I wasn't debating the relative merits of the different systems. I WAS stating that there was no outcry against their existence or the fact that they got included in major distros. Nobody got angry about their presence.
So why are we angry about SystemD ? Because that anger has nothing whatsoever to do with the init tool that comes with SystemD. It's anger at having other tools tightly coupled to that init tool. Tight coupling is ALWAYS a terrible design. There is no such thing as a GOOD reason to tightly couple things. We didn't resist change - we occasionally chose one tool over another because we preferred some changes more than others. SystemD has actively undermined our ability to do that - which is what we are resisting. You call us resistant to change ? On the contrary -we're resisting stagnation, we're resisting letting anybody get in the WAY of changing whatever we want.
Udev is much older than that. The concept first entered Linux as DevFS which was written by Richard Gooch back in 2001. DevFS was ultimately replaced by udev and one of the primary reasons for moving that filesystem to userspace management was that, unlike a kernel driver, it would make the management system autonomous and easy to swap out with other implementations.
The whole point of udev was to NOT be tightly coupled to anything.
> Proclaiming systemd to only be better for maintainers I never said that. I said it is definitely better for maintainers - but the criteria they use are not the same as users.
>that it couldn't possibly be better for users Nope, never said that either. I never said it couldn't possibly better for users. Just that the adoption by systems does not equate to being better for users. It may be better for some users. What I will say is that it cannot possibly be better for ALL users - because NOTHING can be - no piece of software could ever, possibly, achieve that because users do not all have identical use-cases. The lower-level and more basic the software (and systemD operates at a pretty damn low level) the more impossible it becomes for there ever to be a universally 'better' solution.
>making bizarre analogies between 747s and motorcycles (WTF) There was nothing bizarre about that analogy - though apparently you aren't very bright. The analogy simply demonstrated what I said: there is no such thing as a demonstrably better tool - 'better' is contextual and depends on use-case. There is no one tool that is better for EVERY use case. Hell even at the kernel level no two distro's ship with identical kernels - they all have their own custom patches and configs to fit their target demographic's major use-cases and even then most distro's ship with several different kernel configs to meet more than one likely use-case. Ubuntu for example ships a very different kernel with their server edition compared to their desktop edition.
If there's a strawman here - it's your bizarre misrepresentation of my argument and utterly flagrant lies about what I said.
> and attempting to conclude that because of this it couldn't better Again - I never said that. I proved that there is no such thing as BEING 'better'. It CAN'T be better. It can only better AT something. It is fairly universally 'better' at meeting distro maintainers needs - it is NOT universally better at meeting user's needs - it's EASY to please distro maintainer (a small number of people) - and impossible to better for users (a massive and very diverse set of people with extremely divergent use-cases).
I never claimed that SystemD couldn't be better for some users, I did say (and proved) that it's impossible for it to be better for all users.
Had it stuck to being an init system (you only compared it with other init systems) - this would be more viable, it is possible to create an init system that is very good for a very large percentage of people - which is why there were never very many init systems. A few different ones met almost all needs. But systemD is a LOT more than an init system and NOTHING can be better at ALL the things it does for ALL the many things people may need. When it added logind - the odds of being better for a significant number of people got cut in half, with every additional tool those odds get cut in half again. Now my actual argument is that SystemD destroys the lego-block model that is the foundation of what made unix survive for nearly 60 years across numerous major revolutions. When Unix was invented nobody saw PC's coming - yet unix smoothly transitioned onto them, nobody had seen smartphones coming but unix transition to work there - it runs everything from the largest supercomputer clusters to the smallest embedded devices. It has survived because it's lego-block design of extremely loosely-coupled tools could be rebuilt into whatever the hell you needed with very, very little effort. SystemD takes an ever-larger set of those blocks and replaces them with tightly-coupled tools which cannot be put together in different ways and cannot have individual blocks easily replaced. This definitely makes it impossible for it to be better for users because the ONLY thing that can ever be universally 'better' for users is being able to change the system to suit your needs- no programmer can ever know everybody's best setup but everybody can know their OWN best setup. But even for thos
You clearly have no idea what a strawman argument even is. You took the adoption of systemD by distro maintainers and made a claim (with no evidence) to explain it. I merely pointed out how meaningless your claim is. I didnt misrepresent your argument by focussing on a tiny bit of it. I addressed everything you said. Its not a strawman when your argument really was that weak.
