coding
Neither of these guys appear to be able to code proper English.
Never one to mince words, Linux kernel chief Linus Torvalds has once again handed a verbal smackdown to a Linux developer, this time for failing to address a serious bug that could prevent systems from booting. The target of Torvald's latest tirade was Kay Sievers, one of the key developers of systemd, a system-management …
My first language is Danish and I've RTFM for English. Swearing is inappropriate in a public venue, and the first-person pronoun has been capitalized since about the 15th century. I'm not perfect either, but anybody that doesn't know those two points is either not trying, or just plain rude. And don't try to tell me the errors are typos.
My first language is English. In fact, my only language is English.
Unless you are correcting someones term paper or something formal, for most of us you are just considered and asshole for going around correcting peoples English.
Communication is transmission of meaning and sometimes swearing is the only way to get that job done. It is not appropriate in some settings but in others it is, especially if it is the only way to get some shit head to listen or sometime to just shut the f__K up.
I am well published and have been researching, writing and doing formal presentations for the past few decades but believe me I have no idea what "first-person pronoun" is. I can guarantee you most us "native english speakers or not" do not give a sh about formal English. Generally with practice it just works, it is what it is because it sounds ok or looks ok.
Are you sure your first language is English? It reads more as American. Americans do have a rather loose grasp of English and manners at times. "English spoken by foreigners" comes to mind. Swearing is often just an expression of limited competence. Like good manners, why should good language be restricted to formal occasions as if it is a foreign concept to the writer?
If you do not know what the first person pronoun is, I doubt that you are a Professional writer or even frequent presenter. If language is a tool of your trade, you should know some basic grammar. Pronouns are fairly basic.
As for rudeness and swearing, having lived in several countries, two of which required and one that now still requires me to use both English and the local language, swearing and rudeness are bad form in both formal and informal writing in most cultures and environments. Swearing is often just an expression of limited competence, imagination and cliche. Like good manners, why should good language be restricted to formal occasions as if it is a foreign concept to the writer? Even you do not care to write the word, "fuck", in full.
Also, as the hallowed Linus Torvalds is writing in a forum that gets rather a wider audience than just the object of his displeasure, he is writing in public and ought to show the respect and manners that he probably expects from others.
> If you do not know what the first person pronoun is, [ ...] If language is a tool of your trade, you should know some basic grammar
No (kind of), yes (kind of), you need to be clear about definitions and here you're conflating two things.
A formal understanding of language is entirely separate from the instinctive grasp necessary for use. I have no formal understanding; I don't know how to parse a sentence and label its parts. I don't know what a pronoun is, never mind the first person type. Or adjectives, or adverbs or gerunds or...
I do have the strong innate understanding that allows me to use english as well as, or better than, many. Look over my previous posts. Therefore I'd say this demonstrates that a formal understanding is unnecessary. Honestly, what would I gain from it?
Notes.
1) english is my first and (regrettably) only language
2) The spelling of english uncapitalised is by choice.
3) Transmission of a clear message is far more important than the minutiae of precisely 'correct' spelling (whatever that is), but the former is never so bitched over as the latter. When some commenter says "you could have said that in half the number of words" instead of picking on the greengrocer's apostrophe, I'll cheer.
> My language
Yours? Got a receipt?
> all the while ignoring its larger point I'm making.
When presenting oneself as some sort of linguistic expert, it would almost certainly be better not to have made quite such a ballsup of said expertise...
Vic.
> Yours? Got a receipt?
Très glib. It's mine by use, as it is yours by use.
> some sort of linguistic expert
I did not claim this, I claim I can use english competently. I'm not e.g. David Crystal who has earned that title.
> quite such a ballsup of said expertise
A disagreement between us does not necessarily comprise a ballsup on my part. It may, and you're free to show where I failed, but I've put my point as cogently as possible. I'd like this discussion to be constructive, please.
> It's mine by use, as it is yours by use.
It is neither mine nor yours byu use. It only holds value when everyone agrees on what sounds or markings actually mean. Thus your attempt to use it in ways other than the accepted norm serve only to devalue the language and demonstrate your own lack of capability. "I does it different like" is utter bullshit.
> A disagreement between us does not necessarily comprise a ballsup on my part
Indeed it does not - but your hopeless grammar certainly demonstrates a lack of that "innate" understanding you claim to have.
Vic.
@Vic
You're more correct than you let on. Disagreement between two or more parties on the meaning of a given sound or marking is what all civil wars/internal uprisings are fought over and most international conflicts as well. Law! Law is 99.990% comprised of nothing more than providing evidence of historical acceptance of the meaning of a term within a given context.
It is just extraordinarily silly for someone to argue that developing their own rules for the use of a language is a valid use of that language if the target audience doesn't also use the same rules. Things like 'inalienable Hunan rights' are a fine thing, until some guy decides you aren't Human, or only three-fifths of one. Likewise with otherwise simple words like 'perform' and 'work'. 'Performing' the wrong kind of 'work' or at the wrong time, can get you anything from a union grievance filed against you in a factory or get you nailed to a tree if you're a Jew in Roman Occupied Israel.
So yeah, sounds, markings and agreement on their use and meaning are rather important things. You can't just go mixing them up as you please until you've first demonstrated a comprehensive understanding of the accepted ways of their use. You've got to have A Portrait of the Artist as a Young Man before you can have a Finnegan's Wake. Otherwise you're just a loon :)
@Don Jefe, @Vic
It seems some people are really struggling to get what I say. It should be clear enough.
> It is just extraordinarily silly for someone to argue that developing their own rules for the use of a language is a valid use of that language if the target audience doesn't also use the same rules
This is completely true and I fully endorse it, however it has no relationship to what I originally said, that one could grasp and apply the informal rules of natural language without being taught the formal rules of the language.
