paths differ and there is no right or wrong side simply because all sides have its vision.
now, my 2 cents and my personal opinion, foss developers should really look at the nature. natural selection is one way to select the right one. more competition with weeding out the weakest links could only bring faster development and less fragmentation
Edited 2011-03-12 15:54 UTC
Ok, I think I understood _most_ of that.
It’s not about “meeting my criteria” it’s “making sense”.
How can I/we be expected to reply with accuracy to what you’re trying to say, when what you’re _actually_ saying makes no logical sense?
It’s not fair to anyone if I’m guessing as to your exact meaning. If it was in your mother tongue, and google spat out garbage, I could find someone who actually spoke both English and <foo>. Wouldn’t be that hard. You should probably do that.
Again, I’m not saying “SPEAK ENGLISH RIGHT OR DIE!!!” I’m saying “English is a shite language, and so you’re not speaking it properly. As such I cannot say I got any point you may have been making with any veracity”.
As an aside, I’m honestly curious what you intended by “Bio? No, thanks. Not a vegetarian”.
I’m sure the joke’s pretty funny in whatever language in which it started, but in English it doesn’t really work.
just my 2 cents
if you don’t understand, do not respond and do not go into all-out offensive. if you need, mark it down and it will disappear. does every point need your response?
his point actually made sense (to me at least, but i had to try really, really hard to understand, but it made sense), just not to you.
and just curious, was my somewhat rough translation unclear as well?
haha, and here… after you can’t even bother to try to think what original poster said. you propose this?
yes, i can imagine my self flying over the internet. “anyone speaking foo? someone wrote a post and i want to know what he said so i can respond”. yeah, right. now imagine whole internet working like that
reminds me on this one
http://xkcd.com/386/
and yes, bio is really a local joke here. google translated it exactly like it is written. in fact, knowing the joke, translation sounds even funnier to me because it is downright stupid. it was never meant to be funny to others. now imagine posts like that.
btw. i’m a strong believer in right to express your self and equality.
“i disapprove of what you say, but i will defend to the death your right to say it” from evelyn beatrice hall is probably my favorite rule.
Edited 2011-03-12 21:03 UTC
It’s beyond me not understanding, it’s what they’re saying being intensely unclear.
Especially since this is _a troll_.
In my experience trolls like to be unclear. It gives them weasel-room.
From post one this humanoid was stupid, hostile and aggressive. If you’re going to be a moronic jerk, at least be clear about what you’re being wrong about.
“flying over the Internet”? What?
I know people in most regions of Europe, and people that speak the Asian languages I do not.
It’s not hard for me to get an accurate translation of something, for free, in an hour or two. At most a day.
Your circle of friends is more narrow? That’s common.
I’d ask for the portions of the posts that made no sense to be explained, but in my experience on this site when you do that the gibberish gets worse, and comes with insults.
I’m left with replying to every theoretically possible meaning of a given paragraph in a bunch of if/then statements. If I’m going to waste my time, I’d rather waste it learning a new language.
Edited 2011-03-12 21:38 UTC
“flying all over internet”
roughly translated… now imagine your self looking for translation of every post since every post is written in different language. maybe you think this is sane. 99.99% of people don’t
in your inexperience is more appropriate term. if you think someone is troll… mark their comment down and most importantly DO NOT RESPOND!!
could you believe that some people actually like it like that? me personally, i value privacy and few good real life friends more than 10000 casuals. i have no social network interaction. i never post my personal info… why? my life is full more than enough as it is. even this forum posting is just casual break while working. in my spare time, i wouldn’t even think to look at computer when i want to converse with people. hell, i think i’d rather have http://en.wikipedia.org/wiki/Tamagotchi
and…
expecting insults for explanation? hopefully not from me, i doubt i gave that impression.
the only people that get me pissed off are people who attack others by using insults rather than disputing their argument. but i think i explained clearly what my belief is
Also, I’m sick of English being treated as the global lingua franca.
I grew up with it, and every language I’ve studied since has made more sense (save what little Spanish I’ve picked up, but that’s all informal).
I wonder what the next one will be? we (humanity) will find out in a few hundred years. The status quo cannot last.
Try the German or the French grammar and see if it makes more sense than English to you
Although my personal favorite in terms of grammar is Japanese, English is not bad either. It has those silly irregular verbs, granted, but that’s about it. No feminine and masculine objects, articles are not randomly modified depending on their position in the sentence and the time of the day, modal verbs are used properly instead of creating a different way to conjugate verbs for each use case people have come up with, transcribing oral English to text is fairly straightforward…
Edited 2011-03-12 15:46 UTC
English makes perfect sense if you are English, much like any other languages I would expect.
