Flame bait: Why BSD is dying, or How I learned to stop worrying and love Linux

Status
Not open for further replies.
You don't. /usr/ports/misc/compat8x... for example. Then, "rarely go smoothly"... I've the upgrade procedure in motd, read the Release Notes,
have backup. I even once updated with *much* difficulty direct v6 to v9. Desktop applications quaint? Multimedia support lacking? Broad overgeneralizations that are in my experience untrue. ( Of the remaining points I have an opinion on, I've posted elsewhere. )
 
Yeah. And I remember an Ubuntu upgrade making my PC unbootable (it wasn't an upgrade between LTS versions, though).

What about "There's no rapid deployment system like Debian's fai, or Redhat's kickstart."? My understanding is that Bsdinstall is scriptable.
 
IMHO It sounds like someone is afraid of a little hard work, Not sure why he would of gone with Ubuntu over something like CentOS.

I think this is just the regression of System Administrators shying away from really understanding their Systems. Also as much as it would pain the zealots of their favourite OS to say, not every OS is suited for every application and expecting to tweak out Ubuntu to perform like FreeBSD under certain conditions would be a pain.

Horses for Courses
 
These rants have mostly to do with how accustomed is a user with a specific operating system - when it comes to Linux I'd say how accustom is to the specific flavor's mindset - and much less has anything to do with the 'real value' of the OS in question. Reading them, mostly is a waste of time.

As an example I'll tell you that for me it's very easy to do everything on Gentoo. I know exactly what files to edit in etc, to make portage extremely personalized.

Once I tried to do manual kernel recompile on a debian system. I lost something 2 days, without success having to re-read pages of documentation in order to understand a process that in theory should made the task 'way easier'. The debian kernel recompile process however, didn't make any sense to me. Any sense at all. I still consider it as the most stupid approach I've ever saw in an otherwise simple task.

But if you ask a debian user, he might come with much of crap about portage and non-automatic kernel compiling solutions (although gentoo has one or two), etc.

So, if he has problem with the FreeBSD upgrade it's because he didn't read the manual or didn't approach the problems in the right (freebsd's) way.
 
I've found that FreeBSD's ports/packages system is easier to automate (though I know plenty of Linux variants have come along and in some ways surpassed FreeBSD) and dependencies are usually correct as long as you pay attention and keep the ports tree up-to-date. the base and kernel are much easier to automate and take approximately 0 attention once you have done them once - and the same methods and automation scripts I used in 1998 still work. If you are a passable script author, you can write a fairly short script to handle installation - using xapply you can rapidly perform these tasks on many hosts. FreeBSD is excellent for the kind of admin that isn't looking to point-and-click their way through installations, and has been known for security, stability and performance for a long time.

Many Linux-compiled binaries, if you have Linux compatibility packages set up, actually run faster in FreeBSD, and pretty much all major Unix-type free software will compile in it (though whether one wants to is purely up to the individual). Much like in the old guard proprietary Unix variants, you can go to one central place (this website and forum) for most relevant information about how to do what you want in FreeBSD. The kernel is here. The base and CLI userland is here. Various native apps are here. The maintainers of ports and packages are delineated on this site and not especially difficult to find. FreeBSD will also run binaries from many other X86-based Unix-type systems, also better in many cases. FreeBSD is clear about supported platforms, and you can find out about all of them on the main site. I also find FreeBSD's basic design and layout (device naming, location of various binaries) to be consistent - with the exception of new subsystems found in ascending full releases, many things are exactly the same and have been for a long time.

The main reasons big corporations have moved to Linux IMO are recognition of the name and the GNU project, longer-established fulltime enterprise support for certain distributions (though an enterprise incapable of doing most of its own support is in a precarious state as far as I'm concerned), and fairly best-of-breed SysV layouts and userland (closer to non-deprecated versions of Solaris). Many free software projects start as Linux-based code, although mature multiplatform codebases are likely to have a FreeBSD version. The learning curve is a little lower for many tasks, particularly if one is more used to GUI-based frontends and administration - but many, like myself, find those additional API layers to add rather than reduce complexity.

I have done production and test/dev installation, administration, and auditing on many different Unixlike OSes including a ton of Linux variants, and I still prefer FreeBSD when possible. I would happily choose it for any possible production-quality applications, though there are various strengths and weaknesses when compared to Linux.

