Linux certainly has the corporate backing(1), popular mindshare, and lots of workforce, but there is something about the general cohesive feeling of BSD systems that some people really like. I do at least!
I consider the BSDs as more “cathedral” while Linux is more “bazaar”. Linux is also a kernel with a base GNU user-space (written by other people), and now systemd, all packaged by a distro. In BSD-land the base user-space is released by the same team as the kernel. So it is different, and some people enjoy that difference.
If you were a FreeBSD user, you may also wonder why use DragonFly? As a FreeBSD user, I love that DragonFly is trying new things and focusing on cluster computing and high performance. Diversity is great!
(1): something like 90+% of contributions to the Linux kernel are apparently corporate sponsored
Why can’t someone just implement a Linux patch to bring that level of SMP to Linux?
Software is soft so yes, someone could do this. But it’s not like a 10 line patch, it’s a fairly large architectural change in how the kernel works. This is the cause of the DragonFlyBSD and FreeBSD split, the FreeBSD leads didn’t want to make those changes.
So it would be a lot of code to change plus the political battle to get people into it.
I think only you can answer that. What is your usecase for Linux or the distribution you use?
Why can’t someone just implement a Linux patch to bring that level of SMP to Linux?
I think in this case it’s largely about design decisions and how hard migrating to new code is. There is just more than way to do many areas of IT. Operating Systems, a bit like programming languages differ mostly in those different design decisions, philosophies and user interface (or syntax).
Have you seen this post by Matt Dillon yet? I am sure this could be done in Linux, even in a more Linux-y way. That doesn’t mean it will be done. In the end that might not even be a bad thing, as going in different directions when developing software sometimes leads to an overall worse system, where you run into bugs, have a lot of complexity, have to make all sorts of compromises, etc.
Again comparing this to programming languages might make this more clear, since philosophies and design decision are more dominant there, even in general purpose languages, where in general purpose OSs these seem to be exceptions (OpenBSD for example has a strong emphasis on what it considers good). When it comes to programming languages you see JavaScript and Perl as two programming languages that took ideas and philosophies from pretty much everywhere else - Perl did that more outside of stdlib though thinking about Perl Moose. At the same time we see languages like Go, Python and some functional languages that try to keep certain philosophies and opinions dominant. Note that neither programming languages nor operating systems prevent doing it different however, people are still opposed.
In the Linux world this actually is a big part of why patch sets exist. They both show it can be done, and sometimes still won’t be taken into to main source repository, for various reasons, technical and philosophical - and that despite Linux likely being way more open minded about ideas that might or might not be forgotten about in just a few years.
@trousers: Could you add a dragonflybsd tag. Make searches/filters more easy. Thanks!
I certainly will next time. Alas, I don’t think I can add tags after a certain period of time. At least I don’t see any way to do so…
[Comment removed by author]
Linux certainly has the corporate backing(1), popular mindshare, and lots of workforce, but there is something about the general cohesive feeling of BSD systems that some people really like. I do at least!
I consider the BSDs as more “cathedral” while Linux is more “bazaar”. Linux is also a kernel with a base GNU user-space (written by other people), and now systemd, all packaged by a distro. In BSD-land the base user-space is released by the same team as the kernel. So it is different, and some people enjoy that difference.
If you were a FreeBSD user, you may also wonder why use DragonFly? As a FreeBSD user, I love that DragonFly is trying new things and focusing on cluster computing and high performance. Diversity is great!
(1): something like 90+% of contributions to the Linux kernel are apparently corporate sponsored
Software is soft so yes, someone could do this. But it’s not like a 10 line patch, it’s a fairly large architectural change in how the kernel works. This is the cause of the DragonFlyBSD and FreeBSD split, the FreeBSD leads didn’t want to make those changes.
So it would be a lot of code to change plus the political battle to get people into it.
They go into detail in this whitepaper:
https://www.dragonflybsd.org/presentations/dragonflybsd.asiabsdcon04.pdf
my use case is I can’t use OpenBSD and I have a lot of cores = DragonflyBSD. with vmm I don’t even have to mess with FreeBSD and bhyve ever again!
I think only you can answer that. What is your usecase for Linux or the distribution you use?
I think in this case it’s largely about design decisions and how hard migrating to new code is. There is just more than way to do many areas of IT. Operating Systems, a bit like programming languages differ mostly in those different design decisions, philosophies and user interface (or syntax).
Have you seen this post by Matt Dillon yet? I am sure this could be done in Linux, even in a more Linux-y way. That doesn’t mean it will be done. In the end that might not even be a bad thing, as going in different directions when developing software sometimes leads to an overall worse system, where you run into bugs, have a lot of complexity, have to make all sorts of compromises, etc.
Again comparing this to programming languages might make this more clear, since philosophies and design decision are more dominant there, even in general purpose languages, where in general purpose OSs these seem to be exceptions (OpenBSD for example has a strong emphasis on what it considers good). When it comes to programming languages you see JavaScript and Perl as two programming languages that took ideas and philosophies from pretty much everywhere else - Perl did that more outside of stdlib though thinking about Perl Moose. At the same time we see languages like Go, Python and some functional languages that try to keep certain philosophies and opinions dominant. Note that neither programming languages nor operating systems prevent doing it different however, people are still opposed.
In the Linux world this actually is a big part of why patch sets exist. They both show it can be done, and sometimes still won’t be taken into to main source repository, for various reasons, technical and philosophical - and that despite Linux likely being way more open minded about ideas that might or might not be forgotten about in just a few years.
SMOP!