There are oddities but that is because we been invaded by the Romans, the Saxons, the Vikings, The Normans etc etc.
Also many people don’t speak English, they Speak “American English” IMO. Words are pronounced wrong quite often because “it make more sense” … what they don’t get is that … this isn’t the point, it is a language oddity.
I suggest you be a little less arrogant about someone else’s English skills, because I’ll verbally slap you around in 6 different languages, friend.
That’s an impressive skill.
Your comment changes nothing, because, as I said, I neither post nor say serious things in another language unless I know what I’m saying, or at least don’t expect anyone to take me seriously should I do so.
I could post some pretty ludicrous stuff in German, but that wouldn’t serve any purpose. Zero communication would take place.
However, say I didn’t speak English. I could post in Japanese, and a few people would be able to understand me. That’s better than zero, which is the number that would have a chance if I tried to rely on a dictionary and High School English.
Edited 2011-03-12 10:33 UTC
It’d be a welcome refreshing change of pace.
That was very constructive, and everyone learned so much.
Thanks Thom.
-_-
Portability and innovation are not necessarily mutually exclusive. Sometimes they are, sometimes they’re not.
Is BSD lagging in the desktop space? From the average users point of view, definitely. On the other hand, BSD kicks Linux ass in some other areas. OpenBSD, for example, leaves Linux in the dust when it comes to being a network infrastructure platform (firewal, router etc).
Portability also goes both ways. There are applications in BSD that is not easily portable to Linux (OpenBGPD, for example) due to them taking advantage of BSD-specific features. These are usually not on the desktop though so I guess that’s why they dont get as much tabloid press.
Poettering’s talk about ignoring POSIX and only coding for Linux is nonsense though.
5 years ago what you said is kinda true. Lead OS in embedded for routers is Linux these days. Yes firewalling in Linux has passed BSD. Major reason for some devices still choosing BSD is lack of requirement to release source code.
Basically time BSD guys wake up innovation is required soon. You have sat still fair to long and getting passed.
So when it’s about Linux market-share is a good meaurement of quality and innovation. Lets apply that to Windows vs Linux, shall we?
Sorry, but you just have no clue on this topic.
Edited 2011-03-12 06:30 UTC
How about you give us one. What does OpenBSD do that Linux doesn’t?
pf and a proper, simply and elegant packet filtering DSL. State-full packet filtering that can be sync’ed between multiple, failover firewalls.
bgp that integrates seamlessly into the packet filter and understands VRRP/CARP states.
CARP
IPSEC implementation that can be understood and configured by mere mortals.
Great man pages.
I’d go on but I have other things to do today.
oh yah, and OpenSSH. you know, the dominant SSH implementation but since it’s actually portable I guess it doesnt count in this context.
[/q]
I asked for things that OpenBSD provides that Linux doesn’t.
That’s mostly a list of features that Linux already provides. Provides with various opensource projects, kernel modules, or VM appliances like Vyatta. In fact, Vyatta makes it easier to setup network routing and services than it is in OpenBSD (although that’s a matter of exposure and experience).
The rest are nit-picky rhetoric. pf? I like pf. I liked it when it was ipfilter, and OpenBSD decided to make their own. I prefer it over Linux firewalling, but they both provide Layer 4 firewalling.
The problem is, Layer 4 isn’t enough anymore. Attack vectors are mostly Layer 7 now, SQL injections and the like. OpenBSD has nothing for that, save for what you can run on top of it like ModSecurity (which you can do on Linux, and better, because Linux almost always trounces OpenBSD in performance and scalability).
Decent man pages? That’s a matter of opinion. And in the age of Google and project communities, man pages are nice, but not a viable differentiator. Certainly if that’s what a project makes it stand on, it doesn’t have much to stand on.
IPSEC by mere mortals? I think you’re suffering from what’s known as “the curse of knowledge”. Most of us find it incredibly difficult when we understand something intimately to comprehend how someone else could find it difficult. Case in point: http://www.symantec.com/connect/articles/zero-ipsec-4-minutes
That doesn’t look “mere mortal”-ish to me, but I don’t do OpenBSD IPSEC configuration. I understand Cisco CLI (both IOS and NX-OS) like the back of my hand, but it drives new comers insane (as it did me).
It’s portable and ubiquitous, pretty much the only game in town.