This does not relate to mutating the rules of the language. I mentioned not capitalising some words to make clear it was by personal choice not typo, so you wouldn't get all smug on me. If that single item has managed to distract you from my main point, that is depressing.
> You can't just go mixing them up as you please until you've first demonstrated a comprehensive understanding of the accepted ways of their use
I thought I had.
> It seems some people are really struggling to get what I say
We're not - it's just that what you say is wrong.
Many people are wrong much of the time - and that's just fine. But you set yourself up as being some sort of expert, whereas your knowledge of language is actually insufficiently thorough even to understand how wrong you are. So you get called on it...
Vic.
> it's just that what you say is wrong
just saying 'you're wrong' repeatedly doesn't advance the debate.
> But you set yourself up as being some sort of expert
likewise your inability to read my prior post disclaiming exactly this.
> just saying 'you're wrong' repeatedly doesn't advance the debate.
I quoted the bit you got wrong. Feel free to go back and look at it. If you can't work out why it's wrong - I suggest you read a grammar. It really ought to scream at you...
> likewise your inability to read my prior post disclaiming exactly this.
Oh., I read that post - as well as the other one where you claimed exactly what you disclaimed :-
I do have the strong innate understanding that allows me to use english as well as, or better than, many.
That's the nice thing about the written word - it tends to leave a trace...
Vic.
> I quoted the bit you got wrong.
what, this bit?
"
> It seems some people are really struggling to get what I say
We're not - it's just that what you say is wrong.
"
Erm, can you even distinguish between a formal and an informal understanding?
And me saying "I do have the strong innate understanding that allows me to use english as well as, or better than, many." simply is me claiming that I'm good with it, but that's not expertise. I'm not an expert. Now I've said it explicitly twice.
You do seem to be reading into things what you wish.
@BlueGreen
You're right, I did ignore your larger point - as it was nonsense. Not to mention your claim that it is your language is demonstrably false.
Every language has rules. Without those rules, you don't have an effective or usable language. Yes, there are limits to what every day people need to know to use the language but people still understand the rules, generally. Its why we teach them in school.
Your basic understanding of English does appear to be lacking - to the extent where your message is warped and difficult to understand.
Think of it this way - web browsers use formal specifications in order to interpret web pages. However, they have an issue when they come across a web page which doesn't comply with those rules. So, they then have to basically guess what to do. You then end up with the things like the Internet Explorer box model problem, or you end up with a completely malformed page displayed to the user.
If all web designers and web browser makers followed the specifications properly, that would never happen and everyone would get to see what the designer intended without interpretation. The same concept applies to the English (and other) languages. Without following the rules, people have to engage in a lot of guesswork to figure out what you're saying and often, those guesses end up warping the message.
> Every language has rules
Exactly! I agree totally.
There are formal rules about sentence structure etc. which are taught in schools, and there's the innate grasp which comes from exposure to the language. I have the latter only, which includes an informal but still strong set of rules, which embody the formal rules, and go much further.
The idea that one can't use language unless it's taught in class is as bizarre as saying one picks up the meanings of words only from a dictionary.
> Think of it this way - web browsers use formal specifications in order to interpret web pages.
There's a fundamental difference between natural language and formal languages. Formal languages are for a limited domain and *require* an unambiguous definition. English is *not* a formally defined language. If you don't realise the difference, you're going to struggle in IT (Incidentally I do have a background in formal semantics though I've forgotten most of it).
> to the extent where your message is warped and difficult to understand.
Seriously, what in my original post was 'difficult to understand'?
> Every language has rules. Without those rules, you don't have an effective or usable language.
Unlike French (Académie française), Italian (Accademia della Crusca), Spanish (Real Academia Española), German (Gesellschaft für deutsche Sprache) there is no central authority for English. This means that English does not have rules as such, it has conventions. The conventions, which includes spelling and grammar, change through usage.
> Unlike French ...
English has found it preferable to have a catalogue tracking its actual usage, thus allowing the language to grow and evolve rather than be tied down.
Of course these changes upset those of us who acquired their experience at an earlier time or in a different place.
Minor nit:
Rather than saying there are limits to what people need to know, I'd say the threshold for basic understanding is fairly low and can mostly be learned by listening and reading without formal training.
Otherwise spot on.
I'd also note that English does make it rather harder to notice some of the distinctions you learn in formal training. Therefore learning a "foreign" language in a formal setting, especially the Romance languages helps one understand the English rules better. I've forgotten almost all of my Latin, Spanish, and German, but my enhanced understanding of English remains.
No, 'English' when used to describe either the language or the people originating from England is a proper noun, not a pronoun and is therefore capitalized.
The danger of being the pedant, which I now risk as well, is to make sure you are not in error when making a correction.
If the word is written lower case I believe it refers to the spin one imparts to a spherical object in sports or leisure activities. For example, one might impart english to a cue ball in pool to make a trick shot. Of course, that might be a local idiom and non-standard usage.
"I am well published and have been researching, writing and doing formal presentations for the past few decades but believe me I have no idea what "first-person pronoun" is."
In that case you might be well published but you haven't published well. First person means the subject, generally 'I' or 'me', second is 'you' and third is 'he', 'she', 'him', 'her', 'it'. I (first person) learned that at school. What were you (second person) doing when we (first person plural) were taught that?
Please be understanding. Grammar has been removed from the curriculum in the UK, it is considered useless - one of my English professors from the UK was unable to answer one of the questions I had about English grammar, dismissing the question with: "We are no longer taught grammar at school, so I have no idea."
"My first language is English. In fact, my only language is English."
"I have no idea what "first-person pronoun" is"
There is some connection between these two sentences.
I was in school in the UK in the 60's and 70's and they never taught us the parts of speech, just some vague blather about verbs and nouns. Being unable to describe the construction of a sentence in ones own language makes it near fucking impossible to be gramatically correct in other languages.
Thank you, fucking British education system, crippling your clients just to make things "easier".