You know it's weird, but there is literally not a single thing on your list that Linux hasn't been used for successfully and doing successfully for the better part of 20 years. None of these are new problems. So since systemD is only 6 years old and most major distro's didn't adopt it until the last 3, I guess all of us were just suffering some mass delusion when we watched all this stuff working beautifully back in 2000 when nobody had ever concevied of systemD - and progressively get better every year since.
Now nobody is saying that there cannot be better solutions for this - what I can tell you is that a better solution CANNOT come from a massive bunch of tightly coupled tools with opague interfaces that are so utterly cross dependent that none of them can run (at least without massive hacks) unless you also run all the others.
The ONLY way to EVER do a good solution - especially at the system level - is to build it out of lots of LOOSELY coupled tiny bits that don't care how you put them together or what you put them together with (including pieces that the creators never knew existed). That design has allowed an OS first compiled in 1969 to scale to the largest supercomputers and the smallest embedded devices alike, to survive 50 years of computing history jumping from platform to platform and architecture to architecture, resilient across one major revolution after the other - because it could adapt to any need and any use-case. Because you never had to redesign it to meet a new challenge, you just had to add a few small tools to the mix, and put the others together in a new way.
The lego-blocks approach is the heart and soul of the unix philosophy - and it's a philosophy worth preserving because that philosophy is literally the ONLY thing that has caused Unix to be the single longest-living architecture in computer history. It's an architecture that's so easy to evolve that no revolution was out of it's reach. From mainframes to PC's to phones - it went where the hardware went and was consistently the most reliable and cheapest and fit-for-purpose answer because it was designed to be easy to rebuild by simply taking the blocks and hooking them up in a new way that Kernighan and Ritchie never imagined.
In other words - everything SystemD is not.
We love doing things in new ways, we love change - but we're GOOD at spotting the difference between progress and regression - and systemd is NOT progress, systemD is doing on Linux the exact same mistakes that every operating system besides unix in history has made. If it remains dominant too long - the outcome will be that Linux goes the way of Multics or VMS because, like those, it will not be able to survive the next revolution.
Better for distro maintainers != better for users or better for Linux.
Better is an ENTIRELY subjective thing. Better at what ? Better *how* ?
Whether something is demonstrably better depends on what your chosen measurements are. That's like saying a Boeing 747 is demonstrably better a motorcycle. Whether or not the statement is true depends entirely on the job description. If the job description is 'ferrying lots of people from coast to coast" then it's true, if the job description is "getting to the other side of town with minimal traffic problems" then it's utterly false.
No systemd is NOT better than anything by many, many measures. The only thing it is consistently better at is making distro maintainers' jobs easier. That's not a bad thing, but it's the wrong metric. Here in my country we have a similar issue in the medical insurance field. The largest local insurer by a long shot is also demonstrably the worst insurer you can have. They frequently refuse to pay claims they are liable for (relying on the imbalance of power their wealth gives them should a client choose to sue). Their customer service is absolutely atrocious.
So how the hell did they get to be the biggest insurer ? Because the deals they offer employers is demonstrably the best in the market. They save employers lots of money, so employers make them the default insurance offered - and employees are stuck with the worst insurance imaginable.
That's pretty much the relationship with systemd and distro-maintainers versus users.
Indeed. We didn't mind simpleinit, or upstart or openRC or slackware's BSD-init - all of these were different init systems in the past. We didn't mount autofs or any of a dozen mount helpers added on top of the unix basic during the years.
To suggest that the opposition to SystemD is generically opposing change is to ignore that the people opposing it have been embracing change in all the areas where it plays for decades and are STILL embracing change in those areas - we're just not embracing THIS change because we believe it's badly designed. Having this many basic tools in a common code-base with massive interdependency that makes it near impossible to swap tools out with other tools or run any of them without running all of them... THAT Is a terrible design.
