FreeBSD's Executive Director Calls For Linux + BSD Devs To Work Together

[SystemCrasher]

With that attitude; Linux wouldn't even have existed. We would still be knocking away with Minix.

Linux appeared thanks to attitude of mr Torvalds, who preferred to solve real-world problems under reasonable assumptions and wrote something he at least could and would use himself to further develop that. I dare to say a lot of BSD ppl miserably fail at this part, mumbling some crap about windows, mac or something. That's where it gets hard to take them seriously. Most interesting part about Torvalds? He can eventually be quite toxic, however, if you come with good technical reasoning, he would hear it and consider it - and suddenly become far less toxic it you've got the point.

Competition is important.

I would somewhat agree, with remark that joining efforts and improving same thing could be more efficient than bikeshedding, reinventing the wheels amd competing - as long as development process isn't stuck/runaway. Competition is a watchdog against getting development processes totally wrong - but it comes with plenty of overhead (duplicated efforts).

I am also very happy the OpenBSD project "NIH" developed OpenSSH or we would still be using an SSH implementation that is license encumbered and lacking.

First opensource implementation hardly qualifies as "NIH" since you can't, say, grab existing implementation, patch it, improve it and make it doing what you need/want. That's what makes proprietary software so inconvenient and annoying - it brings tons of artificial problems that shouldn't exist in first place. It also what eventually obliterated various proprietary BSDs and *nixes in favor of Linux...

Also, Linux didn't take the lead. Windows kinda did

That's depends on how you define "lead". I've referred to "technical" side rather than marketing, Linux used thanks to its performance and features rather than thanks to marketing. By looking around me I see dozen and half of "computers" (devices capable of running my posix programs) - none of them run Windows. Most of them wouldn't even boot windows, as it lacks drivers/cpu support to begin with. Even casual users carry billions of android devices, that are "linux" (yes, rather shitty and troublesome) - far more numerous than PCs with windows. Same for "embedded devices" like TVs, routers and so on. And, erm, there was funny post by Windows dev explaining why Linux works faster than windows. Actually, speaking for myself, Linux performs much better than Windows for me and I'd say expertise in Linux system internals proven to be more rewarding for me than I ever had chance in windows. No matter what I would do, it hard to impossible to adapt windows to use cases I could be interested in, not to mention its licensing is prohibitive as well, just as much as I hate forced reboots, not really excited about sending all I type to MS (=NSA) servers "to improve quality of service" or suddenly failing "activations" and so on. So uhm, well, if someone thinks Windows got the lead (in technical terms), they are perfectly free to try to borrow techs (and get sued by MS, geez). I don't get how praising windows benefits anyone in opensource world though. This said I'll admit my desktop somewhat resembles what has been in win9x/2k/xp - however, implementation and details are different, it just overall concept that survived and ironically MS recently departed from it far enough to get unusable for me. So if I have to describe MS and windows in just 2 words, it would be... "good riddance"

[pal666]

It'd be really cool if AMD, Valve, GhostBSD, and Wine all teamed up to make SteamOS 2.0. Just being able to leverage ZFS would be damn spiffy; like per game datasets and zones...just imagine being able to snapshot a game before the update, updating it, and then being able to revert without needing to redownload anything (bandwidth waste) and wasting time on a scan like with Steam now; and all of that in a sandboxed environment so we don't have to worry as much about cheaters spreading malware, games that bug out, etc.

everything is available under linux. zfs is obsolete by design, use btrfs for snapshots. and don't even try to start imbecilic mantra about btrfs stability - 1) zfs is stable on solaris, not somewhere else and 2) your imaginary steamos 2.0 not just unstable, it doesn't even exist.
nobody is insane enough to support os with no developers and no users because some clueless forum commenters have succumbed to propaganda

[wizard69]

You do realize that corporations develop all sorts of code for internal use that never sees the light of day, right?

Beyond that hat I really don’t see any benefit to the community here to condemn one license over the other. Open source can be a good thing for various reasons and because those reason vary so much it is important to have a range of licenses to choose from.

If you want to contribute to both, you have to water down your work to a crappy permissive license. I don't think that's the right way to go, as you let corporations take your effort and use it to further their proprietary software.

[AdrianBc]

adrian

Sorry for not quoting your comments, but it's little hard from mobile phone. I'm using Linux since 1999 if I remember correctly. Even then Linux was outperforming FreeBSD in network performance. Probably not always, but it wasn't slower. There were problems with malloc library, but it was enough to replace it for another one to fix this. Even then Linux had superior filesystems. FreeBSD got one from Sun, so the old times weren't much different. Maybe 1993, but not 2000 for sure.

I have been using both Linux and FreeBSD continuously since 1995 and despite the permanent jokes about when there will be the year of the Linux desktop, I have never had any problem with using only Linux on all my desktops and laptops since 2004. Before 2004 I had to use dual-booting for Linux and Windows because there were still Windows programs without Linux equivalents.

While Linux became more convenient for desktops very early due to the availability of more programs and more device drivers, FreeBSD was better for many kinds of servers and for certain purposes it is better even now. I have always used both operating systems, each for the appropriate tasks, so I have no reason to make false claims about them, like those who obviously do not have any direct experience but feel entitled to have opinions about what they do not know.