I’m all for variety, but the BSDs haven’t kept up. They have niche uses (and incredibly good ones) and I don’t think they should be abandoned (I use FreeNAS for instance, although its ZFS implementation was a joke).
Data centers are moving towards virtualization because power and cooling are running out. iSCSI, FCoE, and 10 Gbit are where storage is moving to. Security is being breached with SQL injection and other Layer 7 attacks.
The BSDs aren’t leading in any of those areas. They’re following, or just outright deficient.
Linux is following MacOS and Windows on the desktop. It’s the copycat of Solaris etc. on the server. Leading? Linux is a UNIX mock-up, hopping around between technologies because they don’t have a clue about long-term goals. You know, software design instead of mere hacks, that just work.
Can you make a point without trolling Linux?
Yes Linux is a Unix clone a clean room interpretation of UNIX if you will. Its goals is whatever one can think of and more!
I remember back in the 90’s when the UNIX old guard used to laugh at Linux and say it matters not because it does not run any big servers or super computers, it does not scale beyond 4 cpus its only usable on x86 and mk68 cpus etc…
now Linux leads the way in terms super computer use, its more portable than NetBSD, it scales up and beyond 4,096 cpus. Now if I can go back to thoser usergroups I used to read and say how do you like Linux now? maybe they would be speechless!
Sounds just like how many Linux users treat BSD these days…
Why would i as a *BSD user care about what the linux users think/say?
Hell i even have a linux install aswell. What ever gets the job done and do it in the best posible way. I don’t care if it’s *BSD or Linux.
And I gave you a number of them.
Uhm, not in my book.
Welcome to 5 years ago. Perhaps you should keep up to date with what’s going on in OpenBSD instead of blindly assume they’re lagging.
No, of course it isn’t but it’s still important. And we’re talking about networking and firewalls here, not web hosting.
Right and what do you run on virtualization? Operating systems. like BSD. and Linux. and Windows.
Virtualization is great because the host supports all these technologies like iSCSI, FCoE etc so that the guests doesn’t have to worry about it.
And what? Linux is? KVM is awesome but not exactly a leader.
Edited 2011-03-13 08:08 UTC
[/q]
You were asked for what bsd does that Linux does not. I see a BSD person who has not been following what Linux can do. Yes all those features do exist for Linux.
CARP and VRRP are done differently as user-space plugins into firewall. Particular with VRRP that has a patent issue this avoids the GPL license of the linux kernel. http://off.net/~jme/vrrpd/ and http://www.ucarp.org. Technically VRRP support in kernel could come back and legally effect you. Also there are other solutions linux can use as well. http://www.linuxvirtualserver.org/ That BSD has no support for and was pre both VRRP and CARP. So a lot of areas running pure Linux have had no reason to implement either of VRRP or CARP since LVS covers all the networking layers and VRRP and CARP dont.
IPSEC is not that hard on linux.
Functionally in Linux is done different. Since netfilter on Linux is a hybred kernel design. Part userspace part kernel space. And it really makes bugger all difference for performance if part is userspace or kernel space. Makes a differences for secuirty and issue handling. If carp or VRRP goes wrong in Linux you can kill the services and restart them no stuff up. Same cannot be said of openbsd implementation.
And to pfsync BSD golden card. Is not so golden. conntrack-tools. Another userspace solution addon gives pfsync state between servers. What is the advantage of this. conntrack-tools and kernel don’t have to match. So all machines can be on the same conntrack-tools even if they are running different kernels. So state syncing is not kernel dependent for operation. Also in broken sync transfers conntrack is simple to terminate. Of course most BSD users have never enabled conntrack on Linux to find that lot of the missing firewall features magically appear once conntrack is enabled with is userspace parts. Userspace parts are responsible for the sync.
Simple fact your claims were valid 5 years ago. Not valid now. Linux has implemented the features differently that is all. The problem is taking the BSD implementation way and looking at Linux the same way and saying features are missing when they are not. Linux kernel + a set of userspace parts is the Linux firewall system. Where BSD is more contained to the kernel.
Even bgp support also can integrate seamlessly into Linux firewall.
Actually, since i’ve been using BSD, Linux and Windows daily in a professional capacity for more than 10 years I know exactlly what they can and can’t do. The point is, and I wasn’t really clear on that, is that I get all the features I need for a solid firewall out-of-the-box with OpenBSD. I don’t have to go hunting for solution X and somehow make that work with solution Y. It’s all there and it’s well integreated from the moment installation is done.
Perhaps Linux has some or all of these things now but there was a time when it didn’t.