"Dismissing someone who is leading the biggest and most important software project in existence . . . "
Who's dismissing Linus? Certainly not me. I'm suggesting that someone of his stature and coding ability could apply a little of that skill to what is simply another language. Also please note that the repeated use of the lower case pronoun was by Kay Sievers. I have the greatest respect for both of these people when it comes to software.
"Dismissing someone who is leading the biggest and most important software project in existence based on "he used naughty words mummy". Grow the fuck up."
Hmm. There are rather a lot of software projects in existence, far more than either of us know; so this is an extravagant and unfounded statement. I suspect that the code used to programme the firmware (which is just software at a level nearer "the metal", real time software code in automated power plants, military, nuclear and other systems could be considered important. After all, there are lots of UNIX variants and look-alikes and many of them would, do and did function perfectly well in the role played by Linux variants. One could argue even that GNU is more important as it provides software providing critical user interfaces for developers, designers and other users across many systems from Linux to OSX to Solaris to BSD to whatever you like.
But then, your hackneyed coarseness at the end may be intended as a hint that you are not serious or that you are very unsure of yourself and so think that the equivalent of drunken bellowing will prove your point.
Swearing at a child's baptism or First Communion is, probably, inappropriate if you aren't the child. But in a 'public' forum, why is that inappropriate?
'Most important software project in existence' and 'leading' are two highly subjective terms, but I'm not sure anyone who is just one letter away from a pubic forum site for curly hair fetishists has even a cunt hair of wiggle room to be telling anyone about perspective.
'That Welsh bitch with the swollen tits is licking herself in the way only those Welsh bitches do when they've had more than three inside at once' - 'That's right Tom, it is very distinctive behavior and as near as I can tell, a Welsh Corgi giving birth live has only happened here once before, at the 1941 Westminster Kennel Club Dog Show. Hope those parents watching have updated their metaphor libraries, there's not a single bird or bee or stork in here tonight'.
See, that ^^^ is perfectly appropriate. It's all about perspective, yeah... Just like from the perspective of the guy who has his name on the thing you're working on, and want recognition for, if you're fucking it up he's got all the right in the world to tell you to fuck off. Kay can get it straight, whine or just quit and go home, it isn't Kaylix, but it could be and that's the thing. Kay can stay, or Kay can go, but until he has his name on the tin it doesn't fucking matter.
Linus Torvalds is a creepy little man I would have put into orbit if he was on the same continent as my daughter, but we have an understanding, so it's not an issue. I also think he has outlived his usefulness and is now serving as the best possible representation of what business people despise about Linux and until he goes away or somebody kicks the box out from under him and he's hidden by the podium the entire Linux movement is being denied the commercial support it so desperately needs so that 'free' isn't the only thing the general public thinks when they hear 'Linux'. But again, doesn't fucking matter, it's his name on the box. He can tell people to fuck off, or not, that's his call.
@Don Jefe " the entire Linux movement is being denied the commercial support it so desperately needs so that 'free' isn't the only thing the general public thinks when they hear 'Linux'. "
Red Hat, SuSE and Novell, all of them sell commercial Linux support, from the kernel up to font rendering in the browser. And those are just the three largest companies that do Linux support. There are many more.
So, what the hell are you talking about?!
If you mistook my comment for a swipe at Linux that wasn't what I intended. I had a bit of a poke at Linus, not Linux. I have to say I've been impressed with the maturation of Linux, the technical achievements are solid, but I'm most impressed that the entire Linux community hasn't accidentally killed itself in a misguided attempt to find some form of functional structure. It said it before and I'll say it again, I was, very, publicly wrong in my forecast of the development community's ability to find any sort of functional structure. I honestly didn't think any of the communities could maintain cohesion when there was so very little incentive for them to do so. I was wrong.
But selling support commercially and commercial support are wildly different things. We'll just run with Red Hat as the example: Red Hat sells support commercially for their flavor of Linux, but as tech companies go, they don't have a lot of commercial support and that's directly reflected in their revenue.
Everybody likes to point to the big name companies that run various flavors of Linux internally, but the Enterprise market is a wee, tiny portion of the market within any country or the entire planet. About 75%(ish) of all global revenue is generated by companies with less than 300 employees (less than 40 employees in the US) and guess what 85-90% of those companies use internally? It isn't Linux.
Before you get all defensive, that wasn't a Linux vs anything but itself comment. Linux is its own worst enemy. Linux is perceived by the overwhelming majority of the global business community as a thing that weird tech companies and the fat guy with the unkempt look and some sort of mobile recycling center in his car use for whatever it is they do.
Don't get me wrong, I find great humor in the fact that many scathing comments about Linux are composed on Android devices and served up on Apache for all the world to see, but what don't you see there? I realize what Apache is, but the point is still valid, nobody likes to say Linux. Even the 'LAMP' box fad came and went quickly. It's still a LAMP box, sure, but the term hasn't hung around like 'cloud' has.
That's the bitch about perception. You have to manage it, or it manages you, and that's what has happened with Linux. Take a look, even here on El Reg, what are the bulk of the Linux stories about? Either some technical thing that's already been covered by 9.765M other sites, or the latest antics of a Finnish Gnome with poor choices in suits (or tailors) and little understanding of group dynamics. There aren't many business related stories or investment related stories about Linux. Why? Android has plenty of those stories. Why not Cinnamon or MangoEarwax or whatever?
There are a lot of reasons, but one is Linus Torvalds himself and another is the totally ridiculous ways the very vocal community gets attention drawn to itself. Look at that guy up there with the 'Most important software project in existence' comment. That's extreme extremism and business just doesn't want that, it's fucking weird.
It's also a major element in the mess that saw every significant VC group and high risk investment firm move as far away from desktop Linux as they could get. Maybe the trees are blocking your view of the forest, but go take a gander at the most prevalent people say 'Linux is the best'. Go ahead, go take a look. What did you find? Yep. 'Free'.