Hell, we don't even do that on the desktop where it may almost make sense. For over a decade KDE has had performance improvements if you run KDE apps in a KDE desktop - but never, once, did we have a KDE app you couldn't run under Gnome or OpenBox or any other DE you want. The coupling was always weak - use the features when available, don't depend on them. And vice versa - all the apps ran under all the desktops. You didn't struggle to run gimp or libreOffice if you chose KDE as your desktop - despite neither of them being written for it. In fact, there were patches you could install to integrate them better which were entirely optional.
Yes... and ought to. It's a job that absolutely belongs in a desktop environment and absolutely does NOT belong any lower in the stack than that - because there is no universally 'correct' way to deal with removable media - the common pattern is ONLY correct in a D.E. - lower in the stack any number of options could be correct for the use-case up to and including completely preventing the mounting of any removable media for security reasons.
Thats an interesting approach. Though where do you draw the line between publishing company and communications company ? Does a newspaper become a communications company if it published 'letters to the editor'? Can it filter which letters to publish ? And what happens if it is a foreign company with Brazilian users ? Must they forfeit the rights they have at home for visitors from Brazil ?
In principle I am in favour of that approach but it seems rather fraught with legal complexity.
Why is paying for a vote illegal anyway ? It makes no sense. You can buy candidates. You can spend a fortune trying to sway elections. Why should only politicians get to be bribed to support something ? Why not let them buy voters too ?
It would be the greatest piece of entirely voluntary wealth distribution in history. That may just make up for the insanity of who will rule.
On the other hand, if you think that's crazy, welcome to the anti-money-in-politics side of the fence.
If "Will they lie" was the only concern - the amateur would win.
But since the question here is "which person should have the fucking nuclear launch codes" - I will choose to go with the person who isn't a fucking textbook example of antisocial personality disorder ("psychopath" in common parlance).
Now in a sane world - this would be the time to vote third party - unfortunately no third party candidates in the US deserve a vote either. No really. Neither Gary Johnson nor Jill Stein deserves a vote no matter how much you may agree with either of their policies. Because neither of their parties are willing to the hard labour. You want a third party. Get candidates running at ground level. Run for dogcatcher. Run for sheriff. Run for town councilors (not just mayor). Get a few towns and show us how you govern. Then when you have a nice big track record with a few hundred towns - THEN start running people for state government, and not just governor, run for state congress and state senate. And when you have a track record of your policies and governance at state level THEN start running for the US congress and senate. And only then - when you have at least several dozen in each house, and governance records at every level of government from there-on down, THEN you can try running for president and deserve a vote. Then you will BE a major party. The US will actually BE a multi-party democracy. Then when you run for president you can say "Look how our policies worked in Bummsville Idaho. Look what we did to Alabama's unemployment rate", then you'll be a *real* party - then we'll vote for you. If any third party does that - you may have a viable third party candidate for president in 2040-something.
While that may be a good argument - twitter is concerned about their own risk. They were already charged with abetting terrorism in France this year - they were acquitted in the end, but that kind of case is bad PR, costs a lot of money and risks getting a bunch of laws passed you have to comply with. It's cheaper to pro-actively root out users who put your business at risk of such things.
In a way it does - the US constitution is not unique - most countries have very similar laws, for a very good reason, it's just about the only *sane* way freedom of speech can work. Unless you are personally happy to let BOTH the local SJWs AND the local neo-NAZIs spraypaint slogans on your living room wall.
To add to your example. Lets assume the guy then starts beating you, you pull out a gun and shoot him. You claim self defense in court - you will be convicted of murder. Self-defence in almost all jurisdictions on earth is precluded as a defence for 'first aggressor' and no sane judge or jury would NOT consider your words there to make you the 'first aggressor' - after all, they are clear an example of "fighting words".
