3 ms·

Systemd continues to remind me of windows. It keeps fixing what wasn't broken, and you don't get to say 'no thank you' without a complete OS switch. Yes you ca

Systemd continues to remind me of windows. It keeps fixing what wasn't broken, and you don't get to say 'no thank you' without a complete OS switch. Yes you can argue it's the distros decision, but that ship has sailed and fairly soon there won't even be an effective choice.

dale_glass 6y ago
Seems perfectly useful to me. The stated intention is to use that to provide things like debugging tools in setups that are deliberately minimal by design and don't include any troubleshooting functionality, maybe not even a package manager to install any with. A bit of a niche use perhaps, but seems to be very useful to me. And it's good that it's in systemd because otherwise you'd have to hack the boot process of every distro you might want this in. And I wouldn't say it wasn't broken, it's just a need you personally didn't have. I do actually have this need in a virtualization context, and may well making use of it.
- anotherhue 6y ago
  Would an approach like this not also serve your needs? https://www.unix.com/man-page/FreeBSD/8/rescue/ https://www.unix.com/man-page/FreeBSD/8/rescue/
  - dale_glass 6y ago
    There's other approaches would work for me, yes. I could also download a tar file with what's missing and unpack it over root, I suppose. But then root would need to be read-write, like your solution also requires. Also, FreeBSD wouldn't work. We use many Linux specific features, and I have no interest whatsoever in porting all that stuff to BSD for no gain in functionality. Also, I see no issue to somebody else solving my problems and not needing to write more stuff that's unrelated to the actual purpose of what I'm doing.
Chris2048 6y ago
> you can argue it's the distros decision Is this not a fact, versus a mere argument? Do you think the decision is not based on merit? b/c there's all the effective distros switching, versus you saying nothing is broken - one of you must be wrong? Maybe the Linux ecosystem is broken, you just can't say it because you've already adjusted you expectations, or solidified your use case(s)?
- anotherhue 6y ago
  No other package upgrade causes me to wonder "what will this subsume today?"
  - Chris2048 6y ago
    Few other packages are as fundamental
- lmm 6y ago
  RedHat made the decision based on their commercial interests, not technical merit, and since RedHat de facto controls GNOME anyone who wanted to keep shipping GNOME had to fall into line. Debian were probably the only distro who had enough clout to push back against GNOME, and there was a bitter political fight there (the decision was certainly not based on technical merit, nor did the committee even follow their own rules). Slackware wisely chose to drop GNOME but most distros didn't dare; Ubuntu tried to go their own way for a while but they couldn't maintain everything on their own. Honestly the whole thing made me more or less give up on open-source as a development model. Good programmers are expensive, especially if you want them to do the boring maintenance work, and that funding has to come from somewhere. The "give the program away for free and sell consulting" business model works up to a point, but it distorts maintainers' interests away from what would actually benefit the majority of users. While I do blame RedHat, the real problem is that no-one except RedHat is paying people to work full-time on GNOME or any of dozens of similarly important projects (KDE seems to be mostly funded by the EU, which is a much healthier model, but I don't know if you could make a whole distribution from projects like that).
  - 6y ago
hnarn 6y ago
> Systemd continues to remind me of windows. How is it anything like Windows? You're honestly, without bad faith, comparing it to a closed source operating systems that has telemetry enabled by default and forces you to run system updates? I have no idea what systemd-oomd is, but from a cursory reading of the article I can tell that it's a .service, and from the description that it's a "userspace out-of-memory (OOM) killer". So either this service will (worst-case) be enabled by default, in which case I presume that you can disable it or mask it completely like any service, or (best-case) it will be disabled by default, in which case I really don't understand why it matters. In addition to systemd fully allowing you to disable and re-configure much of its functionality, it's also fully open source. You can re-build it however you want, if you want to use it in another manner. This is what passes for being reminiscent of Windows?
- anotherhue 6y ago
  With windows, (and I suggest systemd) you're either all in or all out. And as the scope continues to increase, you're still either all in or all out. As the distros fall into line, you're either all in or all out. Try and not use systemd, and then tell me it's not a problem. I think it's honestly easier to not use windows at this stage. The BSDs (love them as I do) just don't have the clout or the support for many day to day developer needs.
  - galgalesh 6y ago
    > With windows, (and I suggest systemd) you're either all in or all out. This is not the case with systemd. Although these projects live in the same repository, distributions can pick and choose them depending on your priorities. Granted, this is obviously not something most end users come in contact with and even know since very few people actually build distributions. As an end user, ripping out a core part of a distro to replace it is always a painful experience. Regardless of whether that component lives in the systemd repo or somewhere else. Systemd actually makes ripping out core parts easier, which is why arch adopted it so quickly, but that's another story. You're just throwing vague and incorrect statements out there, sprinkle "Windows" on top in the hopes of eliciting an emotional response.