Approximately In the decade 1995 - 2005, I have never seen Linux outperforming FreeBSD in network performance. During that time I was designing network equipment and I had to test its performance. Neither Linux nor Windows were able to reach the hardware performance in all test conditions, but only FreeBSD, so that is what we were using for network testing.

Of course, for certain hardware configurations it could happen for Linux to outperform FreeBSD, if that hardware happened to have good Linux device drivers but bad FreeBSD device drivers. FreeBSD had usually very good drivers for Intel hardware, e.g. Intel NICs, but less good drivers for hardware from other vendors.


"Even then Linux had superior filesystems. FreeBSD got one from Sun"

The FreeBSD UFS file system is not from Sun. The FreeBSD and Sun file systems are different and both are independent evolutions of the older BSD file system.

Which operating system had superior file systems depends on the criterion. The Linux ext2 file system was indeed very fast, but that was because it cheated, so it was very unreliable, which made it inappropriate for many servers.


Around that time almost everybody transitioned to journalled file systems. Nevertheless at least the initial versions were far less reliable than advertised.

I was living in a place were power outages were frequent, but I was still not able to afford UPSes. Because of that I was forced to witness a lot of unwanted experiments about what happens to a file system after a power loss.

Despite their journals, sooner or later ext3, xfs, reiserfs and NTFS became corrupted after a power loss, some times so badly that an OS reinstallation was necessary.

On the other hand, the only file system which never had any corruption problem (except for the possible loss of some data that was being written when the power was lost), was the UFS of FreeBSD.

UFS was not using a journal but a careful ordering of the disk writes. Unfortunately that method is probably no longer relevant today, because it seems that the code was well understood only by the original maintainer so it might not work as well today, after many revisions. Moreover. the modern hardware might reorder the writes in another way than in the order that the software tries to enforce.


In conclusion, while the Linux file systems were faster, UFS was far more dependable, so for me it was the superior file system. I was able to move most of my data to Linux file systems only after I could afford UPSes for all computers.

[wizard69]

I might add that Open Source simply isn’t good enough to be used in mission critical systems on its own. Honestly how much kernel code or user space code has been proven to be correct? I don’t mean some yahoo crying out that we tested it but rather code that has been mathematically proven to be correct.

While i I love the idea of electric vehicles, the idea that our roads will soon be populated with self driving cars scares the hell out of me. We all ready have had several deaths directly related to the control software in these cars making serious mistakes, mistakes that a human paying attention would not make.

So unless people in the open source world want to go through the extensive validation processes seen in other industries the idea that open source has a singular play in critical software, where human life is at high risk, the issue simply has no play in this discussion.

A number of people died from automotive sudden acceleration. It had nothing to do with them being bad drivers. The vehicles randomly went out of control as if the gas pedal had been floored and the brake pedal was ineffective at stopping them. The brakes were destroyed trying to stop an engine at full throttle when it happened at high way speeds. People in other vehicles could have been killed too. You have a sick mind if you mean to suggest that those deaths were a good thing. Such failures can kill anyone, including you.

By the way, people using/putting OSS software under licenses other than the GPL was what was claimed to be abuse. I had been replying to that. Asking what stops people "from following the rules of [the GPL]" when the GPL is not involved is nonsensical. Also, the L in GPL stands for license. saying "GPL license" does not make sense. You are effectively writing the word license twice.

[pal666]

Competition is important. I am also very happy the OpenBSD project "NIH" developed OpenSSH or we would still be using an SSH implementation that is license encumbered and lacking.

we are not using openssh. we are using its linux fork because upstream openssh does not support linux.

Also, Linux didn't take the lead. Windows kinda did

what planet you are from? on earth the most used os is linux

[pal666]

Apple didn’t rape anybody. More importantly they have contributed significantly to the open source world and especially BSD.

you mean, freebsd has access to apple kernel and userland code?

[wizard69]

Where did did you get this idea that brakes must be able to stop the engine. If this is true there are plenty of examples where this isn’t true. The fact is brakes are only effective for a limited time, anybody every driving down a steep mountain road can attest.

Small note: While it's true that some floor mats where badly designed leading to potential unintended acceleration, there have been no destroyed brakes. The brakes in cars are by legislation always more powerful than the engine, and so far there have not been a single verifiable situation where applying the brakes does not stop the vehicle (only anecdotal claims exist).

[pal666]

Of course, for certain hardware configurations it could happen for Linux to outperform FreeBSD, if that hardware happened to have good Linux device drivers but bad FreeBSD device drivers. FreeBSD had usually very good drivers for Intel hardware, e.g. Intel NICs, but less good drivers for hardware from other vendors.

for random hardware configuration freebsd doesn't have drivers, linux has drivers. that way freebsd can't compete with linux and never was able to and never will. linux has software, freebsd doesn't have. just look at their discussion of how git is "not designed to support their huge repo" - linux kernel-only git repo has same size as all of freebsd kernel+userspace git repo and kernel's history starts from 2005 when freebsd from 1993

[Volta]

adrianb

I still cannot agree with you. UFS wasn't so good and it was hardly compatible between implementations. It corrupted more often on power loss than Linux file systems and it was slow as you said. When comes to networking there was problem with broken gnu malloc library on Linux. It was enough to replace it. I didn't see Linux loosing to FreeBSD in such benchmark and Linux very quickly conquered the server market.