Commercial support and 'free' don't mix and the fact the Linux community consistently misses that is just fucking sad. It's worse than sad, it's the behavior of insane people. THE SAME PEOPLE who run around screaming 'Windows is a ripoff, software should be free' are the same people camped out at the NoScript Socialist Club singing the glee clubs latest rendition of 'we are people, not a product' and laughing with excitement every time Facebook announces a new way to fuck users faster.
What's the message there? 'Free is bad unless it's Linux? You'll know you've hit the big time when Lilliputian software developers in apparently stolen clothes castigates you in public for your shitty contributions to the only 'free commercial computer operating system'? Good fucking luck with that pal. I've spun some serious bullshit but not even I can do anything with abortion of a desirable concept.
Again, the Linux community has surprised me before, maybe they'll do so again. Stranger things have happened. But until somebody gets in there and takes the reins of perception management the technical aspects of Linux are going to continue being marginalized with mass general adoption of (flavor)Linux, not Android or some other disguise, being no closer than it has been these last few decades.
It's no different than anything else, as any product matures and evolves, those who control/lead it must change and evolve as well. The transitions must happen together as well. If you change leadership but not the product it won't work, ask RIM. If you change the product but not the leadership, that won't work either: See Windows 8 for more information. Click the 'Start' button to...
Those things don't work because they are change for the sake of change. Attempting to force something instead of going with the flow and taking advantage of the free momentum. Linux has for the first time, in my opinion, reached a point where real momentum exists. Not the idealistic delusions of momentum, but an actual system, accelerating under its own power and it's a great time to begin rolling in some professionalism and actually participating in the markets and economies as a viable entity instead of being the lurker in the corner at all the parties.
Professionalism doesn't mean not swearing at people, or not threatening people's families with violent deaths, not at all. Professionalism means you do those things behind closed doors so as not to embarrass anyone or be implicated in missing persons cases. Professionalism means talking about what you offer, and why it has value. There's nothing wrong with value. There's nothing wrong with actually making a living doing what you enjoy versus doing what you hate/for who you hate so you can afford to go home and do the same thing, but for free.
There's this idea that a successful software company has to be worth zillions of dollars and be full of Elisions and Gates and Jobs', that's not correct. It can be a place full of geeks who aren't (all) dicks, still make scads of money, not 'sell out' or make your actual profits selling use data to Amazon, and I want to know, what's wrong with that? It would seem to me the problem is on the people, not the technology.
Microsoft had a policy of ranking staff and automatically firing those at the bottom so what exactly is your point here ?
And why the hell do you care so much about Linux to come here and preach us about marketing and professionalism ? It's free and you're free using it so if you like it please stop enlightening us and if you don't, then move to Windows and we can still be friends.
By the way, in case you didn't get it, there is no Linux behind close doors because each time something is hiding behind closed doors, it is not for the public good.
While I fully agree with you (and I mean completely), the idea of being able to sell an image change (to change inbound perceptions) to the "Linux Crowd" still seems a bit distant. My own perception of "black-hatting" generally involves a geek (like myself) or group of geeks, chatting it up using some obscure or even, unknown, platform whilst preparing to "dominate" this or "pwn" that...from some slef-brewed Linux derivative.
From a "professional" and "business-like" perspective, I don't want a system in my organization, running a platform that the most devious of us know all too well how to compromise...better than how my own security team knows how to protect.
Linux is that image...I hope you are right and that they are on the upswing - it's a great system, once the negative image is no longer in view...
There is nothing wrong with those words. If you reduced the swear words you use to make room for a broader vocabulary, you would know that a female dog is properly called a bitch; in this context, to say a dog was giving birth would be the equivalent of describing a woman as a man giving birth. These commentators are commenting on an event for professional dog breeders and showers, so they use the standard, English terms.
I am aware that USA English emphasises the derogatory usage (why not the same for "cow" or "pig" I wonder?); but this is an international forum set up by a British entity, so I assume that British English is the basic standard even if the editors are too lazy or permissive to edit USA-source material in the way they would with, e.g., French or Russian or Pidgin.
In short, ignorance and bad language are inappropriate in most contexts, formal or informal and simply alienate even those who may agree with the writer or speaker while distracting from the point the person is trying to make and confirming the opponents in their view, as this thread shows.
If LT can not control himself and think of his audience, can I trust his judgement when designing software and user interfaces?
This post has been deleted by its author
"Dismissing someone who is leading the biggest and most important software project in existence based on "he used naughty words mummy". Grow the fuck up."
Your obvious bias shows with that comment, but leaving that aside, I'd say Obama is a more important person as a leader oof something, and I'm sure you wouldn't expect him to behave the same way.
It seems to me that there are bigger and more important things than "leading the biggest and most important software project in existence". Things like respect, kindness to others, basic tact and diplomacy, in my mind, overrule any perceived greatness. I believe Linus Torvald's would have enhanced any greatness he might have by dealing with this situation in a quieter, more private and lest shouty way.
Local vs Global.
It's okay to swear in the sense of 'freedom of speech,' but it's not okay in the way that people don't want to hear it when anger is intoned. It's kind of obvious.
No one's dismissing Torvalds. I get angry and curse in public occasionally too, but I am always wrong to do it. It does not advance the biggest and most important project in existence: society.
But you know all this. Have a pint.
I believe the following work is widely regarded as definitive on the topic:
http://www.amazon.com/The-Elements-Style-Fourth-Edition/dp/020530902X/ref=zg_bs_11981_1/189-0798329-8339531
Although the following seems to be a local favorite:
http://www.amazon.com/Eats-Shoots-Leaves-Tolerance-Punctuation/dp/1592402038/ref=zg_bs_11981_5/189-0798329-8339531
and if your not up for either of those, you could always resort to the reliable:
http://www.amazon.com/English-Grammar-Dummies-Geraldine-Woods/dp/0470546646/ref=zg_bs_11981_9/189-0798329-8339531
I am reminded of what friends who have gone to Japan to teach English have said:
"I speak real English. I can construct an entire sentence out of nothing but the word f*ck in all it's various word forms with the occasional use of the intensifier 'mother'."
