71–80 of 224 posts
It's really cool, but I get worried anytime I see something that seems to want to be a user friendly, mainstream distro, after seeing how hard Manjaro tried with the whole "Bring simplicity to the masses" stuff and how many issues they still have. Is this an experimental or hobby distro or does it intend to be something grandma can use? If I had to choose this or BSD, it looks like I'd very strongly considering this…
I feel like they made a lot of good decisions in being simple enough without too simple, and backwards compatible enough without too much legacy... but imho wayland doesn't fit in that same space for me (at least not yet.)
> Chimera is a general-purpose OS born from unhappiness with the state of Linux distributions. https://xkcd.com/927/ It seems to have a nice build system. But I am not sure this solves any issue any other distribution has. And with the amount of opinionated choices this distribution has I imagine there are few who it fits. I think for many, Void would be a better choice. They do defend/document their choices well. Th…
cbuild is also designed to solve real issues, particularly proper sandboxing and linting of builds to allow for a small team to maintain it while at the same time being fast and ensuring correctness (distro build systems often tend to be bash/make monstrosities that do nowhere near enough where they need to while doing too much where they shouldn't)
i used to work on void, and there are good reasons i went to start a new project instead; this includes poor quality packaging for a lot of void stuff (by part caused by xbps-src design choices), poor platform support (all non-x86 stuff is buggy, partly because of cross-compiling and not running unit tests for things being built) and state of infrastructure, design issues of xbps itself (lack of solver, poor implementation of shlibs system and virtual packages, etc.), lacking service management (no oneshots, no transparent/standardized support for user services, no support for dependencies, no support for service-integrated dbus activation etc.), lots of stuff sticks with legacy nonsense where systemd has taken over and improved things in the meantime (e.g. handling of dbus, user management and tmpfiles management, user services as mentioned before, etc), and so on and so on
we are also addressing the whole thing with "everybody has adopted systemd and its functionality, and distros without systemd have instead stuck their head in the sand and pretend systemd is the devil and implement none of the stuff people want" which is leading to increased dependence on systemd in various upstreams, and an increasing amount of bad hacks in non-systemd systems
supporting a lot of architectures is not something to regret, it's a good goal to have and aids keeping the ecosystem healthy as well as giving people the choice of running the system on hardware they like; any pain caused by supporting multiple archs pays off several times
I think I'd be a lot more excited about this if it didn't seem to be bought into the wayland hype train. I feel like they made a lot of good decisions in being simple enough without too simple, and backwards compatible enough without too much legacy... but imho wayland doesn't fit in that same space for me (at least not yet.)
Earlier quoted context omitted.
[musl is] slower than glibc, and in occasionally difficult to predict spots Note that sometimes (/often) this purely due to the allocator, and as far as I remember Chimera uses a different allocator than the stock musl one.
Chimera uses the scudo allocator, which is Ajay also the default on Android and Fuscia. Yet I can't find many benchmarks. Frustratingly, it's listed here, but not actually included in the results: https://github.com/daanx/mimalloc-bench Might be fun trying to run the benchmarks myself.
that said, scudo is highly configurable, and the performance reflects the configuration (you can even swap out the primary allocator, secondary cache, you can tune all the parameters, the TSD registry implementation has a big impact on multithreaded code, etc.), so it's not 100% representative - chimera's current configuration is further tweaked for both better performance and lower memory usage
scudo being this configurable is excellent because it is what enables such integration in the first place (other allocators e.g. frequently rely on ELF TLS, i.e. __thread and the likes, which would make libc integration very difficult and would require major changes to the dynamic linker, with scudo we instead implement a custom TSD registry and simply shove an extra pointer in the pthread structure)
Earlier quoted context omitted.
This has been my experience with alpine as well. musl just ends up causing much more trouble than it is worth
For the longest time using anything but gcc was a problem. But people found clang/llvm better in many respects and persisted. Today clang/llvm is a drop in option, even to compile the Linux kernel. Similarly we should persist with musl. It’s infinitely cleaner, simpler and understandable.
I am in favor of standardizing the behavior of libc and having multiple implementations that are commonly used.
Earlier quoted context omitted.
Good to know. That was my first question. I hope this means they also get rid of nonsense like pulseaudio and logind. I’m a bit skeptical of the “shared session bus” though. Linux supports inode watches for things that want to react to configuration changes, and session busses add a ton of complexity (including a second authorization language, and a userland SPOF)
They use pipewire IIRC (which is 1.0 now!!). And they have a custom alternative to login AIUI.
For now they seem to be using a different alternative to do at least some of it. I think it was called elogind?
Earlier quoted context omitted.
Android uses the Linux kernel but it is not what people mean when they say “Linux”. Speak for yourself. In my mind, Linux is home routers, smart TVs, digital signage, tablets, smartphone, fridges, and yes, servers and workstations.
Are you saying that just to be contrarian? When somebody tells you they're running Linux on the laptop, you actually think they are owning a chrome book (considering that there are likely more chrome books than laptops with other linuxes), when you read a job add which requires Linux admin knowledge, you think they mean being able to install apps on android?
Earlier quoted context omitted.
[musl is] slower than glibc, and in occasionally difficult to predict spots Note that sometimes (/often) this purely due to the allocator, and as far as I remember Chimera uses a different allocator than the stock musl one.
Happy chimera user here. It feels fast in regular usage though I use ssh much more than desktop. What’s actually amazing about chimera is the musl + clang/llvm/compiler-rt combination. It allows you to do all kinds of amazing things easily from a programming perspective. The apk 3 package manager, build system (cbuild) is also excellent. Everything feels modern and clean. Package availability is still limited and you…
Any examples?
Earlier quoted context omitted.
GNU is defined as what it is; that it is an operating system and that it is free software. "What is GNU? GNU is an operating system that is free software—that is, it respects users' freedom." https://www.gnu.org Although GNU stands for "GNU is NOT Unix" that doesn't count as its a boomer hacker meme that started before GNU.
I'll add that most anti-GNU people I know in the BSD space and related projects are not at all against GNU politically speaking or it's ideals or the free software movement. We mostly think that a variety of terrible technical decisions have been made and allowed to metastisize over the years, and that they are largely unfixable so long as Linux remains an operating system devoted primarily to ease of use, popularity…
One thing that possibly comes to mind for me is macOS' launchd, which doesn't seem to give me the same headaches as systemd even though they're kind of the same thing (see also: windows services manager.)
Another is/was clang/llvm vs. gcc, but gcc seems to have improved since clang originally came out. And clang is readily available on linux distros.
BSD pioneered containers with jails (and zones on Solaris), but I have gotten used to (and perhaps come to appreciate) Linux's a-la-carte namespace/cgroups design. I don't really enjoy working with docker and kubernetes though.
One thing I like about FreeBSD is its documentation, which seemed (to me at least) to be reasonably clear, complete and well-organized.
I've liked NetBSD for a variety of reasons (somewhat coherent classic BSD organization, synchronized kernel and userland releases, great multiplatform support, support for classic systems and architectures, pkgsrc which works nicely across platforms and without root access, /usr/games, and just having a different kernel implementation) but I'm not sure I can point to individual utilities being better than their GNU alternatives.