You should read Karl Popper. Tolerating intolerant views (such as homophobia) is, in fact, a logical fallacy and NOT tolerating them does NOT in any way make you intollerant, you CAN'T be tolerant UNLESS you are refuse to tolerate those views - since tolerating them will lead to the destruction of ALL tolerance.
People are free to believe that what twitter are doing is wrong - but if they are not also inviting everyone to spraypaint slogans on the walls of their living rooms then then they are being hypocrites.
I do not want neo-NAZIs spraypainting swastika's on my living room wall. That's my right as well. There's a reason why free speech laws end where they do - because ANYTHING MORE is automatically an infringement on equally important rights of other people, such as the right to freedom of association and the right to freedom of thought and the right NOT to express something you don't agree with.
Considering how flaky exchange and SQL server is, that's no endorsement at all - after all, when the things it manages are utterly unreliable pieces of shit, whose going to NOTICE if the management tool also sucks ?
But, SUN lost their case - just as Oracle effectively lost this one. They were however a victim in a separate, criminal case by the DOJ - but that was for a totally unrelated crime (antitrust violations).
Now the question becomes - is google perhaps guilty of antitrust violations ? That's actually trickier - since that law is not so straightforward, the exact same behaviour in one context could violate it and in another comply with it. It's only an antitrust violation if the actions decrease competition in the market, in this instance google's actions actually INCREASE competition in the market over-all by any reasonable measure. Chromebooks added a new type of competitor to the PC market, android broke Iphone's near monopoly over the phone market (and made smartphones cheap) etc. etc.
There are plenty of sectors where one could try to bring an antitrust case against google (advertising for one) but java isn't one of them.
Antitrust law isn't copyright law and even though in that instance it involved the same players around the same behaviour it had very different outcomes because these are different laws. You can't sue for antitrust violation - it's a purely criminal law and cases have to be brought by the US government.
Slashdot Mirror
Sounds about right yeah :(
I wasn't debating the relative merits of the different systems. I WAS stating that there was no outcry against their existence or the fact that they got included in major distros. Nobody got angry about their presence.
So why are we angry about SystemD ? Because that anger has nothing whatsoever to do with the init tool that comes with SystemD. It's anger at having other tools tightly coupled to that init tool. Tight coupling is ALWAYS a terrible design. There is no such thing as a GOOD reason to tightly couple things. We didn't resist change - we occasionally chose one tool over another because we preferred some changes more than others. SystemD has actively undermined our ability to do that - which is what we are resisting.
You call us resistant to change ? On the contrary -we're resisting stagnation, we're resisting letting anybody get in the WAY of changing whatever we want.
Udev is much older than that. The concept first entered Linux as DevFS which was written by Richard Gooch back in 2001. DevFS was ultimately replaced by udev and one of the primary reasons for moving that filesystem to userspace management was that, unlike a kernel driver, it would make the management system autonomous and easy to swap out with other implementations.
The whole point of udev was to NOT be tightly coupled to anything.
I live in Cape Town but I'm originally from the North.
I'm not in the US. I was talking about Discovery Health.
Other South Africans here who have ever had dealings with them will very likely agree.
> Proclaiming systemd to only be better for maintainers
I never said that. I said it is definitely better for maintainers - but the criteria they use are not the same as users.
>that it couldn't possibly be better for users
Nope, never said that either. I never said it couldn't possibly better for users. Just that the adoption by systems does not equate to being better for users. It may be better for some users. What I will say is that it cannot possibly be better for ALL users - because NOTHING can be - no piece of software could ever, possibly, achieve that because users do not all have identical use-cases. The lower-level and more basic the software (and systemD operates at a pretty damn low level) the more impossible it becomes for there ever to be a universally 'better' solution.
>making bizarre analogies between 747s and motorcycles (WTF)
There was nothing bizarre about that analogy - though apparently you aren't very bright. The analogy simply demonstrated what I said: there is no such thing as a demonstrably better tool - 'better' is contextual and depends on use-case. There is no one tool that is better for EVERY use case. Hell even at the kernel level no two distro's ship with identical kernels - they all have their own custom patches and configs to fit their target demographic's major use-cases and even then most distro's ship with several different kernel configs to meet more than one likely use-case. Ubuntu for example ships a very different kernel with their server edition compared to their desktop edition.
