As a long-time Asm programmer who has read a lot of compiler-generated code (usually for RE purposes), I completely sympathise with Linus - despite all the research and improvements, compilers are still very stupid, and will routinely generate code that no sane human would. The widespread notion that compilers can almost always generate better (faster or smaller) code than a human is probably closer to myth than reality. The one thing that is certainly true is they can compile orders of magnitude faster than a human can, and that's usually the main reason for using them.
I think the x86-64 ABI is also partly to blame for this mess - IMHO it's far too complex (e.g. the "red zone") and makes it easy for these bizarre edge-case-bugs to occur.
"Spilling a constant" is something I've never seen before and I'd probably WTF the first time I saw it, but thinking about it more carefully, it might actually be a wise choice in some cases - x86-64 unfortunately allows very few instructions with 64-bit immediate values, and those few instructions are very long (opcode + 8-byte immediate), so it could make sense to store some constant in memory, on the stack, for a while and put it in a register when it's needed with some moves - 64-bit r/m move with 8-bit displacement is 3/4 bytes. Of course that's not what GCC is doing here; that last move instruction is 6 bytes alone (mov + mod/rm + disp32.)
The tone difference is certainly stark! The bug report feels like a process is in play, while the mail has a tone of a rant in a closed room.
I often take it for granted the compiler is always right (and it darn well just about is), but bugs like this look exactly like mis-cast magic to me. Learning how the compiler makes decisions seems like a rabbit hole, but may be fascinating enough to be worth it. (And think about being a maintainer of such a complex beast!)
I think it is more related to the fact that it is an external project (I assume with different community interactions). So all that "Linux being himself" it's not that accurate.
Can anyone give a layperson's explanation of what exactly GCC 4.9 is doing wrong here? I'm not fluent in Assembly and/or compiler internals. What's changed from 4.8?
It is doing two things wrong, it is moving a constant value to the stack and it's doing it before increasing the stack pointer, so it's momentarily outside the stack area. If an interrupt hits during the instructions the stack is broken, the interrupt service routine will increase the stack for its own use and overwrite that value.
See the bug report linked by e40[0] for explanations. Interestingly, the bug actually exists since 4.5.0, it just doesn't trigger because of different optimisations being applied elsewhere.
I've noticed it also has idiotic warnings. I have some code with macrology that generates lexically scoped typedefs, which are sometimes not used. I do not wish to be warned about these, damn it! Who cares about a typedef at lexical scope that is not used? It's compile-time only. What's next; a warning about every file-scope typedef that has not been used? "Hey you included <stddef.h> but never used ptrdiff_t!"
I really enjoyed reading this even I don't understand any of what the bug is actually about. There's something about people who have taken swearing to a virtuosic level.
How the hell is this abusive? Notice how he isn't cursing at any one person? Notice how he points out exactly what gcc is doing wrong? Notice how he made a bug report to get the problem fixed? Notice how no one in that email thread had any problems with his tone?
That is throwing the guy he's responding to and the whole GCC compiler team under the bus with a very thinly veiled personal attack.
If he was just hurling insults at the code, I could defend him, but that would look more like:
"What the 4.9.0 compiler is doing to my scheduler is fucking insane. It can't be spilling constants like this because we can get hit by an interrupt while the stack is insane and then we'll be screwed. And this other shit is worse and just has to fucking go."
Less pointing at the other devs and screaming that they're incompetent and more pointing at the code and screaming that its going to hose the shit out of the scheduler.
It really isn't that big of a deal, even though it does come off as immature.
A professional would handle it like this:
1) He or she wouldn't choose to use such phrasing himself or herself, but also won't shy away from being firm and asking hard questions if they need to be asked.
2) If somebody else chooses such wording, the professional will engage his or her thick skin and ignore it, instead focusing on the parts that actually matter. It's resolving the problem at hand that's truly important, not wasting energy getting worked up over phrasing that somebody chose to use when describing the problem.
That may be so but I'm pretty sure that pretty much every study that's been done on this says insulting people is demotivating and generally destructive behaviour.
Linus can believe what he likes but the evidence contradicts him.
He has not insulted any people though. He has said bad things about the piece of software called gcc-4.9.0, but he has not once in that email adressed a person in a negative way.
Plus, the bug report he filed towards gcc is entirely matter-of-fact and polite.
As I've worked with more and more younger developers ("Millennials" or "hipsters" or whatever you want to call them), I've noticed that many of them consider criticism of somebody's work to be a criticism of the work's creator himself or herself.
I don't really know for sure why this is. Maybe they're more attached to what they've created, seeing it more as a work of art or something like that, instead of just a tool?
"He has not insulted any people though. He has said bad things about the piece of software called gcc-4.9.0..."
A number of people have made that point.
So if I were to say "Hacker News' emotional maturity ranks below that of 12-year-olds, who generally recognize that contempt expressed for an inanimate entity is actually targeted at the persons who created the flaw in question", that would not be taken personally?
Well, the GCC developers that I know or have otherwise worked with at one time or another are mature men in their 30s or 40s, if not older at this point. Being adults and professionals, they accept that they may just accidentally make mistakes when working on rather complex tools. They don't really care how the problems are pointed out, as their interest is in getting the problems fixed, and not taking it personally.
These people are very different from, say, the young Rubyist who will have his feelings hurt when legitimate flaws are pointed out with something he's worked on. The GCC developers won't cry if you report a bug, and they won't deny that it exists to try to save face. Getting the bugs fixed is what matters to them, so they don't waste time allowing their feelings to be hurt by hyperbolic mailing list discussion.
You're obviously implying that the GCC developers know of the Linux kernel mailing list archives, they could potentially read the emails, the content of the emails could potentially hurt their feelings, this could potentially be considered insulting to them, and that makes it wrong, and so on and so forth.
What I'm saying is that the scenario you've built up isn't very realistic at all. The GCC developers I've dealt with would not waste their time getting offended. What was said in the mailing list, and its tone, is pretty much irrelevant. Getting the problem fixed is what matters to them.
Not at all. I was making no positive logical argument.
Merely pointing out the flaw in the one I responded to.
To put it plainly, the post I responded to made the claim: "Linus was nice when he posted it to the GCC list; therefore everything is fine." This argument doesn't hinge on the GCC guys having thick skin but rather on the apparent unlikeliness of their seeing the more offensive messages in the kernel list.
Further, even if we take into account the 'thick skin' argument, it does not excuse the behavior-- bad behavior is bad even if the affected party doesn't think so. This is why often times the State will prosecute domestic violence cases even when the affected spouse/partner does not wish to do so.
> This argument doesn't hinge on the GCC guys having thick skin but rather on the apparent unlikeliness of their seeing the more offensive messages in the kernel list.
