4 ms·

> "I'm not going to defend Lennart's approach to community interaction, but what I will say is you have to admire the willpower and determination of somebody wh

> "I'm not going to defend Lennart's approach to community interaction, but what I will say is you have to admire the willpower and determination of somebody who at work one day says 'I'm gonna write my own init system- actually no it's going to be a full on system layer for linux' [...] and then he gets it put into pretty much every major linux distribution." Yeah, I suppose I also admire the willpower and determination of Genghis Khan. Corbet is almost as dismissive as Lennart himself of the life-disruptive bugs real people encounter due to Lennart's approach to software engineering and project management. "It has bugs? It's software" This attitude is positively correlated with the amount of hate systemd (and pulseaudio) have gotten over the years.

HarryHirsch 8y ago
If planes fell out of the sky due to buggy software as often as systemd crashes due to yet another bug, air travel would get much hate, too.
- freddie_mercury 8y ago
  If systemd cost as much as a plane, it would have fewer bugs. We can go back and forth with dumb comparisons all day but I don't really think we're advancing the conversation.
  - peterwwillis 8y ago
    I'm in the dumb comparisons camp, so here's mine: Systemd is communism, BSD init is capitalism.
  - HarryHirsch 8y ago
    I disagree about the value of the conversation. Airplane software is meticulously tested because of the safety implications. If we can't have proper testing for a widely used, sprawling piece of software that runs with root privileges and PID 1 we may be better off without that piece of software. Also this: the equally widespread SQLite seems to be programmed with a much better level of discipline and comes with much better testing. It isn't nearly as bug-ridden.
    - djsumdog 8y ago
      It's a good comparison because sqlite is used in many embedded aviation applications. I do get frustrated with sqlite a lot because it's missing so much of what I expect in SQL (no ALTER tables, having to rewrite a lot of right joins into other types of joins, etc.) but I have never had a situation where I got frustrated with sqlite because it lost my data.
      - adontz 8y ago
        And still there are bugs in SQLite, despite unprecedented testing, best in industry by opinions of many.
        shakna 8y ago
        And SQLite respond to bugs with a lot more professional attitude. They apologise, they fix. They try and ensure it won't happen in future. They do not make excuses. They don't dismiss concerns - they validate them.
        adontz 8y ago
        I really don't think that often hated "Not my problem" kind of answers Pottering gives are that unprofessional, because in 99% of cases, he points to exact party who's problem it is and mostly they are distribution maintainers, which makes sense.
        Karunamon 8y ago
        Except, defaults matter. If you ship something that violates decades of well-known behavior, it is at least very unfair (and probably a lot of other unpleasant adjectives) to expect downstream to enable the —suck-less flag to restore the expected behavior. The friendly and not bull-headed thing to have done was to ship the new behavior, but enable the compatibility flag by default, rather than expect other people to clean up after systemd's messes. This mentality is the main reason bag on systemd all the time. If systemd does something to annoy you, it is automatically your fault for doing something Wrong or not enabling some obscure config file option. It's never systemd's fault when systemd unilaterally violates the principle of least surprise.
  - fiddlerwoaroof 8y ago
    If you add up all the engineering time used diagnosing systemd bugs at companies that use systemd distros, you might discover it’s enough money to run a small airline.
- darkpuma 8y ago
  Exactly. Generally speaking people like things that work well and hate things that don't. Any other justification citing 'philosophies' or similar nonsense are retroactive attempts to cast their feelings in a more intellectual light. When somebody says "I dislike systemd because it violates the Unix philosophy" what they almost certainly actually mean is "systemd has caused me a lot of grief." You don't hear complaints about firefox or chromium violating the so called "Unix philosophy" because generally firefox or chromium will satisfy the overwhelming majority of users. That's why nobody uses Uzbl. This is what systemd advocates (like the article's author) don't seem to get when they point out that systemd is a launchd ripoff so people who like launchd should like systemd as well. People like launchd because launchd doesn't cause them grief. People dislike systemd because systemd does. Design philosophies have jack shit to do with it. Anyway the only real 'philosophy' Unix ever had was "worse is better" and systemd certainly seems to pay homage to that.
agumonkey 8y ago
PID HUN
JdeBP 8y ago
You have erroneously put Benno Rice's words into Jonathan Corbet's mouth.