If there's a strawman here - it's your bizarre misrepresentation of my argument and utterly flagrant lies about what I said.
> and attempting to conclude that because of this it couldn't better
Again - I never said that. I proved that there is no such thing as BEING 'better'. It CAN'T be better. It can only better AT something. It is fairly universally 'better' at meeting distro maintainers needs - it is NOT universally better at meeting user's needs - it's EASY to please distro maintainer (a small number of people) - and impossible to better for users (a massive and very diverse set of people with extremely divergent use-cases).
I never claimed that SystemD couldn't be better for some users, I did say (and proved) that it's impossible for it to be better for all users.
Had it stuck to being an init system (you only compared it with other init systems) - this would be more viable, it is possible to create an init system that is very good for a very large percentage of people - which is why there were never very many init systems. A few different ones met almost all needs. But systemD is a LOT more than an init system and NOTHING can be better at ALL the things it does for ALL the many things people may need. When it added logind - the odds of being better for a significant number of people got cut in half, with every additional tool those odds get cut in half again.
Now my actual argument is that SystemD destroys the lego-block model that is the foundation of what made unix survive for nearly 60 years across numerous major revolutions. When Unix was invented nobody saw PC's coming - yet unix smoothly transitioned onto them, nobody had seen smartphones coming but unix transition to work there - it runs everything from the largest supercomputer clusters to the smallest embedded devices. It has survived because it's lego-block design of extremely loosely-coupled tools could be rebuilt into whatever the hell you needed with very, very little effort. SystemD takes an ever-larger set of those blocks and replaces them with tightly-coupled tools which cannot be put together in different ways and cannot have individual blocks easily replaced.
This definitely makes it impossible for it to be better for users because the ONLY thing that can ever be universally 'better' for users is being able to change the system to suit your needs- no programmer can ever know everybody's best setup but everybody can know their OWN best setup. But even for thos
Says the pug.
You clearly have no idea what a strawman argument even is. You took the adoption of systemD by distro maintainers and made a claim (with no evidence) to explain it.
I merely pointed out how meaningless your claim is.
I didnt misrepresent your argument by focussing on a tiny bit of it. I addressed everything you said.
Its not a strawman when your argument really was that weak.
You know it's weird, but there is literally not a single thing on your list that Linux hasn't been used for successfully and doing successfully for the better part of 20 years. None of these are new problems. So since systemD is only 6 years old and most major distro's didn't adopt it until the last 3, I guess all of us were just suffering some mass delusion when we watched all this stuff working beautifully back in 2000 when nobody had ever concevied of systemD - and progressively get better every year since.
Now nobody is saying that there cannot be better solutions for this - what I can tell you is that a better solution CANNOT come from a massive bunch of tightly coupled tools with opague interfaces that are so utterly cross dependent that none of them can run (at least without massive hacks) unless you also run all the others.
The ONLY way to EVER do a good solution - especially at the system level - is to build it out of lots of LOOSELY coupled tiny bits that don't care how you put them together or what you put them together with (including pieces that the creators never knew existed).
That design has allowed an OS first compiled in 1969 to scale to the largest supercomputers and the smallest embedded devices alike, to survive 50 years of computing history jumping from platform to platform and architecture to architecture, resilient across one major revolution after the other - because it could adapt to any need and any use-case. Because you never had to redesign it to meet a new challenge, you just had to add a few small tools to the mix, and put the others together in a new way.
The lego-blocks approach is the heart and soul of the unix philosophy - and it's a philosophy worth preserving because that philosophy is literally the ONLY thing that has caused Unix to be the single longest-living architecture in computer history. It's an architecture that's so easy to evolve that no revolution was out of it's reach. From mainframes to PC's to phones - it went where the hardware went and was consistently the most reliable and cheapest and fit-for-purpose answer because it was designed to be easy to rebuild by simply taking the blocks and hooking them up in a new way that Kernighan and Ritchie never imagined.