I made no such argument; in fact my reply to you specifically recognized the fact that a GCC developer might come across Linus' message on the Linux kernel list. My argument was that such a GCC developer would recognize that the Linux kernel mailing list does not belong to him or his project, it belongs to Linus and his project, and the fact that Linus' tone when he posted to GCC's bug tracker was quite different shows that Linus understands the difference too.
Maybe it would help if you read the conversation again. I'll paste here for convenience:
>>>>>> Linus just being Linus (paraphrased).
>>>>> That may be so but I'm pretty sure that pretty much every study that's been done on this says insulting people is demotivating and generally destructive behaviour.
>>>> He has not insulted any people though. He has said bad things about the piece of software called gcc-4.9.0, but he has not once in that email adressed a person in a negative way.
>>> Uh, do you think GCC just spontaneously sprung forth into being? No, it was created by people, and Linus is indirectly insulting them.
>> As a number of people have already commented, when Linus posted the bug report to the GCC project, he was careful not to insult them.
> Do you think those people exist in some sort of vacuum where they don't have access to (or will become aware of) the kernel mailing list thread?
Did you really intend to say that a person can only be insulted if they are insulted in every place a particular issue is discussed? Or are you saying that if a person is treated kindly in at least one avenue, then it doesn't matter how many others they may be insulted in (i.e., the one kind treatment negates all the other offenses)?
Since you brought it up in your quote, I should clarify that I agree with the person who said that insulting a piece of software is not the same as insulting the person who created it, not even "indirectly".
That said, my own argument had nothing to do with "what counts as an insult", whether Linus being non-insulting on the GCC bug tracker "cancels out" what he said on the kernel mailing list, etc., etc. I was simply pointing out that the fact that Linus was not insulting when he posted the bug on the GCC bug tracker indicates that he had a genuine intention to help get the bug fixed. (The GCC bug tracker thread makes it clear that the GCC developers recognized that too.)
In other words, I (like the other commenters I referred to) was focusing on whether or not Linus was actually trying to help or not, and what actually got accomplished as a result of his raising the issue. You are focusing on what kind of language he used and where, whether it was "kind", whether "kind" language in one place "cancels out" insulting language in another place, etc., regardless of the substance of what does or does not get accomplished. I was not making any claim whatever about the questions you raise.
I think if you read the entire thread it should be clear that the topic of conversation was in fact "the lack of kindness in Mr Torvalds' public persona". It is in fact your attempt to focus on whether Linus was actually "trying to be altruistic" or not which is in fact incongruous (i.e., discontinuous) with the rest of the conversation. One can "try to be altruistic" all they want but if they cannot package it in nice words, then they are merely going to be [rightly] ridiculed for the "attempt". Also, as I pointed out before, the mere fact that Mr Torvalds can get it right in one context but not in another shows that he in fact knows the difference between right and wrong.
> One can "try to be altruistic" all they want but if they cannot package it in nice words, then they are merely going to be [rightly] ridiculed for the "attempt"
But that's not what actually happened, as far as I can see; Linus' post on the GCC bug tracker appears to have gotten the GCC developers to pay attention to the bug, not to ridicule Linus for using nice language on their bug tracker but insulting language on his own mailing list.
> the mere fact that Mr Torvalds can get it right in one context but not in another shows that he in fact knows the difference between right and wrong
I would say it shows that he knows the difference between one context and another. Whether or not that difference is also "the difference between right and wrong" is one of the points at issue; you can't just assume that your personal opinion about that is shared by everybody. As you pointed out in an earlier post, it's up to each of us to decide that for ourselves.
> But that's not what actually happened, as far as I can see; Linus' post on the GCC bug tracker appears to have gotten the GCC developers to pay attention to the bug, not to ridicule Linus for using nice language on their bug tracker but insulting language on his own mailing list.
The 'ridicule' I was referring to is this firestorm of a thread: the mere fact that we are even talking about this on HackerNews (and the wider Internets) and that a large majority of the commenters here agree with me that it was, in fact, deserving of ridicule.
> Whether or not that difference is also "the difference between right and wrong" is one of the points at issue;
Whether one can [in his own reasoning] get all the way to 'wrong' from 'adolescent, silly & childish' is really quite a minor issue to me. But, I think it's pretty obvious to most adults that it was in fact pretty adolescent, silly & childish (i.e., socially under-developed) and that's enough for me.
Further, if I had to choose one form of overly adolescent behavior, it would be the fun type coming from 'brogrammers' and startup culture rather than this trite anti-social and uber-negative poison from 'hardcore Linux(TM) hackers' (but I can certainly understand that this is merely a preference of mine in particular).
> The 'ridicule' I was referring to is this firestorm of a thread
Which is not going to have any effect on the Linux kernel or the GCC compiler. I'm not participating here because I think it will change anything substantive; I'm just doing it because I feel like it.
> a large majority of the commenters here agree with me
Have you actually counted?
> I think it's pretty obvious to most adults that it was in fact pretty adolescent, silly & childish
I think you are overestimating how many "adults" agree with you.
Even if we disagree about the other matters being discussed earlier, I do have to agree with you about the misvoting you've been subjected to.
There is no legitimate reason for your comments to have been voted down.
I'm seeing this sort of blatant and unjustifiable misvoting happening more and more often here. It's really making it harder to read a lot of legitimate content, and it's discouraging participation in otherwise perfectly fine discussion.
For the record, I strongly disagree, but I doubt we're going to resolve that here. I would merely point out that one of the main reasons we have different forums for discussion with different rules is precisely to provide different contexts. Wanting every forum in every context to conform to your preferred rules is ridiculous and unrealistic. Part of the price of living in a free society is that you will sometimes come across things that you find offensive, and you'll just have to deal. Understanding things in their proper context is a big part of dealing.
Well, these were both very public forums. And the fact remains that Linus' behavior on his mailing list (i.e., in the comfort of his own lair) was rather silly and childish. I don't think anyone with his level of experience with large complex systems of software should have a hard time imagining how a bug such as this could be introduced in what has to at this point be considered a program with a life of its own (not that much unlike a machine-generated proof or some other reserve of Big Data(TM)). Yet, instead of jumping to this [obvious] conclusion, he rather chose the vastly more unlikely explanation of incompetence/ willful ignorance/ etc and went on a childish profanity-laced tirade so all of his [equally adolescent] hero-worshippers could get a [cheap] laugh. Also, the fact that there were in fact two very different presentations here from Linus is further proof of the fact that one of them was uncalled for.
Only in the sense that you can view them on the Internet. But there are lots of forums that anyone can view on the Internet, with a wide variety of norms. Just as in the real world, there are lots of "public" places where in theory anyone can show up, with a wide variety of norms. Anyone who doesn't like the norms of a particular place is free not to go there.