Well, that’s why we have CARP and admitedly I forgot that it has been ported to Linux. Still, it originated on BSD while Linux was stick with patent-encumbered VRRP. It’s something that BSD had while Linux didn’t.
Btw, LVS is using VRRP.
Perhaps not but it’s harder. Of course, you can just use openvpn instead that works on many platforms proving that innovation is not depending on making stuff Linux-only.
Wow, that’s almost like who you guys do it, only vice-versa.
“Linux is doing it in userspace and thus it’s much better. There’s 5 bilion different userspace ways of doing this and that’s much better because…uhm…that’s how Linux do it”.
“I see a Linux person who has not been following what BSD can do.”
CARP on openbsd is an interface. If it fuscks up you destroy the interface and re-create it.
No, it’s instead depending on application version. Btw, you can run pfsync between kernels of different versions.
Please give me an example where Quagga/Zimbra/XORP understand VRRP state.
At the end of the day, it’s funny how so many Linux fans just can not admit that there are things that Linux isn’t the best choice for.
I have used Linux BSD Solaris and AIX over the last 20 plus years.
Technically no LVS optionally can use VRRP or CARP. The protocal LVS uses by default is IPVS(IP Virtual Server). IPVS is Linux kernel space. No need for userspace parts of VRRP or CARP for a proper functional LVS.
VRRP is Cisco system implementation to save the same problem as IPVS and CARP. Yes we kind have gone overboard with protocols. I think you mixed up who created each protocol.
IPVS is not patented effected and predates CARP. It was the BSD guys here who went and revented the wheel. Think that the first attempt to port IPVS to FreeBSD was in 2005 but support has been left suspect.
Basically you don’t know the protocals here. Linux can support all three.
CARP,VRRP and IPVS. You only have 2 on BSD at the moment that are stable. Yes technically for perfectly safe operation IPVS in Linux is in the wrong location.
I like the Linux-only comment. There are IPSEC setups that work solaris to Linux but don’t work solaris to bsd or Linux to bsd. Linux does have a broader coverage to other Unix implementations of IPSEC than BSD and yes its is possible to use modes that are not BSD compatible just as equally as ones that are compatible. This more shows how far behind BSD is.
There are reasons for doing particular things in userspace. Items that could require different version to kernel to be compatible. Like server to server messaging. Error safe guards. Most cases each feature of the firewall that appears missing only has 1 way in userspace todo it.
Funny part not always. CARP table memory size in Linux can be capped. So issues from runaway don’t bring nightmares. If there is a implementation error CARP is kernel bounded so restarting the interface is not enough. You need to replace it. User-space form of CARP and other protocols like it are the most secure stable way due to the restrictions that can be placed on userspace that cannot be effectively placed in kernel space.
What happens if someone typos the implementation. It has happened before with BSD all you have todo is go back through the bugs. Application version is as stable as same kernel to same kernel. It was done they way it was in Linux because Linux developer doing this learnt from BSD mistake.
State tracking on are talking for those in connection tracking. Yes it was done by userspace addons for the means to support new protocols without having to rebuild kernel.
Iptables/netfilter can understand vrrp state with the connection tracking features load. Yes firewall level it works. Of course appears missing if you did not setup and start connection tracking. Advantage less overhead when you don’t need that functionality. Ie desktop and phones don’t need connection tracking most of the time.
I really don’t know why you are mentioning Zimbra. Not like it has any issues on LVS with Vrrp, carp or ipvs. Quagga funny enough does not have problems either. You really really need to sit down build a LVS and find out surprise that a lot of the stuff you are saying are issues really are not if you don’t go the hard ways.
I find it funny how many times a BSD user claims Linux cannot do X when in fact it can. Also complain about particular compatibility errors between Linux and BSD not being aware that those errors also exist BSD to other Unix OS’s. So are pure example of where BSD are lacking.
That’s my biggest issue with trying to do anything on Linux: you have to search around for Solution X, figure out how to integrate it with Solution Y, discover it doesn’t have Feature 0 so you start over with Solution Z. Rinse, repeat, figure out how to cure stress-induced baldness, retire.
Yes, you can do pretty much anything network-related on a Linux machine. Doesn’t mean it’s easy, integrated, or worthwhile.
Ever tried to configure vlans, bonding, bridging, and openvpn on a Linux system? It’s not fun, and requires the use of multiple configuration tools each with a different syntax: brctl, vconfig, if-enslave, ifconfig, tun/tap stuff. And that’s only if you get the right versions of each of the above.