This is just my experience, and I will work with any recent (or less recent, if need be) Unix-type OS. We're still all in this together when it comes to Unix-style OSes based on freely available software and source code, and the specialization of certain distros (e.g. I really like BackTrack Linux as a liveCD used to audit security) makes them better or worse for various purposes; still, i will continue to choose FreeBSD for general Unix server-type purposes when plausible.
 
Well, I'm not quiet sure but still think that GNU/Linux may be faster than FreeBSD (that's why some companies begin to choose it instead of *BSD`s) - even CERN has Linux machines on LHC system, but not FreeBSD :)).

Linux kernel has real support for A LOT MORE hardware than FreeBSD does (you don't need to find hardware for os, but os for hardware, again that's why some companies begin to choose it instead of *BSD`s).

To sum up, BSD is ancient (and maybe slow...).

For regular home-dummy user:
Versions of programs in ports of FreeBSD are really old compared to originals made for GNU/Linux.
As for using a laptop I'm not satisfied neither with FreeBSD, nor with OpenBSD - all of them have poor acpi and modern hardware support (second BSD is even a lot slower than first one).
FreeBSD may be good for oldschool admins with great beards (who are real GEEKS and enjoy having a possibility to infinitely tune just every smallest detail in system - polish system until it shines like a brilliant), but it's still not ready for regular-housewife-like user who wanna run this kind of system on laptops or desktops.

Overall, everything is documented in *BSDs, you can set up everything from reading through man`s.
But what I do reeeeeeally HATE is BSD`s very long directory layouts in "/usr" and so on", dead forth language in loader scripts (which I always delete and use loader.rc instead), and no boot to GPT from 32 bit x86 UEFI BIOS.


Thanks.
 
freemason, it's true that depending on the distro, kernel parameters, specific application, etc. that Linux can be faster - but it goes the other way often enough that I consider it pretty much a draw except for certain specific circumstances (Oracle, for example, is very specific about what you must run to be supported, and they don't support FreeBSD, so I wouldn't bother trying to run a RAC/CAD/VDB Oracle cluster in FreeBSD in an enterprise production environment) - performance tests often vary from version to version to such a degree that, as an example, FreeBSD might benchmark with a better score on Apache static HTML page GETs one month, and some variant of Linux might the next month as tweaks are made in the kernel/filesystem/Apache codebase. Linux does support a lot more hardware and many many more platforms, but NetBSD supports a ton of platforms and so might be choice for certain architectures - and I think the base and common natively ported daemons tend to be a little better tested for stability and security in the open source BSDs. Anyone wanting to run the latest and greatest hardware for gaming or audio production will likely find it better supported in Linux, if anything other than Windows supports it.

In the end, a lot comes down to personal preference - I prefer BSD-style userland and directory layout to SysV in general, so i will give FreeBSD a point in my personal book for that. Commercial software in general overwhelmingly supports Linux better, in my experience - and while FreeBSD may run that software just as quickly or stably, you will not likely get Enterprise-grade support with quick-responding Service Level Agreements in FreeBSD for those Linux-native binaries. There could be a great community of folks running the software combination but big companies will not approve 3rd party software if it is not officially supported in a great many cases, particularly if they can get official support for RHEL or similar. Neither OS has a consequential portion of commercial desktops, but Linux has notable installed bases of certain GUI-based applications that will likely never see an official FreeBSD version - partially because Mac OS/X is BSD-based and has loyal fans for multimedia apps that have neither Linux nor *BSD native ports. Linux clusters and VMs do the job well enough that nobody bothers to make the same happen in FreeBSD unless development of these happen in-house at a company that prefers Free/Net/OpenBSD for its own purposes.
 
freemason said:
Well, I'm not quiet sure but still think that GNU/Linux may be faster than FreeBSD (that's why some companies begin to choose it instead of *BSD`s) - even CERN has Linux machines on LHC system, but not FreeBSD :)).

These are the companies who think otherwise; Also when building a Cluster there is far more than Speed alone that is factored in when selecting an OS, including Driver Support, Legacy Software support, it can even come down to what the System Admin likes more or has more training in. So saying that it's "slower" is just conjecturbation.