A statement which is both sad and true at the same time. And by that standard, Torvalds has quite mastered the language.
Er....what about acronyms that include swearing? Or is RTFM appropriate because it's not spelled out? Or are Reg comments not a public venue?
OK, OK, I'm SMILING, really. I'm not the comments police. Nonetheless, there's no avoiding the fact RTFM is not exactly entirely unlike swearing and perhaps undercuts your point eeeeeeever so slightly?
Skål, c
>I find Swedes and Germans speak English better than a lot of British folk.
Thats what I used to think until I started working in an international european company. The europeans all speak with good accents and grammar , but after a while you realise their vocabulary is a lot more limited than a native speaker, and if you start speaking idiomatically or - god forbid - use slang, they tend to get confused pretty quickly unless they've lived and worked in the UK for a long time.
This isn't a criticism, its same with anyone who learns a foreign language - unless you really immerse yourself in the language and culture for a LONG time you'll never speak the language as comprehensively as a native. I know - I used to think my French was ok until I spent a lot of time in France and realised how poor my actual ability in the language really was. And still is to be honest.
I was raised bilingual, as are many others. Both of Linus' parents were Swedish and that language would most likely have predominated in the home, but Suomi would have been the language used in educational institutions, with a strong understanding of English required for most computer-related courses due to the Anglo-centric nature of most programming languages.
That said, Scandinavians typically learn multiple languages in school from an early age as a matter of course – typically from the UK equivalent of primary school and up, as the younger you are, the easier it is to learn new languages. Most will be taught their native language, plus two others, often including English, with German also popular.
For what it's worth, I've always considered programming as mere translation, nothing more. The trick is to understand how the target audience – i.e. the computer – 'thinks', and work within their frames of reference, but that's a given for any language. I used to get weird looks from colleagues when I told them I really could think in the programming languages I was using.
I don't care if their first language is Klingon Linus Torvalds comes off as a foul mouthed 14 year old Halo player...is THAT the kind of image you want for Linux? An OS run by foul mouthed maladjusts?
There was an article recently talking about how the Linux kernel development has become "an old boy's network" because no new blood has come in in years...here's why, nobody wants to deal with bosses who are foul mouthed ego-maniacs. If you can't act like a professional then you shouldn't be doing the job,PERIOD.
You're bitching about their English, others are bitching about their coding and lack of grown-upness, yet surely, when it comes to their ability either to write code or manage a software product/project, The Most Important Issue has to be did either of them support Prop8 six years ago?
/if you need sarc tags for that then you need to grow up and get over your single issue outlook.
actually this sentence:
"Neither of these guys appear to be able to code proper English." is written in a passive tone - not exactly a "no-no", but not a good habit either.
It should be written thusly:
Neither of these people appears able to code in proper English.
That is grammatically and linguistically correct. :-)
(I acutally do a lot of writing on a daily basis and speaking and I speak Geramn as well as native English, but that doesn't mean I am always right - gotta love that spellchecker!)
So let me see if I get this straight: at minimum the bug is 1 year old, as it is claimed that Sievers created the bug and since he the last time he worked on said code was at minimum a year ago, and not only did no other coder fix the problem but Torvalds is only now complaining about it? As well as complaining that Sievers is not fixing it himself rather than find out that Sievers is no longer working on Linux projects at all?
It only leads to one conclusion in regards to this matter: Torvalds=idiot. It is not a fine statement about 'how good open source' is when there is a known bug and the project manager lets it sit for a year while awaiting servicing from the coder that will never arrive due to him leaving the project. Who's the prima donna now?
<QUOTE>Apparently YOU need to reread the article - the AC got the timeline correct but apparently you (and the morons who upvoted you) DIDN'T</QUOTE>
No dear. The bug was in SystemD, not in the Kernel. Linus threatened to exclude him from working on the kernel even though he last worked on the kernel a year ago and had started working on systemd instead. The bug happened WITHIN the last year, up to the present.
I would have used he joke icon if I could.
@AC - "So let me see if I get this straight: at minimum the bug is 1 year old, as it is claimed that Sievers created the bug and since he the last time he worked on said code was at minimum a year ago, and not only did no other coder fix the problem but Torvalds is only now complaining about it?"
It IS a bit amusing when you think about it. Might be interesting to get together a list of other non-contributors to the kernel that Linus should refuse to accept code from in the future:
- George Clooney?
- Lindsay Lohan?
- Steve Jobs?
- Kurt Cobain?
- Vladimir Putin?
- Bill Gates?
- Darl McBride?
- Larry Ellison?
- Kim Dotcom?
>>Man, the open source community is full of assholes.
And arseholes. Why insult donkeys?
Actually, perhaps it's "open source" as in, open arse, voiding its contents (that are more than those of the head of the writer).
Sometimes the writers are correct. But their inability to express themselves even a little bit courteously detracts from their value or even kills all interest in reading their opinions.
The Open-Source world is full of, shall we say, Differently-courteoused persons. That world doesn't select for courtesy--it selects for competence. The For-Profit-Business world is full of Traditionally-courteoused sociopaths who speak what some would call "grammatically-correct English" and what others would call "Marketroid". People don't have "skills", they have "brands". People aren't "liars" or "honest", they are "promoters" and "detractors" respectively. "Customer care" doesn't involve precision injection of prescription pharmaceuticals, or even changing bedpans: it means "emotional manipulation of other people for personal profit." "Excellence" doesn't involve doing anything well, let alone better than others, it means "spontaneous enthusiastic expressions of delusions of adequacy." They never say "shut the expletive-deleted up", but--because this is something they need to say so often, they have whole manuals of phrases that mean the same thing, from "I understand your concern" and "I feel your pain" down to "I don't care what you know; I don't know what you do; but with two weeks' training I could teach an idiot to replace you."