dale_glass 6y ago

Seems perfectly useful to me. The stated intention is to use that to provide things like debugging tools in setups that are deliberately minimal by design and don't include any troubleshooting functionality, maybe not even a package manager to install any with. A bit of a niche use perhaps, but seems to be very useful to me. And it's good that it's in systemd because otherwise you'd have to hack the boot process of every distro you might want this in. And I wouldn't say it wasn't broken, it's just a need you personally didn't have. I do actually have this need in a virtualization context, and may well making use of it.

anotherhue 6y ago
Would an approach like this not also serve your needs? https://www.unix.com/man-page/FreeBSD/8/rescue/ https://www.unix.com/man-page/FreeBSD/8/rescue/
- dale_glass 6y ago
  There's other approaches would work for me, yes. I could also download a tar file with what's missing and unpack it over root, I suppose. But then root would need to be read-write, like your solution also requires. Also, FreeBSD wouldn't work. We use many Linux specific features, and I have no interest whatsoever in porting all that stuff to BSD for no gain in functionality. Also, I see no issue to somebody else solving my problems and not needing to write more stuff that's unrelated to the actual purpose of what I'm doing.

Chris2048 6y ago

> you can argue it's the distros decision Is this not a fact, versus a mere argument? Do you think the decision is not based on merit? b/c there's all the effective distros switching, versus you saying nothing is broken - one of you must be wrong? Maybe the Linux ecosystem is broken, you just can't say it because you've already adjusted you expectations, or solidified your use case(s)?

anotherhue 6y ago
No other package upgrade causes me to wonder "what will this subsume today?"
- Chris2048 6y ago
  Few other packages are as fundamental
lmm 6y ago
RedHat made the decision based on their commercial interests, not technical merit, and since RedHat de facto controls GNOME anyone who wanted to keep shipping GNOME had to fall into line. Debian were probably the only distro who had enough clout to push back against GNOME, and there was a bitter political fight there (the decision was certainly not based on technical merit, nor did the committee even follow their own rules). Slackware wisely chose to drop GNOME but most distros didn't dare; Ubuntu tried to go their own way for a while but they couldn't maintain everything on their own. Honestly the whole thing made me more or less give up on open-source as a development model. Good programmers are expensive, especially if you want them to do the boring maintenance work, and that funding has to come from somewhere. The "give the program away for free and sell consulting" business model works up to a point, but it distorts maintainers' interests away from what would actually benefit the majority of users. While I do blame RedHat, the real problem is that no-one except RedHat is paying people to work full-time on GNOME or any of dozens of similarly important projects (KDE seems to be mostly funded by the EU, which is a much healthier model, but I don't know if you could make a whole distribution from projects like that).
- 6y ago

hnarn 6y ago

> Systemd continues to remind me of windows. How is it anything like Windows? You're honestly, without bad faith, comparing it to a closed source operating systems that has telemetry enabled by default and forces you to run system updates? I have no idea what systemd-oomd is, but from a cursory reading of the article I can tell that it's a .service, and from the description that it's a "userspace out-of-memory (OOM) killer". So either this service will (worst-case) be enabled by default, in which case I presume that you can disable it or mask it completely like any service, or (best-case) it will be disabled by default, in which case I really don't understand why it matters. In addition to systemd fully allowing you to disable and re-configure much of its functionality, it's also fully open source. You can re-build it however you want, if you want to use it in another manner. This is what passes for being reminiscent of Windows?

anotherhue 6y ago
With windows, (and I suggest systemd) you're either all in or all out. And as the scope continues to increase, you're still either all in or all out. As the distros fall into line, you're either all in or all out. Try and not use systemd, and then tell me it's not a problem. I think it's honestly easier to not use windows at this stage. The BSDs (love them as I do) just don't have the clout or the support for many day to day developer needs.
- galgalesh 6y ago
  > With windows, (and I suggest systemd) you're either all in or all out. This is not the case with systemd. Although these projects live in the same repository, distributions can pick and choose them depending on your priorities. Granted, this is obviously not something most end users come in contact with and even know since very few people actually build distributions. As an end user, ripping out a core part of a distro to replace it is always a painful experience. Regardless of whether that component lives in the systemd repo or somewhere else. Systemd actually makes ripping out core parts easier, which is why arch adopted it so quickly, but that's another story. You're just throwing vague and incorrect statements out there, sprinkle "Windows" on top in the hopes of eliciting an emotional response.