In other words - everything SystemD is not.
We love doing things in new ways, we love change - but we're GOOD at spotting the difference between progress and regression - and systemd is NOT progress, systemD is doing on Linux the exact same mistakes that every operating system besides unix in history has made. If it remains dominant too long - the outcome will be that Linux goes the way of Multics or VMS because, like those, it will not be able to survive the next revolution.
Better for distro maintainers != better for users or better for Linux.
Better is an ENTIRELY subjective thing. Better at what ? Better *how* ?
Whether something is demonstrably better depends on what your chosen measurements are. That's like saying a Boeing 747 is demonstrably better a motorcycle.
Whether or not the statement is true depends entirely on the job description. If the job description is 'ferrying lots of people from coast to coast" then it's true, if the job description is "getting to the other side of town with minimal traffic problems" then it's utterly false.
No systemd is NOT better than anything by many, many measures. The only thing it is consistently better at is making distro maintainers' jobs easier. That's not a bad thing, but it's the wrong metric. Here in my country we have a similar issue in the medical insurance field. The largest local insurer by a long shot is also demonstrably the worst insurer you can have. They frequently refuse to pay claims they are liable for (relying on the imbalance of power their wealth gives them should a client choose to sue). Their customer service is absolutely atrocious.
So how the hell did they get to be the biggest insurer ? Because the deals they offer employers is demonstrably the best in the market. They save employers lots of money, so employers make them the default insurance offered - and employees are stuck with the worst insurance imaginable.
That's pretty much the relationship with systemd and distro-maintainers versus users.
That already happened. SystemD has been using it's own udev replacement for some time now.
Indeed. We didn't mind simpleinit, or upstart or openRC or slackware's BSD-init - all of these were different init systems in the past. We didn't mount autofs or any of a dozen mount helpers added on top of the unix basic during the years.
To suggest that the opposition to SystemD is generically opposing change is to ignore that the people opposing it have been embracing change in all the areas where it plays for decades and are STILL embracing change in those areas - we're just not embracing THIS change because we believe it's badly designed. Having this many basic tools in a common code-base with massive interdependency that makes it near impossible to swap tools out with other tools or run any of them without running all of them... THAT Is a terrible design.
Hell, we don't even do that on the desktop where it may almost make sense. For over a decade KDE has had performance improvements if you run KDE apps in a KDE desktop - but never, once, did we have a KDE app you couldn't run under Gnome or OpenBox or any other DE you want. The coupling was always weak - use the features when available, don't depend on them. And vice versa - all the apps ran under all the desktops. You didn't struggle to run gimp or libreOffice if you chose KDE as your desktop - despite neither of them being written for it. In fact, there were patches you could install to integrate them better which were entirely optional.
That's a good design.
Yes... and ought to. It's a job that absolutely belongs in a desktop environment and absolutely does NOT belong any lower in the stack than that - because there is no universally 'correct' way to deal with removable media - the common pattern is ONLY correct in a D.E. - lower in the stack any number of options could be correct for the use-case up to and including completely preventing the mounting of any removable media for security reasons.
Thats an interesting approach. Though where do you draw the line between publishing company and communications company ? Does a newspaper become a communications company if it published 'letters to the editor'? Can it filter which letters to publish ? And what happens if it is a foreign company with Brazilian users ? Must they forfeit the rights they have at home for visitors from Brazil ?
In principle I am in favour of that approach but it seems rather fraught with legal complexity.
Why is paying for a vote illegal anyway ? It makes no sense. You can buy candidates. You can spend a fortune trying to sway elections.
Why should only politicians get to be bribed to support something ? Why not let them buy voters too ?
It would be the greatest piece of entirely voluntary wealth distribution in history. That may just make up for the insanity of who will rule.
On the other hand, if you think that's crazy, welcome to the anti-money-in-politics side of the fence.
If "Will they lie" was the only concern - the amateur would win.
But since the question here is "which person should have the fucking nuclear launch codes" - I will choose to go with the person who isn't a fucking textbook example of antisocial personality disorder ("psychopath" in common parlance).