> the fact that there were in fact two very different presentations here from Linus is further proof of the fact that one of them was uncalled for
No, it isn't, because, once again, the two presentations were made in different contexts for different purposes. On the GCC bug tracker, Linus' purpose was to inform them of the issue and try to help resolve it. On the kernel mailing list, his purpose was to make sure that the compiler issue did not result in problems with the kernel. The latter purpose may well require a different style of presentation than the former; and since it's Linus' kernel and mailing list, he gets to judge what style is required to accomplish his purpose there.
Looks like you chose to nitpick two minor points of my last message instead of focusing on the single substantial point which still remains.
However, I will still respond to your nitpicks: public is public. If you don't want to be seen as an unsavory character, you would do well to not say mean and nasty things about other human beings in public (and as the Donald Sterling case taught us, perhaps even in private). If you want to be a truly virtuous human being, you will not do so either in public or in private but that is certainly up to you to decide. I think it is the case that a person who bad mouths people in their [perceived] absence but not in their presence is a morally compromised individual but that is for you to decide if you want to keep company with that sort or not. Just be forewarned: if they are bad mouthing other people around you then they are likely doing the same about you in your absence.
My response to your second nitpick is essentially the same as the one above. But, I would still be interested to see if you have any thoughts on the [remaining previously ignored] substantial point in my last message to you.
> the single substantial point which still remains
If you mean the fact that in his LKML post (but not in the GCC bug tracker) Linus attributed the bug to "incompetence/ willful ignorance/ etc" instead of assuming that it just snuck in (I'm not sure I agree with that interpretation, but I'll adopt it here for the sake of discussion), for a bug that's severe enough to make Linus prohibit kernel compiles using the affected version of gcc, I'm not sure what the difference is, practically speaking. Either way, the compiler can't be used with the bug present, so, substantively, it doesn't matter whether the bug just snuck in or the GCC developers wilfully ignored the behavior; that version of gcc can't be used to compile the kernel until the bug is fixed.
As you so aptly pointed out, the GCC maintainers are no fools but rather mature, competent professionals. I think the explanation that it was a simple oversight is far more likely than malice or incompetence.
There are many ways in which you can criticise a project without blaming those involved but as a rule when you see the phrase "We're talking "sloth that was dropped on the head as a baby" level retardation levels", that's not the space you're in.
Yeah, actually he did. He explicitly addressed it as "your compiler" and then proceeded to spew venom. That turns it into an attack against a person and not just code criticism.
(just to play devils advocate)
How do you figure "the evidence contradicts him" ?
His attitude: launched & maintains the largest free collaborative dev known to man (?) which has generated squillions as a by-product.
Yours/My attitude: makes some friends
seems to me he has some runs on the scoreboard (and some facts on the ground)
It'd be a lot healthier if malleable, hero-worshiping younglings could get their heads around the fact that sometimes their flawed, human heroes accomplish their greatness despite their outrageous flaws and not because of their outrageous flaws. Then we might end up with fewer kids affecting the asshole-ness, misogyny, alcoholism, or misanthropy of their idols thinking it will make them better programmers, writers, basketball players, or mathematicians.
But noooo, it's all "Linus did X, so I'll do X and be cool like Linus!"
It would have been more appropriate to have said "cooccurrence is not causation", because there isn't actually a correlation between personality flaws and accomplishment, just some notable cooccurrences.
>Unless you believe correlation is providing evidence of causation here:
In both cases, yes, it is evidence of causation. However the low prior probability of causation in each case massively overwhelms the evidence provided by observing a correlation.
Here's an easy way to tell if A is evidence of B: ask the inverse question - is "not A" evidence of "not B". If so, it is trivially provable that the answer is yes.
I'm very much not "make some friends", I'm much closer to Linus's model by default.
I still remember when Linus and friends couldn't write a VM that worked on a server to save their fucking life and all Linux was good for was running xmms on Linus' desktop, but I was stuck trying to run his garbage in production. What goes around eventually comes around.
And we can either all yell at each other and wind up tuning the fuck out, or we can try to be slightly more civil and realize that everyone is trying to solve hard problem and getting yanked in a dozen different directions. I'm willing to give the GCC compiler team the benefit of the doubt that these bugs were not due to the team being generally incompetent or lazy.
To start with, certain types of insults, when done at the workplace, deny workers their civil rights under various laws, including Title VII. A workplace could be sued for allow a hostile workplace should this occur. Even if it's the most profitable company in the US. Hence the question cannot be framed as a balance between "largest free collaborative dev known to man .. generated squillions" vs. "make some friends".
To have a real discussion then requires a finer nuance of what "insult" means. Verbal abuse includes "scolding, yelling, swearing, blaming, insulting, threatening, demeaning, ridiculing, criticizing, belittling" (see http://journals.psychiatryonline.org/data/Journals/AJP/3768/... ). That paper, which looked at verbal abuse of children, concluded "Parental verbal aggression was a potent form of maltreatment. Exposure to multiple forms of abuse was associated with very large effect sizes [comparable to those associated with witnessing domestic violence or nonfamilial sexual abuse]."
> Obnoxious behaviour has become endemic in the workplace. Some of the worst offences are: not turning off mobile phones in meetings; leaving a jammed printer, gossiping, and snapping at coworkers. Yet, it is this small stuff that relentlessly grinds down collegial working relationships. Incivility is at the low end of the continuum of workplace abuse. Workplace incivility is not violence or harassment or even open conflict, although it can build up to any of those things. Links between the work environment and indicators of employee loyalty, commitment, and productivity show this is not a “fluff” issue. In addition, a recent study on workplace incivility reveals that rude employees and managers can cost a company millions of dollars a year. This paper will look at a definition of incivility in the workplace, the causes of poor behaviour, the costs to organizations, and what employers can do to help.
You may also be interested in "Sticks and Stones The Philosophy of Insults" by
Jerome Neu.
I bring this up to point out you have not succeeded in being a devil's advocate. My response is not part of a dialog testing the original hypothesis but instead is only criticism in the way you structured your question.
I didn't realize certain accomplishments allowed someone to be a total and utter dick - to ignore all decorum, respect, and human kindness to others. The more you know...
That's not what's being discussed. Tyrannosaurs made a descriptive claim, mh_ questioned it. You're acting like mh_ made a normative claim.
(And you're using the shitty rhetorical tactic of sarcasm, which allows you to tell someone that they're wrong without making any claims of your own, or even understanding what they're saying.)
Your adding the modifier "shitty tactic" to the use of sarcasm gives away your inability to understand that tone is notoriously difficult to infer from online comments, and predisposition to try to prove how smart you are. But you're unable to see the irony in placing rules around the decorum of conversation in a topic about the lack of that very thing.
But that's just supposition, not a normative claim.
Edit: I wonder if I'd be down-voted so harshly if I was Linus Torvalds.
I wonder if I'd be down-voted so harshly if I was Linus Torvalds.