> The For-Profit-Business world is full of Traditionally-courteoused sociopaths who speak what some would call "grammatically-correct English" and what others would call "Marketroid".
And one of the common tactics when they sell you software that doesn't work is to wait until you start yelling at them in frustration, then feign offence and claim they can't work with you.
@AC: "at minimum the bug is 1 year old, as it is claimed that Sievers created the bug and since he the last time he worked on said code was at minimum a year ago,"
You did not read the Reg article properly, you certainly have not looked at the linked material, and it does not surprise me at all that as a result you called Linus Torvalds an idiot. Allow me to say it does not improve your own (anonymous) reputation...
Executive Summary: The last time Sievers submitted a patch to the *kernel* was a year ago. The bug is not in the kernel but in systemd that Sievers develops and maintains, and the "years" part is Linus's post refers, apparently, to a repeatedly observed attitude.
For those who might be interested in the actual problem, see the bug report. It is obvious that Linus is not even the main person who is annoyed. Quite a number of top kernel developers seem to be of the same opinion (Borislav Petkov, who reported the bug - and later added, "I was right to be very skeptical when considering opening a bug here," - Mel Gorman, H. Peter Anvin, and others). And look at Comment 14 by Luis Rodriguez - Sievers rejected the bug report within ~20 minutes without any discussion. To emphasize, a userspace program reads the "debug" parameter on the *kernel* command line (used for ages), seemingly interprets it as *its own* parameter and starts spamming the *kernel* log buffers with *its own* debug messages so much that the machine fails to boot.
Linus's comment about not accepting patches is addressed not to Sievers but to Greg KH who, apparently, has related patches in his pipeline. Linus is saying he is not willing to risk destabilizing the kernel by code that originates with developers who routinely dismiss bug reports out of hand. He does say that if distros merge said patches and test them he would be willing to consider them.
I have not checked, but judging from the LKML thread the fix for systemd was actually submitted by a kernel developer (Greg KH).
Two points in mitigation:
"You did not read the Reg article properly, you certainly have not looked at the linked material"
Well, I think quite a lot of readers don't look at the linked material. We rely on El Reg to summarise enough of it so that we have a balanced view of the situation without doing all the research ourselves. Thanks, at least from me, for the additional summary.
And in any case:
If the kernel can't protect itself against bugs in user-space programs, it isn't a very good kernel. Linus is free to have as low an opinion as he likes of the systemd people concerned, but he does need to change his kernel to address this. It's a DOS attack vector and if it was in Windows then we'd be queueing up to explain how it proves Microsoft's inherent shit-ness.
> If the kernel can't protect itself against bugs in user-space programs, it isn't a very good kernel.
upvoted as it's a good point, but I don't think systemd is a *normal* userspace process. From wiki "systemd is a system management daemon designed exclusively for the Linux kernel API. For systems using it, it is the first process to execute in user space during the Linux startup process. Therefore, it is also the parent process of all child processes in user space. "
It isn't kernel but it does seemed privileged in some ways so *perhaps* it can be expected to be written more carefully than other userspace code.
(disclaimer: am linux noob)
The KERNEL doesn't have a problem.
SYSTEMD is acting as the system init process... and is so screwed up it doesn't work. KERNEL options are not supposed to be interpreted by ANY userspace application. Yet SYSTMD does.
Totally in violation of the rules.
Second, this is NOT the first time this person has violated the rules. He broke the usb subsystem... and then said the kernel should actually do it instead of actually fixing the bugs he created.
Sometimes it really is necessary to cuss someone out. If you are in the personal presence, you get to do things like pound on a desk for emphases.
In text, you don't have that ability.
> Many programs look at /proc/cmdline
That they *do* doesn't mean that they *should*....
> See "grep cmdline /etc/init.d/*" for some examples.
On my current laptop, I've just got livesys and livesys-late that do (and I don't really get what they're doing there anyway - probably an artefact of installing form a Live image). On my server upstairs, there are no /etc/init.d/ files that look at /proc/cmdline.
I agree with the earliler poster - these are command-line arguments to the kernel, and were already defined in that context long before things like systemd came along. Should a userspace program *decide* to look at the kernel's parameters and act upon them, on its own head be it; it is the one that is overloading those parameters.
Vic.
"If the kernel can't protect itself against bugs in user-space programs, it isn't a very good kernel."
I'd go a step further: If you look at FreeBSD (or any BSD/Unix), you'll see that programs and services that are essential, are all maintained as part of the core. Linux, on the other hand is *only* a kernel plus GNU/FOSS tools. As this particular discussion shows, it can't even boot without third party stuff (at least if the third party stuff has bugs). The philosophy is less than ideal for stability and security.
I'm not taking away anything from Linux (the kernel) here, nor am I saying that Linux isn't any good. Far from it. But this is a problem that has its root in the structure, and it won't be the last time that it causes issues.
As for Linus being a twat and rather agressive... I believe he has to put up with a lot of shite submitted by tons of people. Nonetheless, I find it inappropriate and immature how he voices his criticism. It does seem to work, though... an old issue was fixed shortly after...
"If the kernel can't protect itself against bugs in user-space programs, it isn't a very good kernel. Linus is free to have as low an opinion as he likes of the systemd people concerned, but he does need to change his kernel to address this. It's a DOS attack vector and if it was in Windows then we'd be queueing up to explain how it proves Microsoft's inherent shit-ness."
That. Even if you accept the argument that systemd is special, and should therefore be held to a higher standard in return for privileges, the fact that the system can be DOSed by the absence of any sort of flood control isn't good. systemd needed the fix, but so does the kernel.