HarryHirsch 8y ago

If planes fell out of the sky due to buggy software as often as systemd crashes due to yet another bug, air travel would get much hate, too.

freddie_mercury 8y ago
If systemd cost as much as a plane, it would have fewer bugs. We can go back and forth with dumb comparisons all day but I don't really think we're advancing the conversation.
- peterwwillis 8y ago
  I'm in the dumb comparisons camp, so here's mine: Systemd is communism, BSD init is capitalism.
- HarryHirsch 8y ago
  I disagree about the value of the conversation. Airplane software is meticulously tested because of the safety implications. If we can't have proper testing for a widely used, sprawling piece of software that runs with root privileges and PID 1 we may be better off without that piece of software. Also this: the equally widespread SQLite seems to be programmed with a much better level of discipline and comes with much better testing. It isn't nearly as bug-ridden.
  - djsumdog 8y ago
    It's a good comparison because sqlite is used in many embedded aviation applications. I do get frustrated with sqlite a lot because it's missing so much of what I expect in SQL (no ALTER tables, having to rewrite a lot of right joins into other types of joins, etc.) but I have never had a situation where I got frustrated with sqlite because it lost my data.
    - adontz 8y ago
      And still there are bugs in SQLite, despite unprecedented testing, best in industry by opinions of many.
      - shakna 8y ago
        And SQLite respond to bugs with a lot more professional attitude. They apologise, they fix. They try and ensure it won't happen in future. They do not make excuses. They don't dismiss concerns - they validate them.
        adontz 8y ago
        I really don't think that often hated "Not my problem" kind of answers Pottering gives are that unprofessional, because in 99% of cases, he points to exact party who's problem it is and mostly they are distribution maintainers, which makes sense.
        Karunamon 8y ago
        Except, defaults matter. If you ship something that violates decades of well-known behavior, it is at least very unfair (and probably a lot of other unpleasant adjectives) to expect downstream to enable the —suck-less flag to restore the expected behavior. The friendly and not bull-headed thing to have done was to ship the new behavior, but enable the compatibility flag by default, rather than expect other people to clean up after systemd's messes. This mentality is the main reason bag on systemd all the time. If systemd does something to annoy you, it is automatically your fault for doing something Wrong or not enabling some obscure config file option. It's never systemd's fault when systemd unilaterally violates the principle of least surprise.
- fiddlerwoaroof 8y ago
  If you add up all the engineering time used diagnosing systemd bugs at companies that use systemd distros, you might discover it’s enough money to run a small airline.
darkpuma 8y ago
Exactly. Generally speaking people like things that work well and hate things that don't. Any other justification citing 'philosophies' or similar nonsense are retroactive attempts to cast their feelings in a more intellectual light. When somebody says "I dislike systemd because it violates the Unix philosophy" what they almost certainly actually mean is "systemd has caused me a lot of grief." You don't hear complaints about firefox or chromium violating the so called "Unix philosophy" because generally firefox or chromium will satisfy the overwhelming majority of users. That's why nobody uses Uzbl. This is what systemd advocates (like the article's author) don't seem to get when they point out that systemd is a launchd ripoff so people who like launchd should like systemd as well. People like launchd because launchd doesn't cause them grief. People dislike systemd because systemd does. Design philosophies have jack shit to do with it. Anyway the only real 'philosophy' Unix ever had was "worse is better" and systemd certainly seems to pay homage to that.

agumonkey 8y ago

PID HUN

JdeBP 8y ago

You have erroneously put Benno Rice's words into Jonathan Corbet's mouth.