From me, the answer is yes, and I hope that is true for others on HN as well. I'm sure it's not always appreciated, but I think it's important for influential people to receive honest feedback. Although in the rare cases when I downvote 'pg' on this board, I do wonder if some 'all-future-downvotes > /dev/null' switch might get flipped. For what it's worth, I downvoted your sarcastic grandparent, and upvoted this somewhat more constructive reply.
I think it was a question of whether it was or not (devil's advocate) but I was questioning, in what I realize now was difficult to discern over written comment, whether gains in management effectiveness was a good enough excuse for abusive tactics.
I can show you evidence (China, etc.) of supreme levels of output in the most abusive of situations. That doesn't mean that society doesn't place value on a balance between productivity and the way we treat fellow human beings.
You're veering wildly off-topic. Tyrannosaurs said that "the evidence" shows that Linus's behavior hurts productivity, and that is what was being disputed here. You might value other things, but that isn't what's being discussed.
The evidence is that the Linux kernel is one of the most successful large-scale coding projects ever. So whatever Linus is doing, it appears to be working.
Indeed, but that's not to say it wouldn't work even better if he behaved in the way that studies of organisational behaviour general show result in higher performance.
Studies of behavior of certain types of organizations, perhaps. I'm extremely skeptical that such studies are applicable to "organizations" like the Linux kernel developers. If you have any specific studies that you think would be applicable, please give some links.
Some open source culture, yes. However there are plenty of friendly open-source communities as well. If you want to contribute to free-software and don't want to interact with a Stallman/Torvalds/Drepper/deRaadt type personality in doing so, that's a perfectly legitimate preference, and you can find lots of projects that fit the bill!
The kernel community is old.. you can't walk in there an expect people to conform to your new-age ideas.
That being said you are more than welcome to start new projects and choose not to include more abrasive people.
This is the beauty of culture, everyones is different.
Interesting analogy, but in a way I don't think you meant. Problem bears are translocated or killed. That's not due to taking things personally but to prevent future bad interactions. If your analogy has merit, shouldn't we do the same (metaphorically) with Torvalds?
Good point. I've been misinterpreted by some pretty emotional interpreters, too.
I'm not going to name and shame the interpreter in question, but the person who originally wrote it was the same person who wrote the compiler in question, Richard M Stallman. ;)
Well, for one thing a C compiler is a considerably more utilitarian creation - "Hey, McDonalds, Big Macs taste like shit".
And of course even purely artistic creations have always been subject to quite blunt appraisals - you won't have to look far to find them about, say, Jackson Pollock's paintings.
If it's your product you take any criticism personally especially if it some hack who messed up and blamed your baby instead of coughing to your cockup - as the tech editor of the guardian did to one of my products!
I initially felt the same, but imagine if you'd spent countless hours getting to the bottom of an obscure bug that turned out to be a completely bone-headed compiler problem outside your control. Sometimes I think Linus goes a bit far, but this time I think he's justified in his venting his irritation
People in IT are assholes. I've only worked in retail, food, biotech and IT but IT has been my least favorite because of my coworkers. I see the shitty attitude is just an in-general IT thing and accepted.
a friend of mine once told me that everyone in Phoenix is an asshole. I laughed and pointed out that there was no way she had met even a significant fraction of people, that human relationships are built with all sorts of biases, and there was no way she could make that assessment without generalizing to the point of uselessness.
she called me an asshole in reply.
you have pretty much the same problem here. you have no reasonable basis upon which to make that statement. it is offensive and generalized to the point of uselessness.
Generalize much? I have worked in different teams.
The "being an-unfriendly-*" problem has often to do with tight timelines and problems/bugs that seem completely weird. Frankly, that can drive you nuts.
It's not really an excuse, but it can probably explain some behavior.
I wouldn't say it's accepted. IT and engineering staff are often given very little respect outside of their own departments in organizations, but they're tolerated.
I hope we see the "shitheads get ahead" mentality give way, or at least recognize that we're first and foremost facilitators of business and that working to earn respect from coworkers is an important part of that.
I've had to deal with bone-headed obscure bugs in the Linux kernel... Strange feature regressions, kernel-bundled drivers that used to work but have mysteriously become broken over a year ago and were never fixed because nobody gives a damn about making older hardware work. Seems unfair to treat someone like shit for a crime Linus must have been guilty of himself at some point. Unless he somehow thinks that Linux is so perfectly tested that no kernel ever gets released with really annoying bug that crash production systems. As far as I can see, the GCC people are fairly responsive, they're going to fix that bug, and that's probably the best one can hope for.
> imagine if you'd spent countless hours getting to the bottom of an obscure bug that turned out to be a completely bone-headed [...] problem outside your control
I imagine everyone on HN has been in this situation many times during their careers.
Yep definitely, but compiler bugs are a particularly painful\opaque breed of problem. In the overwhelming majority of cases the safe\sensible way to debug an issue (if you don't want to turn yourself slowly insane) is to take compiler correctness as a universal constant. When this turns out not to be the case it's infuriating. Perhaps this assumption is made in error but as I said it's a reasonable one to make.
It might be fine for Linus to adopt such a tone, because he knows enough to moderate it up and down according to circumstances. It makes him more expressive. Note that the bug in question really does indicate some serious mistakes in the compiler, with serious consequences, and along with the invective he provides plenty of information. The problem is that hundreds of other kernel developers imitate his most severe tones all the time without such technical detail and even for technical issues where Linus himself wouldn't express disdain that strongly. That lack of "dynamic range" makes them less expressive, so nobody can tell when they're expressing justifiable concern/outrage and when they're just being dicks out of habit.
It's like the second amendment of developer communication. That sort of verbal weaponry should be available, but its use still needs to be justified.
> hundreds of other kernel developers imitate his most severe tones all the time
You are making this up.
If you don't, then you should have no problem producing a handful of examples, because 100s of dev doing this all the time will have LKML and other places overflowing with them.
Fortunately, I don't have to spend part of my vacation wading into that cesspit to satisfy your faux-curiosity. Others have already done so. The canonical example, of course, is Sarah Sharp.
Just look for "crap" or "s--t" or "idiot" or "Viro". Note especially that these are presented as humor, i.e. approvingly and not necessarily showing the worst examples that are just mean without even the saving grace of being slightly funny.
This is not an illusion, not a new or unremarked-upon phenomenon. You're the one making things up when you act as though some reasonable person might believe otherwise.
Since other people have responded to the parent by saying he's making it up... no, he's not. Abusiveness is prevalent in my experience throughout Linux and open source culture. For example, getting flamed for asking reasonable questions at a point where it makes more sense to ask than to keep digging.
Sometimes people deserve it, but often times they just don't. The people doing this are not paragons of judgement like Linus (arguably) is.
