Tweet
Originally posted by MoonMoon
View Post
-
Agree, I was wondering whether Ian Jackson is a shill working for somebody because iirc in the past he's also been so over the top against systemd (without having any big valid complaitn against it) that I'm guessing he's either mentally unstable or waging the ani-systemd campaign for money. -
Justus Winter (Debian Gnu/Hurd) thinks that systemd is a lot more portable than sysv-init. So, soon the systemd-haters will be just left with BSD on the list(which they dont use themselves anyway). When BSD moves with launchd-port, this is going to be even more amusing.
Lets get all non-rational systemd haters ("unix way", "its conspiracy","it breaks") and make them write own system! This will at least make them productive instead of destructive.Last edited by brosis; 20 October 2014, 12:17 PM.Comment
-
Well, he could just be one of those people who get butthurt of being told they're wrong and do anything in their power to manipulate reality to make it look afterwards like they won. I wouldn't call such a person mentally unstable but they definitely do not belong in a decision-making committee. Sadly in democratic systems that's typically exactly where you find such people because most people don't want to spend their time arguing in committeesOriginally posted by mark45 View PostComment
-
I guess it's time to make the Debian GNU/kFreeBSD default for these guys!Comment
-
That is good news, however, that might not work out too well, if systemd developers intend to keep the same attitude as when they rejected patches that made systemd work on non-GNU libc versions. Software can only be portable if it is allowed to be ported.Originally posted by brosis View Post
There is no evidence that Ian Jackson is in any way involved with that website. Also what Ian proposed is a very strict and technically limited proposal of maintaining init-system-neutrality in Debian, there was nothing specifically against systemd, and he did not propose to revert the decision about systemd being the default. In addition he has substantiated his complaints with facts and specific bug reports. Please educate yourself and avoid attacking people personally.Agree, I was wondering whether Ian Jackson is a shill working for somebody because iirc in the past he's also been so over the top against systemd (without having any big valid complaitn against it) that I'm guessing he's either mentally unstable or waging the ani-systemd campaign for money.Comment
-
I've hated systemd thinking it was totally intrusive and reinventing the wheel in a very bad way until I actually learned what it does. The problem of the "unix way" today is that we have a myriad of little tools that each do one thing well, but they don't talk to each other, there is no integration. It's a huge mess of many tools doing the same or similar thing because the devs forked, wrote their own instead of coming together and improving one. For example, count how many "whois" packages are there in your distro. Which one do you use? Why is there so many packages doing WHOIS lookups?!
That was the whole presentation of that Datenwolf guy - the lack of integration in GNU/Linux distributions and senseles layering. And despite Lennart being the douche he was, he actually confirmed Datenwolf's objections. There is zero integration that's why ConsoleKit was invented. That's why PolicyKit and I don't know what other abominations were invented. It is beside the point that the were poor implementations, the point is there is no integration. The problem with that is the the "community" can't agree on how to do it and individual implementations are therefore poorly done. There will always be a group that hates one implementation and forks into another. That's the doom of the GNU/Linux ecosystem. The "fuck you, I'll just fork your code and do it my way" mentality.
The binary logging thing? The corruption is a very bad thing, and Lennart and company are jackasses for marking it NOTABUG, but indexing is very cool thing. It provides a unified indexed logging interface, you don't have to guess grep patterns, and are not subject to localization. You are also not required to grep across rotated log files. You can query by date, by date range, by severity, process name, and many many more.
Let's see someone write a tool to colorize and index through log files like that. It would be broken by design because there is no single unified format for logging, no API to write against. Until now, that is.
The systemd-hostnamed? Initially I was thinking why the hell do we need a daemon to tell us the hostname! Well... the problem is that different distros specify the hostname in different ways. No, it's not always in /etc/hostname. Also, how do you set up the system hostname? There are three steps, ffs! you update /etc/hostname or whatever in your distro. You also have to supply the ip-based hostname in /etc/hosts and you have to use the hostname command and/or restart services to pick up the new name.... With a systemd API call, the software would just have to issue one API call to get it or set it.
The same total lack of integration is the reason why so many tools are being reinvented and tightly coupled with systemd. Yes, systemd is an intrusive layer between the kernel and the user apps, and that sucks and it is buggy, broken, etc... but the alternative is a set of tools that are not integrated at all.
Now that I understand how it works, my only problem is that I think it is too young. They're constantly adding new features and it will take a long while before it stabilizes and is production-ready for sensitive systems.
No, I'm not a systemd zealot. I don't care which packge is it, who does it and how it is called. I just think that in 2014 we need way more than a bunch of shell scripts with no higher-level semantics to run our system. It also does not have to be a single package. I'm all for writing a thin layer hub that will integrate exsting tools. But the problem is, there is no unified API to do that, as I said. And nobody dared to venture there except Poettering and company.
Meanwhile, to all of you lamenting about "this is not the unix way". You know what else isn't? ZFS and yet it is the hallmark of the *BSD world. It is a bunch of functions that exist independently but all rolled into one monolithic binary. It does RAID, it does volume management, it does snapshotting, it's a filesystem in traditional sense, it does backup, .... How unix-y! Linux Is Not UniX. If you want UNIX, then please, do use UNIX and leave GNU/Linux alone.
Oh one more thing. I have no idea who is behind the Debian Fork movement. I just know the old Roman "divide et impera" (divide and conquer). Whoever is doing this is working the GNU/Linux community against itself. I wonder who could benefit from that.Comment
-
So long one offers a working patch and agrees to support it in long run? Just the way its done in Linux kernel. They reject patches for a reason.Originally posted by aigarius View Post
They also dont disallow forking.Comment
-
That's a rather dumb argument. I expected something more technical, but I guess it's not technical at all. sysvinit is worse than systemd (and upstart), so saying that it's needed since it "preserves Unix philosophy" is just weird. May be these folks should go to using the original release of Unix itself or something close to it? It for sure should preserve that philosophy the best."We don't want to be forced to use systemd in substitution to the traditional UNIX sysvinit init, because systemd betrays the UNIX philosophy.Comment
-
-
I see you're still providing this useless link written by some idiot. Linux is more ready for desktop than Windows ever was.Originally posted by birdie View PostComment
Comment