!!!, you should really invest a minimum of time into formating your posts, i almost missed this:
Quote:
and this: Quote:
Subtitle: "Lennart Poettering would love it!" |
Further reading for a great time and fun:
http://suckless.org/sucks/systemd |
https://encrypted-tbn0.gstatic.com/i...d7Q0-6ZTYTJWuA
edit: i should reintroduce full quoting. the previous post was MUCH longer, with loads of unsubstantiated and even plain false statements, hence my reaction, trying to calm myself with a nice green colour and thoughts of the queen of england. |
How bout nothing ever braking‽ https://www.linuxquestions.org/quest...ml#post5708564 :eek:
|
Sorry if that's too old a thread to "bump".
I'm just an average ex-windowser for 10 years or so, and I don't know the nitty-gritty things os software design, development and sysadministration, but the notion of "do one thing and do it well" has a tremendous intuitive appeal. I can't believe that a "pro" on systemd at slant.co is "do one thing and do it well." Does "everything" counts as "one thing"? And "well..." counts as "well"? So I was wondering if is anyone (like FOSS councils and the like) considering to slice the systemd code into "minimal" subcomponents, as a way to perhaps address the criticism that it's too gargantuan and doing too many things, becoming a huge, unavoidable dependence for some other software. Instead of having all these things in a single binary, it could be systemd-core, systemd-cron, systemd-whatever and so on. Would something like that help making systemd more easily optional, giving more freedom and room for other init systems to evolve in parallel (kind of akin to different desktop environments), as well as improving systemd itself? Does not doing that presents any advantage other than the glory of conquering the init monopoly in linux? Is the systemd development a "member" of some sort of "FOSS council"/whatever that commits to adhere to some democratic decisions? If such thing exists, rather than being just, "hey, this is what we think it would be better for everyone, but no one need to commit to anything". But, either way, perhaps systemd may be involved in some of those non-binding councils, which could eventualy come to the conclusion that splitting the software would be better. Then they'd be kind of jerks if they'd just ignore it. |
The opinion\philosophy
Quote:
...can walking be one thing done well, won't you hit a wall or be by a car if not also looking‽ |
Linux runs the whole OS and often with blobs too... grumble, grumble and keep that GUI off my lawn!
|
Quote:
Hence, you can be running "walk" and "look" at the same time, which in turn may communicate with "walk" and call any other task as needed. And each task can be more or less highly specialized, which may allow its code to be clearer and more easily debugged, or replaced. It also allows users or sysadmins to play with "walkthisway" instead, while keeping the same "look" application at the same time. Or vice-versa. There's no need for a "do-absolutely-everything" singular program. Likewise, there's no need for systemd to eventually merge with xorg/wayland and provide a graphic environment, or to replace grub, in the other end (not that they're actually planning it). They still can do it, but it could be through a separate "sd-xorg", or "sd-grub", each focused in doing "one thing". I'm not sure that even really applies that much to the systemd debate that much, that's not something I really have any knowledge about. I just feel kind of lost with it. My PC was simply not shutting down for a while, unless I used the sysreq "trick" or the even more brutish "poweroff -f". I just had no clue where to look, and google wasn't helping. Whereas with sysv, at least, there's this feeling I could "hack" into the shutdown script and adapt it to whatever mess I may have done somewheere else, making it work. Fortunately that wasn't even needed, as switching back to sysv just fixed it out of the box. |
Quote:
For me is opinions\philosophy versus (on my Stretch\Sid side lightning-fast, stable(—tho "not supposed to be"(;))?):eek:) reality‽ |
Quote:
Code:
$ ls /lib/systemd/ You seem to want software designed by a committee*? I'm not really convinced there's any one good methodology to producing an optimal outcome in software. They probably all have trade-offs. Though, in a way, there kind of is a committee: the rest of the software in the world. For a while now, some systemd people want daemons like tmux to add systemd-specific code so that they don't get killed when a user logs out**. Unix has had a way of doing this for a really long time, so the tmux and other developers are totally in the right to push back against this, keep filing bugs, and/or turn that stuff off at the distro level by either config options or patches. *What a ridiculous word. Look at all those extra, bureaucratic letters. :p ** https://github.com/tmux/tmux/issues/428 Quote:
|
Quote:
Quote:
At the same time I of course empathize with the concern of those to whom those things could also mean practical troubles. And with those who actually are closer to the actual improvements side of things, apparently developers like systemd's APIs better than whatever they had before. In brief, I was wondering if it there couldn't be a "middle ground", or "the best of both worlds", or "most of the goods of both worlds", possibly a less buggy middle ground. |
|
I'm *absolutely* fine with using systemd, I just don't like it.
|
I think fragmentation will persist, no matter what the "major distributions" demand.
Furthermore, a good architect can easily make every part of the building accessible, it's just that some architects act like wardens and build cell blocks where houses used to be. A secure binary registry to replace the evil plain text, if you will. To protect you from freedom, or something. Just hang this pretty poster over these steel bars and it's just like before, they said. It's not that I hate systemd, I'm just disgusted by it. |
Bitching about the software and not the hardware is stupid! http://www.techrepublic.com/blog/lin...-linux-kernel/
|
All times are GMT -5. The time now is 10:17 AM. |