>Torvalds doesn't want a "code of conduct," saying that "venting of frustrations and anger is actually necessary, and trying to come up with some ‘code of conduct’ that says that people should be ‘respectful’ and ‘polite’ is just so much crap and bullshit.”
This is interesting when one compares it with other projects such as Django which have gone the whole other way in addressing modern views of how to approach diversity, gender issues, conduct etc.
Arent diversity and gender issues orthogonal to this? You can have room full of diverse women, man, blacks, whites, young, old swearing a lot. You can have young-white-male-only group talking nicely to each other (while still being exclusive club rejecting anyone else).
English is not, as far as I know, Linus's first language. I presume that any "verbal weaponry" he possesses is adopted and adapted from the "greater developer lexicon" (I just made that up) he absorbed through years of mailing-list / IRC exposure.
Yes, good debaters shouldn't resort to emotionally laden language, but the best debaters just ignore it in the first place.
Children in Finland start learning Engish in primary school, and Torvalds has lived in the US for the last 17 years, so I think you greatly over-emphasize mailing-list/IRC exposure.
What you said about "the best debaters just ignore it in the first place" is incorrect. Debates are meant to convince others, so the debater must take the listeners into account. If the listeners can be swayed by the opponent's emotionally laden language - and we are all human - then the debater must not ignore it.
The obvious responses are to use emotionally laden language in kind, or to call out the opponent for dressing bad logic behind good clothing.
I think this is actually a fairly common case. Take a hard-left political movement; generally quite diverse (in particular, socialist movements tend to have very high representation of women) and yet also very shout-y and argumentative.
The opposite also happens; a common complaint of highly reactionary commentators is that people aren't responding 'civilly' to their regressive nonsense about the threat posed by the gays or how women's' place is in the home or whatever, but are instead using rude words and personal attacks.
It takes a certain type of person who has the grown a sense of technical entitlement and the necessary confidence to believe they're right and that they belong, even when Linus is making up a dozen colorful insults for the things they've done. That certain type of person could be anyone, but there is a strong bias towards those who can find role models that resemble themselves, those who were encouraged and had the resources to get into tech at a young age, etc.
His reasons are bullshit. I don't care how frustrated you are, learn to control your emotions. Its very much possible to be direct without being an ass. Probably more effective too.
And there we go, talking shit about someone talking shit, all while complaining the when someone else talks shit it is not acceptable.
> I don't care how frustrated you are, learn to control your emotions. Its very much possible to be direct without being an ass. Probably more effective too.
Well, apply the same logic to yourself before requiring it from someone else.
As in, Torvalds being mean about the compiler, or something else? That seems pretty par for the course in the context of a badly-broken thing that people tend to expect never to be broken.
He isn't a peer though. It's like how the hermit at the top of the mountain will smack the novice martial artist for making little mistakes. That doesn't make the novice resentful. Instead, he's glad for the training.
Non native speakers often don't have the instinctual grasp of slang and swearwords in English and so tend to use the F C and S words a lot without fully understanding the severity of the words.
"seriously crazy shit" seems a bit restrained for Linus though
No, we understand them all right. We are just not generally as prudish as English-speakers. Perhaps it's more a culture difference than a language barrier.
Understanding them is easy, because when used as swear words instead of during discussions about fucking or shitting, they don't mean anything. 'Motherfucker' just doesn't mean anything.
>We are just not generally as prudish as English-speakers. Perhaps it's more a culture difference than a language barrier.
There are a lot of native English speakers who swear a lot; they're generally[1] aware of the impression that they're making when they do it, though.
Maybe this is what you see as prudish: it's a somewhat unusual thing about English that we have taboo words that are taboo no matter what context (and weren't once untaboo words that started to be seen as euphemisms, therefore avoided), rather than just taboo things that you can say or taboo modes of speech when you're talking to a superior or an elder. The history of 'bloody'[2] should disabuse you of any notion that swearing in English has to do with prudery or sex.
Noticing that non-native speakers sometimes don't completely get the effect of English swearing generally comes from knowing non-native English speakers and knowing the impression that they're trying to make, and watching them misjudge the tone that swearing introduces and failing at what they're trying to do.
Like swearing while meeting someone's parents, or loudly within earshot of a lot of kids.
I'm sure Linus probably started his swearing thing by misjudging tone, even though at this point he realizes exactly what he's doing.
[1] I say 'generally', because there are differences between groups of English speakers as to which words are magical, and the frequency of use tolerated.
In case you never noticed, swearing loudly within earshot of kids almost always makes them laugh uproariously. And what could be more uplifting and pleasant to the ear than the sound of the laughter of children?
I have more of a problem with cheerfully lying through your teeth to children about deceptive frauds like Santa Claus and Jesus and the Tooth Fairy and the Stork who brought your little sister.
What's worse: kids learning that adults occasionally swear, or kids learning that adults systematically lie to them?
Kids learning that adults systematically lie to them is probably beneficial and adaptive (which I suspect is why humans feel compelled to lie to kids).
I'd prefer for adults to adapt to not systematically lying to kids, if it's not so much of a disruptive inconvenience to your lifestyle, thank you.
Speak for yourself, if you believe the benefits of lying to your own kids outweigh the benefits of raising children honesty. You can have an honest discussion with your children about how OTHER people will lie to them, without having to lie to them yourself to prove it. It's not something that's hard so hard to prove that you have to betray their trust by doing it yourself. Because if they don't trust you enough to take what you tell them at face value, they already know you're a liar.
I believe the main reason some humans feel compelled to lie to kids is simply because they're compulsive liars.
Yep. Most of us understand them just fine. It's not really difficult to know what are the swear words in a foreign language, and what impact each of them has.
Plus, Linus is no "foreigner" in that sense. He has been speaking english for 30+ years, has had professional correspondence in english for 2+ decades, and has worked in several american companies. And he was located in the states for decades too, if I am not mistaken. And of course he is not some naive foreigner, he is hiqhly intelligent and technical, and can understand such things as the emotional etc impact of words very well (he has also explained hilself, several times).
The difference is, as the parent said, he is not prudish. He enjoys their use.
How do you define "prudish"? Does it have something to do with sex or body functions?
Because if that's what you think most native-English-speakers think of when people use these swear words, you're wrong.
There are plenty of swear words that have nothing to do with sex or body functions that shock people just as much or more than those that do originate with those things.
>We are just not generally as prudish as English-speakers.
So you're speaking on behalf of the entire non-English speaking world?
I think it's important here to distinguish between swearing about something ("aww, fuck") versus swearing at someone ("fuck you"). If you are more inclined to swear about something, then yes, perhaps you can say you're "less prudish".
But if you're more inclined to swear at someone, then it has nothing to do with prudishness and everything to do with aggression and rudeness. Is your culture more aggressive and rude than that of many English-speaking cultures? Some are....
I admit I was deliberately generalising, because so was the parent.