Citrix Netscalers
F5 Networks's 3DNS version 3 global traffic manager and EDGE-FX version 1 web cache (NB These are now end of life with 3DNS functionality being moved to the Linux based BIGIP Platform)
Ironport network security appliances
Junos network operating system by Juniper Networks used in their routers, switches and security devices
KACE Networks's KBOX 1000 & 2000 Series Appliances and the Virtual KBOX Appliance
nCircle's IP360
NetApp's Data ONTAP 8.x and the now superseded ONTAP GX (only as a loader for proprietary kernel-space module)
Netasq security appliances
Nokia's firewall operating system
Panasas's and Isilon Systems's cluster storage operating systems
The PlayStation 3 video game console.[30]
Sandvine's network policy control products[31]
Sophos's Email Appliance[32]
St. Bernard Software iPrism web filtering appliances[33]
Panasonic's 2010 TV models (PDP and LCD)
Blue Coat's ProxySG WAN acceleration appliance is partially derived from FreeBSD[34]
Netflix's Open Connect Appliance[35][36]
Other operating systems such as Linux and the RTOS VxWorks contain code that originated in FreeBSD. Debian, known primarily for using the Linux kernel, also maintains GNU/kFreeBSD, combining the GNU userspace and C library with the FreeBSD kernel.[37] Darwin, the core of Apple OS X, borrows FreeBSD’s virtual file system, network stack, and components of its userspace. The OpenDarwin project (now defunct), a spin-off of Apple’s Darwin operating system, also included substantial FreeBSD code. Thanks to the permissive FreeBSD License, much of FreeBSD now also forms the basis of Apple OS X and OS X Server.

freemason said:
To sum up, BSD is ancient (and maybe slow...).

Maybe slow? Do you have no metrics to back this up... This is just conjecture and adds nothing to the conversation. How does the age of an Operating System have anything to do with it's performance? Don't forget your using an Internet that runs on +20 year old Protocols :)

freemason said:
For regular home-dummy user:
Versions of programs in ports of FreeBSD are really old compared to originals made for GNU/Linux.
As for using a laptop I'm not satisfied neither with FreeBSD, nor with OpenBSD - all of them have poor acpi and modern hardware support (second BSD is even a lot slower than first one).
FreeBSD may be good for oldschool admins with great beards (who are real GEEKS and enjoy having a possibility to infinitely tune just every smallest detail in system - polish system until it shines like a brilliant), but it's still not ready for regular-housewife-like user who wanna run this kind of system on laptops or desktops.

Old? Or Stable? No where in the FreeBSD mission statement does it say it intends to be the best Desktop Operating System in the world. The OP was right with his subject name, this type of debate clearly has no point; It could be likend to saying a hammer is the best tool to build a house and makes many assumptions about what the application of the OS is, the skill of the user(s) and the requirements of the software/application.

This is like:
  • Cisco vs. Juniper
  • Holden vs. Ford
  • Andriod vs. iOS
  • vi vs. emacs

No one will win, unless there is a way to rewrite someones mind :)
 
I am always suspicious of one's capabilities when one switches from something like FreeBSD to Ubuntu.

I am also suspicious of this guy's capabilities based on his points, some of which are blatantly false.
 
freemason said:
but it's still not ready for regular-housewife-like user who wanna run this kind of system on laptops or desktops

With no offence, but... this kind of people fall to pre-installed, pre-chewed, pre-digested Windows/OSX/Linux machine. People that has a lot of spare time should not use it to install FreeBSD, this doesn't mean that FreeBSD is an elitary OS, that means you need an OS that is a solid base Unix system, open to any kind of implementation that runs on a wide range of hardware. It's not always easy to implement it as you want, nothing comes easy.

However, there are still differences from a Windows/OSX and a Linux system for housewifes et al, so the die is cast.
 
Ok, to twist the "time consuming" argument a bit : I have little spare time and I still prefer FreeBSD. Ports can compile while I do not watch, not a problem. But I know that, with my urge to tinker around, I need to do a lot more damage to the system to be forced to a reinstall than under Linux. FreeBSD allows me to boot and repair damages which send an Ubuntu user on the search of the install medium. I know this kind of screw up will happen once in a while and I simply have no time to reinstall every time it happens. So, I save time.

But what I see behind some of the "it just works" arguemts provided by the red hat admins is that red hat / oracle have definite views about what they support and so you can shift the blame by telling the PHB that tinkering with the system in the way he wants would void the support contract which is expensive - thus not being asked to do it. And if you did not touch it, it is not your fault if the thing does not perform as expected.
 
lele said:
Not a recent article, but it hasn't been discussed here:

http://www.section6.net/2011/07/why-bsd-is-dying-or-how-i-learned-to.html

EDIT: Seriously, not all of his criticisms still apply. For instance, besides ports, nowadays we have packages available. However, I didn't know that when upgrading the system you had to recompile everything, even things you didn't compile in the first place.
That article is perfect. I couldn't said it better. This is exactly the reason becouse I pretty much terminated professional use of FreeBSD and I moved to Linux (even Gentoo when I need more system customization).