T.F.M, nice write up.
I'd also like to add that people are saying that the kernel should never let a user space app crash it. Well, systemd is no normal userspace application. It's not a word processor or a web server. It's PID 1, the first process the kernel starts and the parent of all other processes. It's responsible for starting everything that mounts file systems, start network services, and the works. If you boot up Linux, some distros show the [OK] after services started, that's PID 1 doing the work (or one of the tasks it created). If PID 1 dies, the system panics. This is the way it has always worked. PID 1 is as *important* to the system as the kernel is.
Second, /dev/kmsg is a file that lets privileged (root only) tasks to write into the kernel logging system. Systemd uses this to write messages into it in early boot up because there's no place else to write to. The filesystems haven't even been mounted yet. /dev/kmsg hooks into the kernels own logging that prints out the messages you see on boot up, like the Linux kernel banner. This is also what it uses to print out oops messages. The output is considered critical and writes to it wait to make sure the data is seen before it continues, as we want as much data out before the system crashes. It's a critical logger, and not something to take lightly.
The bug was that in early boot up, systemd had a bug in it where it would write loads of data into /dev/kmsg, and because this is a critical logger, systemd had to wait till those messages made it out to the console before continuing. If it also had some timeout that would trigger more prints, this could cause systemd to "live lock". That is, by the time it printed out a message, the timeout would trigger, and it would print out the same message, and the time out would trigger again, never letting systemd gain any forward progress. At this point, the system is hung. Remember, systemd *is* PID 1, and if it fails early, so does everything else. Nothing happens, and you can not even log in.
When I read that people say that the kernel should not let userspace hang it, it really did not. It was systemd hanging, but that's pretty much the same as the kernel hanging. The problem is that kmsg is a special file that most userspace is not allowed to write to. If you put too much data into it, it can cause the system to come to a crawl, as things must wait till it finishes. But a patch was created because of this thread that rate limits the data to /dev/kmsg. This patch was written by Linus Torvalds himself. What it does is if there's too much data written to /dev/kmsg, it starts dropping new data. This prevents the writes from taking so much that things stop running. But it also means that you might be losing important data you want to print. There needs to be a balance. You want as much debug as possible printed, but not so much that the system hangs trying to get that data out to the console immediately. And no, you can't let things progress, because the logging is that important that it must get out before the true bug locks up the box.
@Someone Else
nevets23 post was directed at Reg users who are not "in the know" of the Linux Kernel/Systemd development. I.e. outsiders.
Torvalds rant was not "public", it was directed at those "in the know", i.e. insiders, who would understand the context. Therefore for his target audience, it was appropriate. Just like nevets23's post was appropraite for his target audience.
Don't judge him on all you see in the news. I strongly suggest you follow the Linux Kernel Mailing List (LKML) thread a bit before coming to any conclusions: http://lkml.iu.edu//hypermail/linux/kernel/1404.0/01327.html
In particular, he observes in 01504.html that an earlier suggestion of his wasn't so hot:
Yeah, what Andrew said. My suggestion of per-task or per-cred is obviously moronic in comparison.
Linus "hangs head in shame" Torvalds
This is not the statement of a “f*cking primadonna”
" I strongly suggest you follow the Linux Kernel Mailing List (LKML) thread a bit..."
Interesting. "what Andrew said" was that the rate limiting should be applied per-file-descriptor and this was in contrast to per-user. It was then noted that per-user would be more effective against someone who tried to get around the per-file-descriptor restriction by opening several FDs, to which Linus responded:
"I don't think we should try to protect against wilful bad behavior unless that is shown to be necessary. Yeah, if it turns out that systemd really does that just to mess with us, we'd need to extend it, but in the absence of proof to the contrary, maybe this simple attached patch works?"
And indeed it seems to work. Someone had one of the previously afflicted systems booting by Thursday. So it's all remarkably boring and grown-up and productive over there.
And elsewhere in the thread it is noted that the systemd people have fixed their side of the bug, too.
"Those sensitive gay people ruin everything huh? At least the UK establishment has never persecuted gay computer scientists. Oh wait that's right they basically murdered one of the greatest in history".
I am no fan of 'the establishment' but he was not 'basically murdered'. He was treated badly but they did not come close to murdering him. It is far from clearthat Alan Turing committed suicide. It is quite possible he died due to a stupid error but if he did commit suicide it was his own decision and not remotely forced upon him.
I do not understand the desire to turn a talented mathematician into some sort of martyred saint. Let his achievements stand for themselves without the embellishment of martyrdom.
"It is quite possible he died due to a stupid error"
Doesn't it strike you as extremely unlikely that a smart well educated man would accidentally ingest cyanide at his home ? In a lab where they are swapping the labels on bottles for shits and giggles maybe, but cyanide shouldn't really be knocking about at home.
"but if he did commit suicide it was his own decision and not remotely forced upon him."
That assertion has less evidence to back it up that the conspiracy theory.
Turing wasn't some random coke snorting trader, or celebutard, he was one of the key brains involved in cracking codes. In a Cold War you would want those kinds of guys fighting for you, and if one of those guys dies before his time you really *should* be trying to work out exactly what happened - so you can make sure it doesn't happen again.
As it turns out they didn't work out what exactly happened, they chose not to investigate further pretending that nothing happened and everything was OK. It's exactly the same stunt they pulled with Kim Philby...
Nothing to see here, move along - all those dead soldiers and spies are nothing to worry about...
So guy isn't a kernel-type coder, and doesn't interact directly with them - by choice he says. (there's mucho background there fursher)
But coder's widget is fundamental to Linux and affects the kernel people fundamentally.
How do you get dis guy to fix his stuff?
Public shaming, if nothing else works ... ?
But coder's widget is fundamental to Linux and affects the kernel people fundamentally.
How do you get dis guy to fix his stuff?