My culture might seem more aggressive and rude, but more often than not Northern/Central Europeans will tend to be less subtle than English speaking people. This can sometime come across as aggressive and perhaps even rude, even if no swear words are used.
As an example: Linus refers to the term 'Management by perkele', a Swedish term referring to Finnish management, that he himself uses. The word »perkele» is generally considered on par or worse than the English term 'fuck' (in the context of a swear word). Effectively, this term could be translated to 'Fucking Management', but whereas in Sweden this term is used rather publicly, it would hardly be conceivable to have such language in an English-speaking environment.
It's not that we are completely casual with swear words, it's just that they usually have a context, like you said; about something. Against someone is generally frowned upon. (Although, the latter reasoning usually have people objectify the person they are really swearing at.)
I do think the foreigness is part of it. Danes in particular treat English swearing much different than Danish swearing. If you used seriously off-color Danish words in a professional environment, people would be shocked; but you can say "fuck" all you want, because it's foreign and doesn't really carry any weight.
> The word »perkele» is generally considered on par or worse than the English term 'fuck' (in the context of a swear word). Effectively, this term could be translated to 'Fucking Management', but whereas in Sweden this term is used rather publicly, it would hardly be conceivable to have such language in an English-speaking environment.
Doesn't that mean precisely that it isn't on a par with or worse than the English term 'fuck'?
>So you're speaking on behalf of the entire non-English speaking world?
Well, you spoke for the non-English speaking world too, and you don't even belong to it, unlike the parent. If anything, he was less paternalistic.
>But if you're more inclined to swear at someone, then it has nothing to do with prudishness and everything to do with aggression and rudeness. Is your culture more aggressive and rude than that of many English-speaking cultures? Some are....
You can be aggresive and rude without ever uttering a swear word. The only thing that makes swear words appear "worse" in this regard is some kind of prudishness.
There is indeed evidence of significant cross-cultural differences in both [1,2,3]. Cursing itself is universal; but its contextual, relative social acceptability as offense and interjection varies quite a bit across societies.
Put in another way: even cursing that you might perceive as self-evidently "aggressive and rude" --in the limbic sense of the word-- might be perceived quite differently by someone from another culture. And viceversa (eg American-style drunken loudness, to think of one).
I don't think it's even an all-English-speaker thing; rude words are sometimes practically used as punctuation in Ireland, Scotland, and parts of England.
Torvalds has earned his privileges as much as any developer alive or dead. In fact, if his abstract power ranks up there with Fortune 500 CEOs and national leaders, I'd say he deserves his privileges more than most of them as well.
He's stated that GCC 4.9.0 is doing some "seriously crazy shit" as a response to a bug report on the Linux kernel mailing list.
He at no point has insulted the GCC developers, the submitter of the bug, or anyone else who regularly reads the LKML.
I'm not sure why you see defending the Linux kernel code against a problem in GCC 4.9.0 (with exact examples of why GCC is breaking the standards) is exercising any privileges, other than those granted by being the creator of Linux.
I personally think that Linux, Git and recently, Subsurface definitely justify his behaviour here. If you want to argue with him, send him an email. He'll be glad to reply. Otherwise, don't complain about his behaviour on HN. It's the internet equivalent of talking about people behind their back.
It's not about whatever privileges he has, but his aptness for possession of those: I claim that he is not a good leader for Linux, acting irresponsibly, humiliatingly, and inconsiderately. He acts like a dictator, spewing derogatory and impolite phrases around, and people he targets do not have any chance to defend themselves as an army of people volunteer to defend him with childish arguments almost immediately, as if they were organised. A project this crucial to our communities functioning cannot be lead and represented by Torvalds, I say. And this shaves 15 karma off of my HN account -- not that I care the slightest but -- as if it was spamming or whatever, and not a criticism of an important figure in tech world.
Actually, the people he targets do have a chance to defend themselves. What many neglect to mention is that he doesn't blow up at everybody, but only at respected developers (kernel or otherwise) who he expects to know better.
Torvalds' management is one of the major reasons for Linux being successful, and sane. Look at the contemporary Linux userspace: it's in an absolute state of chaos, constant reinvention, poorly documented and composed of all sorts of brittle and baroque libraries.
I give you an upvote for reasonably stating your point, and introducing some very valid points.
I agree that his brash tone, as portrayed on HN, is detrimental to the development of the Linux kernel. I would also posit to say that there would be many examples of him encouraging and helping kernel contributors, but they don't make good headlines.
I think that Greg Kroah-Hartman is actually doing a damn fine job as the stable branch maintainer, and anything Linus says should really just be taken as the passionate words of someone who doesn't want to see their creation diminished.
I'm hoping this is at least partly a troll, because it misses an incredibly important fact - Linux belongs to Linus, he owns the trademark, he is the dictator, and he can literally do anything he wants with the project including shutting it down, walking away, and leaving the world to fork it and carry on.
You may well consider it important and fundamental, but that gives you exactly as much right to decide how to run it as any other piece of private property (which is zero, if I'm not being clear). The fact that he chooses to give away the output has no bearing on anything.
I no longer develop any software, I have lost my interest in tech and these bunch of comments are probably the last ones of mine on HN. Also, I do not need Linux or Unix like I needed before, so I won't even install Linux if I get a new laptop. Why would I, you would wonder, bother who runs Linux. The only reason is I unwillingly empathise with people, those who have words to say him but remain silent because they have to make a living and they make it by getting paid to contribute to Linux, and those who have created software that made FOSS possible and have no obligation to bear Linus' filthy language. You can think that I am a troll, and I can think that Linux belongs to Linus as much as the USA belongs to Washington.
It wasn't exactly well defined in the parent post, but I assumed that he was talking about Linus' rude and condescending way of expressing himself as a privilege.
> He's stated that GCC 4.9.0 is doing some "seriously crazy shit" as a response to a bug report on the Linux kernel mailing list.
> He at no point has insulted the GCC developers, the submitter of the bug, or anyone else who regularly reads the LKML.
I'm not sure how his remarks about GCC can not be insulting to its developers, but each to their own. It's hard to deny the condescending tone of lacing your rant with things like "Lookie here," though.
> I'm not sure why you see defending the Linux kernel code against a problem in GCC 4.9.0 (with exact examples of why GCC is breaking the standards) is exercising any privileges, other than those granted by being the creator of Linux.
Unsurprisingly, he goes on to explain the problem matter-of-factly in a way that I'm sure would have gotten his otherwise completely valid point across on its own, but I'm not sure how his disparaging tone and language contributes to this defense.
> I personally think that Linux, Git and recently, Subsurface definitely justify his behaviour here.
I'm not sure how any of those things could possibly justify his tone and language, so I'm going to assume that we aren't talking about the same thing here.