Doing the same on a BSD system is a breeze: ifconfig. Everything related to configuring a network interface is right there, with the same syntax for everything, nice and neat and integrated.
CARP on BSD is also integrated into the kernel, right into the networking stack.
CARP on Linux is done completely in userspace, and requires a whole shwackload of extra companion scripts to make it work.
CARP+devd on FreeBSD is so much nicer / easier to use than uCARP+?? on Linux.
The greatest strength of the Linux ecosystem is that there’s no “one single Linux OS”; it’s just a conglomeration of projects developed in parallel.
The greatest weakness of the Linux ecosystem is that it’s just a conglomeration of different projects, leading to a complete lack of any kind of standardised syntax for anything, and a complete disconnect between tools (is it fsck.fsname, or fsnamefsck, or something else; why not just “fsck -t fsname”? why is there 15 different tools to configure a network interface instead of just modules to ifconfig? etc).
However this is just opinion and one could say the same thing about a random Linux Distro, but it does not mean its true or false just user preference.
As I know many seasoned Linux admins, I don’t call this opinion, but common experience. In many Linux networks you’ll find some OpenBSD machines. That’s a fact, not fiction.
I also know a few admins who abandoned OpenBSD because its much slower in performance and does not scale well versus Linux, FreeBSD or Windows for that matter.
Sorry, but iptables doesn’t hold a candle to pf when it comes to configuration ability or power.
Yeah, really.
I honestly think this human is just trolling.
Maybe for cheap-ass home routers from Linksys or D-Link or the many off-brand stuff.
But get into the enterprise realm, and you’ll find either custom firewall OSes (IOS) or FreeBSD (Juniper’s JunOS) or FreeBSD (IronPort) or FreeBSD (hrm, not Barracuda, the other big SMTP gateway people). I’ve yet to find any major enterprise networking vendors that use Linux. It’s either custom code or based on FreeBSD.
Same when you get into the 10+ Gbe segment. Most of the driver development and testing happens on FreeBSD, then goes into Linux, then features get passed back and forth.
Not even close. Comparing PF or IPFW to iptables is like comparing a Porche to a dump truck. Sure, you can do more with the dump truck, but it’s not pretty, nor fast, nor easy.
What’s amazing to me is that even now, you cannot get a list of *all* your iptables rules, from all the different tables/chains, in a single command. Nor can you get a list of your current rules output to screen in a format that you can copy/paste into another system. Nor can you get a count of how many packets have matched each rule. Nor can you easily sync ruleset across two systems with live fail-over between them without dropping packets/connections.
And the list goes on.
Right.
Solaris ZFS is portable, and as we all know, it is highly innovative (read what FreeBSD developers says about ZFS portability). The same with Solaris DTrace, and other Solaris techniques.
It IS possible to be portable and innovative at the same time. If you choose not to be portable, that is ok, but do not try to pretend you are forced to not be portable. It is an active choice by the Linux community. And as we all know, Linux likes the Not-Invented-Here syndrome. Instead of porting existing mature software, new (and buggy code) is developed.
First mistake is “Now that Linux is the most popular free Unix-like operating system”. I think, the most popular Unix-like OS, that people get for free (with new computers) is MacOS X. I know it sounds wrong, because i can buy it after all. But when someone get it with new Apple computer, he actually get it for free.
And second sentence is “Choosing Between Portability and Innovation”. I think one is not affecting the other one. I mean, real innovative things do not need things, that already are working inside Linux or any other OS. They are innovative, because no one already used them, or do not made them in user friendly form.
Portability was one of innovations. If developers will make less portability, then their work will be less known, and making app Linux only will make it to gone, before it will shine to become popular.
Edited 2011-03-12 19:05 UTC
This fight is upstairs. It is fight of users, that identity themselves with OS their use (isn’t that childish?), or try to make look better by telling that they use some “better” OS (even more childish?).
On the ground, for developers, i think it is not working like that. They make software for many reasons like: i must make it to do my work on my OS (then developer do not care if it is BSD or Linux), i’m doing nice app for my company (some WebOS gadget at HP for example), i’m making software that i will try to sell (so he will choose Windows because the largest user base, what means more possible customers).
So please guys, do not fight, let us all help to make unified standards, that will make all apps work better in many OS, include Windows. So all Your software (developers) will make more money for You, and Your, many times good, but lost because of little user base projects.