Now in a sane world - this would be the time to vote third party - unfortunately no third party candidates in the US deserve a vote either. No really. Neither Gary Johnson nor Jill Stein deserves a vote no matter how much you may agree with either of their policies. Because neither of their parties are willing to the hard labour. You want a third party. Get candidates running at ground level. Run for dogcatcher. Run for sheriff. Run for town councilors (not just mayor). Get a few towns and show us how you govern.
Then when you have a nice big track record with a few hundred towns - THEN start running people for state government, and not just governor, run for state congress and state senate.
And when you have a track record of your policies and governance at state level THEN start running for the US congress and senate.
And only then - when you have at least several dozen in each house, and governance records at every level of government from there-on down, THEN you can try running for president and deserve a vote. Then you will BE a major party. The US will actually BE a multi-party democracy. Then when you run for president you can say "Look how our policies worked in Bummsville Idaho. Look what we did to Alabama's unemployment rate", then you'll be a *real* party - then we'll vote for you.
If any third party does that - you may have a viable third party candidate for president in 2040-something.
While that may be a good argument - twitter is concerned about their own risk. They were already charged with abetting terrorism in France this year - they were acquitted in the end, but that kind of case is bad PR, costs a lot of money and risks getting a bunch of laws passed you have to comply with.
It's cheaper to pro-actively root out users who put your business at risk of such things.
In a way it does - the US constitution is not unique - most countries have very similar laws, for a very good reason, it's just about the only *sane* way freedom of speech can work. Unless you are personally happy to let BOTH the local SJWs AND the local neo-NAZIs spraypaint slogans on your living room wall.
To add to your example. Lets assume the guy then starts beating you, you pull out a gun and shoot him. You claim self defense in court - you will be convicted of murder. Self-defence in almost all jurisdictions on earth is precluded as a defence for 'first aggressor' and no sane judge or jury would NOT consider your words there to make you the 'first aggressor' - after all, they are clear an example of "fighting words".
You should read Karl Popper. Tolerating intolerant views (such as homophobia) is, in fact, a logical fallacy and NOT tolerating them does NOT in any way make you intollerant, you CAN'T be tolerant UNLESS you are refuse to tolerate those views - since tolerating them will lead to the destruction of ALL tolerance.
People are free to believe that what twitter are doing is wrong - but if they are not also inviting everyone to spraypaint slogans on the walls of their living rooms then then they are being hypocrites.
I do not want neo-NAZIs spraypainting swastika's on my living room wall. That's my right as well. There's a reason why free speech laws end where they do - because ANYTHING MORE is automatically an infringement on equally important rights of other people, such as the right to freedom of association and the right to freedom of thought and the right NOT to express something you don't agree with.
Plenty of non-Christian countries had LGBT rights literally thousands of years before white christian Europeans even existed.
Considering how flaky exchange and SQL server is, that's no endorsement at all - after all, when the things it manages are utterly unreliable pieces of shit, whose going to NOTICE if the management tool also sucks ?
Wait... Monica Lewinsky had a kid and he is on slashdot ?
But, SUN lost their case - just as Oracle effectively lost this one.
They were however a victim in a separate, criminal case by the DOJ - but that was for a totally unrelated crime (antitrust violations).
Now the question becomes - is google perhaps guilty of antitrust violations ? That's actually trickier - since that law is not so straightforward, the exact same behaviour in one context could violate it and in another comply with it. It's only an antitrust violation if the actions decrease competition in the market, in this instance google's actions actually INCREASE competition in the market over-all by any reasonable measure.
Chromebooks added a new type of competitor to the PC market, android broke Iphone's near monopoly over the phone market (and made smartphones cheap) etc. etc.
There are plenty of sectors where one could try to bring an antitrust case against google (advertising for one) but java isn't one of them.
Antitrust law isn't copyright law and even though in that instance it involved the same players around the same behaviour it had very different outcomes because these are different laws. You can't sue for antitrust violation - it's a purely criminal law and cases have to be brought by the US government.