After all this years, the lack of a serious binary packages serious upgrade system is simply unacceptable.

Do they want to stay in a ghetto? It is ok for me, there are lotsa alternatives this days and honestly I don't miss all the lost time upgrading and updating FreeBSD from ports.
 
Why does Ports VS Binaries come up so much? I always perceived source code to be the standard write once compile anywhere goodness. I am a novice sysadmin deciding on what OS to use for my next server/project. This is a great discussion btw. :stud
 
jwele said:
Why does Ports VS Binaries come up so much?
Becouse it is the core of the problem. They said they do not have enough resources to put up and regularly update binary releases and this is what can definitely put them in a ghetto. Sure, if they base them decisions on this forum feedback they could probably die of misconception considering many people writing on threads here and supporting ports to death are not professionals, they are supposed geeks just spending time playing with something they think cool. Many of them don't have any idea about what is needed to make dozens of systems up, ready, secure and before of this updated systems day after day, in the faster time possible, with the less downtime and with the less effort.

And considering that is my job and I do work to get the money and I have so many interests after my day job, I don't want and I don't have the time to loose time with stupid ports (often broken) and to find solutions just to support an OS pretty much remained at the glorious nineties times.

Also the kernel, this days, seems less robust to me. I personally tested the memory leek reported many times here and hardware support is quietly poor. Considering, this days, updating hardware is often less expensive compared to loose time to work with an outdated, problematic system, also for this FreeBSD (totally poor new hardware support) is out of the options a lot of company evaluate before to load an OS.
 
I would reply constructively to your post, but almost every sentence contains an outright insult. Would it not be wiser to solve linux problems over on the linux forums, or seek FreeBSD solutions here on these forums? Particularly since the type of post should never happen once you have search engines, with which the problems can readily be solved or answered or asked?
 
So you are saying that FreeBSD needs a binary package management application to "stay in the game"? I still feel like FreeBSD is harshly underfunded to do as much as they do currently.
 
jb_fvwm2 said:
I would reply constructively to your post, but almost every sentence contains an outright insult. Would it not be wiser to solve linux problems over on the linux forums, or seek FreeBSD solutions here on these forums? Particularly since the type of post should never happen once you have search engines, with which the problems can readily be solved or answered or asked?

Piggy is a professional troll. Since he makes his living being one. He contributes absolutely nothing to the project. Just a tiny user who can't see the forest though the trees. I find it very hard to help those who can't help themselves.
 
In fact, FreeBSD has at least two binary package systems at present, the old pkg_install and the brand new and highly regarded pkgng. There's another package system from PC-BSD with their PBIs.

What we don't have are "stable" and "current" package repositories. Some people want that badly, in the belief it will add stability to the ports/package system. Others think it will add more overhead for already overworked, volunteer port maintainers and committers. One of these two viewpoints is absolutely correct, the other obviously wrong. I forget which is which.
 
FreeBSD cannot so easily die.Because it's the foundation of other 3 mainstream OS'es:

Windows NT was inspired by it, Mac OS (X) is based on it and Linux borrowed many basics from it.

I personally use FreeBSD as a testing and development environment especially while compiling open source codes, i.e. not resorting to pre-compiled packages.

Nowadays, only an ARM port of FreeBSD is really missing. We need to have the ARM version of FreeBSD releases in no time.
 
With modern CPUs getting faster and faster, and the upcoming switch to LLVM. I would have thought ports compilation times would become less of an issue?
 
I ignored this thread for a while, but since it is gaining momentum...

I don't know where the attitude of "I don't use FreeBSD 'coz I don't want to have to compile EVERYTHING maaan!" comes from, but I find it completely untrue. Oddly enough, I suspect it came about due to the excellent ports system.

I do however reluctantly agree with some of the views of FreeBSD's insistance that it is not a Desktop OS. This may be true! But I strongly believe the reason GNU/Linux has become so popular - to use Ubuntu as an example - is because of its pandering to the desktop user. Remember, people will choose to run as a server something they are comfortable with. And it is much more likely future system administrators will first experience GNU/Linux, which will perpetuate things further.

I don't have an answer for what I've posed as a problem, I don't think this project has teh resources to tackle it. But I don't see the tide of "UNIX" applications that are really "Linux" applications stopping anytime soon.
 
Status
Not open for further replies.
Back
Top