Make standard libraries, that will work same in Linux, BSD, Windows ans Solaris. Make innovative libraries, fast and stable, then add layer – new innovative software, and spread it to the whole world, and get fruits of Your work! This is my wish for all devs
You’re talking of software you want to sell, we’re talking of open source. Apples Vs oranges? Well, you have a pragmatic view as professional developer, you do it for money and there is nothing wrong with it. But other people do it for fun, self-fulfilment, for the community. So yes people “fight” for their operating system of choice. Why? Because they don’t have a pragmatic view, they are the heart and soul of the community, without them there would be just another Windows or Mac OS, with some guy at the head, crying buzzwords at the customers.
Some people are more fighting than others or they are fighting just for fun. We call them Trolls. It’s a huge difference between fighting with facts or fighting with personal attacks, like some language zealot in this discussion.
The true is, i’m not developer at all. I’m just trying to put myself in this role. I see myself as someone who want to make software. What can be reason for that, money, girls, parties etc. Well, if i would do coding for fun, then still, i would do it for myself, so i do not car about other operating systems. If i make it for community, then i would have to define what it is “my community”. So it may be the user group, that like same thing that i do. For example, if i liek Linux, then this will be my community. But then, why should i care other OSs? And where is the onnovation vs portability then, and why should i care about it, i do it for fun!
And one more point, Red Hat did nice open source business, so open source do not mean making software only for fun and express myself. It may also mean money
There’s nothing nice to say about this level of short-sighted selfishness.
As I said there is nothing wrong with it. Everyone needs money. But it’s not the only motivation in life for many people.
From a server perspective, operating systems are fairly commodity. The BSDs got left behind with server virtualization. VMWare has 90% of that market, with Hyper-V (Microsoft) and Citrix (Linux also) taking the remainder. There are no compelling enterprise *BSD solutions.
You can run the BSDs on top of those, but other than die-hard fans, what’s the point? The biggest advantage that FreeBSD has over Linux in the server space is ZFS, although the feature set is limited from what you can get in Solaris (and it’s a joke in FreeBSD 7). I wouldn’t trust my data center on ZFS on FreeBSD.
OpenBSD has a solid reputation for security, but all the security attacks now are application layer for the most part, which OpenBSD is irrelevant. Plus the scalability issues that OpenBSD has had in the past (usually always dead last when compared to Free/NetBSD and Linux) make it a limited option.
ZFS is a great file system, and I was sold on it, however I can’t really use it. OpenSolaris has all but been abandoned, and Solaris has gone closed and enterprise.
So really, ZFS is a great concept, but it’s not all that practical anymore.
From a desktop perspective, all the FOSS are a mess. I teach networking and virtualization classes, and not one of the FOSS desktop OS’s would work for me (specifically for the second screen issue). I’ve got work to do, I don’t have time to futz around with Xorg settings only to have them broken with the next update.
The exception is Google ChromeOS (a paradigm which may very well be the future) and Android, but they’re not quite desktops yet.
Uhm…so? the BSD’s run great as guests on all of these platforms so there’s no problem. Who cares if the paltform itself is not BSD? Heck, KVM run OpenBSD better than any other virtualization platform.
I hope you realize that this is exactly the point of virtualization: to run other OS’s on the virtualization platform. I guess KVM should just drop support for all OS’s other than Linux since you are suggesting that there’s no point to run guests that are different from the host. Makes you wonder what we’d run on vmware…
The point? That they provide the features you need or that you simply like the way they do things better. You know, the exact reason so many Linux-fans stick with Linux.
Sssh, don’t tell that to all those data centres using FreeBSD+Jails.
There’s also a rising number of sites using FreeBSD+VirtualBox.
Sure, in the high-end enterprisey virtualisation arena there’s not native BSD solution. But that doesn’t mean FreeBSD isn’t/can’t be used in those situations. For example, we’ll be using FreeBSD+ZFS as the storage backend to our VM setup this summer.
But you would trust it running Linux + “filesystem du jour”?
ZFS is available via:
– FreeBSD
– ixSystems’ FreeNAS
– Nexenta
– GreenBytes
– Solaris
It’s even available on Linux, although the only full-featured implementation is via FUSE, which includes its own set of issues.
It’s not like ZFS is suddenly gone and no longer usable.
Between Nexenta and FreeBSD, ZFS will continue to be developed as open-source software. If Oracle never releases another line of ZFS code, development will continue.
This comes to my mind when I think about BSD guys attempting to collaborate with linux developers:
http://sourceware.org/ml/binutils/2000-11/msg00155.html
I don’t have to argue with you. You don’t know anything about the history of this. Either you believe what I say or go away. The value used is correct and is not to be changed.