> If you want to argue with him, send him an email. He'll be glad to reply.
I have no interest in arguing with him.
> Otherwise, don't complain about his behaviour on HN. It's the internet equivalent of talking about people behind their back.
I'm not complaining. If you care to read my initial post in this branch again you will notice that I am simply saying that his accomplishments don't justify his manner of speaking. I'm willing to admit that there could be some other factor that does, but I simply can't see the justifiable connection between engineering something like Linux or git and treating people badly, and this is not a criticism towards him as much as it is of the idea that his accomplishments somehow justify this behavior.
Besides, you are talking behind his back as much as I am, which isn't a great concern to me because he is a person of great public interest, posting these things in a public mailing list.
It'd be better if he knew some manners though. I can't believe the amount of people who will most selflessly throw themselves between Linus and criticisms.
Some of us have worked for corporation where all problems were sugar coated in nicely sounding phrases. They got hidden so nobodies feelings are hurt and criticized person did not even recognized there is a problem with his work.
The whole system becomes dysfunctional in its own right.
And some of us have worked for businesses where everyone is so busy being aggressive and showing how much of an alpha male they are that nothing gets done and all the customers are scared off. Not to mention potential employees.
Whether or not Linus admits it, the same thing is happening with the kernel (wrt to contributors, the "employees" of an open source project). I know system developers who could make valuable contributions to the kernel but who refuse to work in that kind of atmosphere. And I can hardly blame them.
So, if we can't work in sublime subtlety, we better become savage barbarians, as if “balance“ was hostile to civilisation. To not insult and to not swear is not to speak in “sugar-coated phrases“.
There is a difference between sugar coating and showing a basic level of respect. Clearly expressing valid criticism without resorting to vitriolic language is something most people will have to work on, but as you say, it's very important for a functioning workplace.
Sure, I'd love to see Linus be more civilized in his criticisms. As it happens, nobody is perfect and I think Linus's imperfections are laughably trivial compared to many other people's.
Breaking out the strong language is a clear way to separate the technicality from the opinion. People respond to taboo words differently than mundane ones.
If he was designing a video game, your assessment would be viable. As it happens though, Torvalds writes code that makes half the world function reliably; some would say his exceptionally direct attitude has served him well in this endeavor.
He writes code that helps most of the servers running important code function, lets remove linux totally from the computing environment and you wouldn't even get a DNS request through let alone a webpage returned.
Professionals come in various shapes. Torvalds is not the most polite man in the worlds, far from that. However, he achieved a lot, whether because or despite of it. So, yeah he reserve respect and deserve to be called professional. Although not to be called "the most polite man".
I'm really baffled about this. Maybe I'm from older hacker generation (and I'm also a Finn). I just did not see personal insults going on. I would like to see more neutral discussion about cultural differences (between hackers and nations) and not opinions and own cultural biases stated as normative.
Here is my personal cultural bias:
1. Insulting people personally is not same thing as insulting their ideas or their actions. If someone insults my actions or my ideas, I may get emotional, but I don't take it as an insult. For me it means that I'm being challenged.
2. Cursing is not unprofessional or insulting in itself. Cursing is to be used for emphasis and to get people involved emotionally (good thing). Cursing has its place to signal frustration and get trough people. Too much cursing is like underlining everything.
3. (Anglo American) business culture is not necessarily something to be emulated. Being overly polite invites all kinds of side stepping, passive aggressive behavior and ineffectiveness. I feel that "being professional" is used here as thought terminating cliche.
4. Some hacker cultures can be both fiercely competitive and cooperative at the same time. Being heavily criticized and challenged is part of that process. Kernel hackers compete to get their ideas incorporated into the same codebase. Javascript hackers borrow snippets from each other and do their own things (no need to be as competitive as forking is easier). This might lead to different hacker cultures. What is accepted in one culture is not accepted in other.
5. Management by perkele[1] is management strategy that can be very effective with the 'Shut Up and Show Them the Code' hacker culture.
6. Pacabel may have a point. Younger generations may no be used to their work or behavior being harshly criticized and take it as personal insult. https://news.ycombinator.com/item?id=8089706
>If someone insults my actions or my ideas, I may get emotional, but I don't take it as an insult.
The fact that people get emotional in response is exactly the problem. Knowing that an insult wasn't technically directed at them doesn't stop them feeling shitty, and being unproductive as a result.
>Younger generations may no be used to their work or behavior being harshly criticized
Younger generations are used to their work being criticised constructively and dispassionately.
>Knowing that an insult wasn't technically directed at them doesn't stop them feeling shitty, and being unproductive as a result.
Again, this is not culturally universal reaction. In some people and cultures it gets your blood flowing and it encourages you to work harder.
>Younger generations are used to their work being criticised constructively and dispassionately.
Is dispassionate interaction something you should automatically want? I also don't agree that with the constructive criticism and passionate delivery are somehow exclusive.
I don't personally use the language that Linus uses, but I strive in the environment where people are very competitive and compete against each others by engaging personally (both negatively and positively). It's inevitable that we associate some part of our ego into our ideas and work and feel a sting when it's criticized. I love it when I'm right and others are wrong and little hurt when I'm wrong and and get called for it. I assume others feel the same. This does not mean that I am personally attacking or hating people. If someone can't cope with it, I have to find other way to communicate with them of course. But if I'm in good competitive alpha-team where nobody is taking it personally I feel that we should be able to keep our culture and not change just like very PC and non-emotional "professional" team culture should not change if they feel they are in good and productive place.
I think that the overly critical culture of many European nations prevents younger people from trying out or voicing ideas because humans are human and don't like being humiliated , especially in public. This is something I think we should not emulate.
>overly critical culture of many European nations prevents younger people from trying out or voicing ideas
I don't think this is what is happening here.
>humans are human and don't like being humiliated , especially in public.
This is just the cultural difference I'm talking about. Someone criticizing your work or behavior does not universally signal humiliating you as a person. I acknowledge that this may be so for you personally or in your culture.
I read the bug report thread and that was as professional as I would expect from grownups. I was responding to your statement about anglo-american culture.
I have observed this "it's OK to be direct" motto from some parts of modern Europe and, interestingly for me, Israel.
Everything we interpret depends on our training, but my observation is that this motto is very commonly interpreted to allow rude, condescending and arrogant behavior. I spent a lot of time in India. In this respect correct behavior (as taught to us by our parents) aligns with the Anglo-American view of politeness in many respects.
I have come to believe that the directness championed by some current European cultures may perhaps be a modern phenomenon, not so common before the 1940s in those countries?
I agree that too much politeness can be perverted into passive aggressive behavior but so can (and I think is) this "Let's be direct" attitude.
There has been much talk on HN and everywhere else of how to develop a code of conduct that allows honest exchange but maintain civility. I think politeness and respect is a central part of this code.
People are not computers. You can not kick them when their output does not please you. You can not kick them when you think you are better than them.
Well, perhaps you can. Perhaps being rude and arrogant leads to a better world.
I would like, however to live in a world where everyone tries to respect each other and show that respect.
nabla9 -> I have a technical question about HM comments. I see two interesting things about your reply
1. I can not see a down-vote button (To clarify I personally never down-vote anything, just upvote things)
2. I don't see a reply link to your comment.
This does not seem to be a nesting limit (I can see a reply link for other comments elsewhere here)
What is this - is there some option one picks when responding to a comment?
The downvote button (and presumably the ability to downvote) is removed for direct replies to comments you have authored. It is also removed for top level comments on articles you have submitted.
My theory about this is that Linus treats his mailing list like a semi-public IRC chatroom (I mean, it is his mailing list for his kernel, so it makes sense) and HNers are not accustomed to that kind of cross-over if they've even ever idled on an irc channel run by hackers. It usually gets much more crass in there, so in that context these kind of comments make sense.
But when you consider Linus is responsible for, I dunno, billions? trillions? in spending on technology based on his OS, it's also easy to understand people getting nervous or offended by IRC-style banter. It's like seeing the CEO of a large company make jokes about retards. Pretty off-putting.
(And no, of course Linus is not a "CEO of Linux" and he doesn't owe anybody anything, i'm just saying I get how people could feel shocked or offended by his behavior)
Seriously. If a community/somebodey cares more about how something is said more than what is said, said community/person lacks the allegedly so important professionalism.
at least the conversation around it is amusing. I may never enjoy anything as much as the convoluted logic of people who want the world to work hard to not personally offend their sensibilities
FWIW, I unflagged this submission again, because amongst the comments there was some interesting discussion, specifically about automatic reduction of test cases and the kinds of tool that are used to do it today.
(I did flag the submission at first, though. Someone being rude isn't particularly interesting, and usually isn't worth bothering to read either since any underlying point worth exploring will invariably be made more concisely and politely elsewhere.)
Linus has earned the right to that tone, care about Linux or not. Now when some unknown emulates it, that's the time to take note (and ignore that person ever after).
This attitude is breeding the jerks and aholes in this industry. It is far too common to encounter teams and communities where in professionalism is accepted on a sliding scale of "accomplishment". If you accept this type of disrespectful/demeaning behavior from one "special" person then you will get it from everyone. Eventually, people tire of this type of nonsense and leave. You may be able to backfill their talent, but regular churn per mantle inhibits the creativity and velocity of the group. Linus is extremely fortunate that folks are paid very well to take it, and many very large companies (e.g. IBM, Red Hat, Intel, etc) have a significant portion of their long term livilihoods dependent on Linux.
Specifically the issue is that if offensiveness is considered a perk of being amazing, then people who think of themselves as amazing are more likely to be offensive to demonstrate that they are amazing. Sounds silly, right? But that is what leads to prima donnas.
Well quoted. It's always fascinated (and worried) me how the human brain allows for the elevation of certain people to God-like status, and then excuses behavior they would normally find abhorrent.
The important thing to note about Erik's behavior is that occurred on Usenet, and unlike the mailing list or places like HN, on Usenet, there were (I have not looked closely at Usenet for about 12 years) no "owners" or moderators for a newsgroup and no way for the newsgroup considered as a "community" or a "collective" to enforce behavioral norms (e.g., around politeness) on individual posters.
Individuals regularly complained about Erik's nastiness, saying (correctly, IMHO) that he was scaring people away from Lisp.
Also, I never saw any of the intense admiration for Erik on comp.lang.lisp when Erik was alive that I have seen on Hacker News. The quality of his Lisp code and the insightfulness of his posts won him respect from many on comp.lang.lisp (specifically, those who accepted that there were almost no limits on speech on Usenet), but he would definitely not have been tolerated for long in most places on the internet where technical subjects are seriously discussed today (particularly HN, which has higher politeness bars than most such places) if he were posting with the same hostility that he did on comp.lang.lisp.
All good points, but I would debate about whether newsgroups would not be considered a "community". A few I'd been involved in in (lord, so many...) years past certainly had all the hallmarks of one. c.l.l was large enough to have perhaps diluted that, but the lack of an owner I don't think is necessary.
But yes, those were different days in different contexts.
You know what, here's a thought. HN is a place that emphasizes civility. Would Linus be allowed to act like that here, or would we say that he's earned the right to act like that here? Double standards suck because something has to break to accommodate them, and often the things that break are the things that everyone else likes when the world is a nicer and more innocent place.
The most striking thing is the number of downvoted gray comments. Probably 10% of them are like that, at the moment!
It kind of gets distracting when so many are like that. With that many being grayed out, I know a lot of them aren't obvious commercial spam or anything like that. So I end up trying to read them all, which of course is awkward because of how they've been grayed out.
Unfortunately the linux kernel still relies on some gnu C extensions for some low-level control of the output. I don't know if comparable stuff exists in llvm and clang, and if so, if linus has any plans on supporting them. However this is just bugs in a new compiler, and the Linux kernel is probably one of those things that hits cornercases.
I'm not sure which C extensions do you mean, but Clang supports most of the same GNU extensions that GCC does. Indeed, it's one of the goals of Clang to be compatible with GCC whenever possible. However, there might be some things that are not (yet?) supported.
And the kernel can be compiled using LLVM and Clang, but not quite in all configurations. But compiling mainline kernel for x86 using LLVM was achieved some years ago.
A lot (if not the majority) of GCC extensions used by Linux exist in GCC because the Linux kernel developers requested them in order to have a great level of control in terms of the generated machine code, so it's not exactly surprising that the Linux kernel relies on a lot of said GCC extensions.
Many of these extensions which the Linux kernel relies on have been added to LLVM, but obviously not all, and from the looks of it some never will, which means either Linux needs to refrain from using said extensions, or building Linux using LLVM will continue to require patching.
The optimiser in LLVM does some even more "seriously crazy shit" - usually revolving around obscure edge cases involving officially undefined behaviour. Linus would definitely not be happy with that.
I think the x86-64 ABI is also partly to blame for this mess - IMHO it's far too complex (e.g. the "red zone") and makes it easy for these bizarre edge-case-bugs to occur.
"Spilling a constant" is something I've never seen before and I'd probably WTF the first time I saw it, but thinking about it more carefully, it might actually be a wise choice in some cases - x86-64 unfortunately allows very few instructions with 64-bit immediate values, and those few instructions are very long (opcode + 8-byte immediate), so it could make sense to store some constant in memory, on the stack, for a while and put it in a register when it's needed with some moves - 64-bit r/m move with 8-bit displacement is 3/4 bytes. Of course that's not what GCC is doing here; that last move instruction is 6 bytes alone (mov + mod/rm + disp32.)