Public shaming, if nothing else works ... ?
That isn't how open source works. The idea is if you don't fix the bug someone else will. Too often of course the premise fails: 1% of the user base have the ability to fix the problem and of those high-earners perhaps 1% have the time and inclination to do something about it. Before you dispute this consider how many long-standing security bugs were recently found that can ultimately be traced back to the MIT X release. It didn't work there, did it?
The open source contract works both ways: essentially it reads as "Here is what I have done, knock yourself out with it". It doesn't mean "This is my baby, you must feed it, and if you don't have breasts you must grow them".
Don't get me wrong, I am generally pro-open source, but the quid pro quo is that no one has any duty to do anything, no matter how much you might like them to.
>It doesn't mean "This is my baby, you must feed it, and if you don't have breasts
>you must grow them".
That would be the case if Kay weren't being paid for his work and was doing it out of the goodness of his heart *AND* the systemd project hadn't merged in udev etc that are fundamental parts of the Linux userland. If they didn't want to make sure their stuff works and support it i.e. run their project as a toy project they shouldn't have done things like that. "It's opensource and free, do what you like with it, if you don't like it don't use it" is the line that Lennart uses to get rid of people that bring up realities he doesn't like. It would make a lot more sense if they hadn't created a situation where it's almost impossible not to use their stuff. Then you have the problem that they refuse to merge patches to fix problems. They won't let people help them.
I predict we'll eventually have exactly the same situation we had with glibc.
"Don't get me wrong, I am generally pro-open source, but the quid pro quo is that no one has any duty to do anything, no matter how much you might like them to."
When you lobby so hard to be the default init daemon of pretty much every major Linux distro then, yes, you bloody well do have a duty to make sure your daemon plays nicely with the Linux kernel.
Ha fucking ha. Kay and Lennart get your shit together. You guys browbeat everyone that makes any negative comment about systemd (to the point of hijacking a speakers talk at a conference) but time and time again you prove that you shouldn't be trusted with vital system components.
Bit late here and had a beer, but am I right in the following
1. systemd does this http://en.wikipedia.org/wiki/Systemd and used by a lot of linux distros
2. it is therefore quite important and needs to work
3. a linux kernel dev found a bug that affected the kernel
4. the guy that produced said bug, said he was not going to fix it http://lkml.iu.edu/hypermail/linux/kernel/1404.0/01327.htm
5. he got called a F*&K*ing wanker ( could be me, paraphrasing :-)
6. said guy, starts prancing around, saying, yeah so, after he refused to fix it
Linus does seem quite mild mannered at times IMHO :-)
off to bed, could be tl;rw (too late; read wrong)
OF COURSE he refused to fix it - if you read the article, he is no longer on the project and hasn't been so for at least a year.
It is comparable to moving on to a new employer and the old one complaining that you refused to come back and correct a necessary project - let [me] repeat: "I don't work there any more!"
If Tolvalds has a problem with an individual making a choice to quit (his) project and move on, well too bad. A person is not married to him, nor his project, ad infinitum simply by making a contribution - people move on. Tolvalds needs to learn how to do the same.
What do you mean he is no longer on the project? Sievers is still maintaining systemd surely? He made a change to systemd that hijacked a command-line parameter that was intended for the kernel. When other people described this as a bug, he refused to fix it because he said that it was not a bug. That is what is causing all the problems I think.
To be fair, it _isn't_ a bug, not in the technical sense; it is most definitly a "feature"
It should come as no surprise to any decent coder that turning on debugging can cause a data flood.
As others have said, if this causes a problem for the kernel, then THAT is where it should be limited.
> It should come as no surprise to any decent coder that turning on debugging can cause a data flood.
Well it might surprise you to find that switching on *kernel* debugging also switches on systemd debugging via a dodgy backdoor resulting in a flood of distinctly non-kernel debug messages.
That was essentially the problem at hand.
"...What do you mean he is no longer on the project? Sievers is still maintaining systemd surely? He made a change to systemd that hijacked a command-line parameter that was intended for the kernel. When other people described this as a bug, he refused to fix it because he said that it was not a bug...."
This sounds exactly like Alan Cox. He said the same thing: that the Linux kernel is broken and should be fixed. Linus yelled at him and said the bug was in Alan's code. Alan said there was no bug, the flaw is in the Linux kernel. Alan Cox quit. Just like this systemd guy, and numeous others. Linus himself said "I have scared away all the normal unix gurus, now there are only the weird ones left" - or something similar.
http://felipec.wordpress.com/2013/10/07/the-linux-way/
http://apolyton.net/showthread.php/130212-Linus-Torvalds-is-a-terrible-engineer-Alan-Cox
"Linus is a good developer, but is a terrible engineer," said Cox. "I'm sure he would agree with that."....Cox explained that he and Torvalds sometimes have different approaches to fixing a problem, due in part to their different responsibilities. As the maintainer of the development kernel Torvalds needs make sure the kernel code is easy to maintain, while Cox is more interested in kernel stability and is not so worried about "hacking" the code to get it to work...."One of the hard problems to fix are design errors," said Cox. "These are a pain because they need a lot of refactoring. Linus' approach is to re-write it to a better design. But to get a stable kernel you tend to do small horrible fixes. Linus is very keen to have maintainable code, while to have a stable kernel I'm keen to have code that works."
No, "professional" does not mean that one is tied to every employer one ever had. One's current employer may object to one spending time on, possibly a rival's, software regardless of the authorship.
Most firms have a legal contract with the employee, that any software, invention etc. done for that employer belongs exclusively to that employer. Often, the contract forbids outside work without approval or even at all. Conversely, that means the employer takes full responsibility for work done for them and if, for whatever reason, the worker moves away, it is the employer's responsibility to ensure he covered this possibility. It is the professional's responsibility to do his work to a proper standard for his current employer, to keep to his contract and to protect the interests of the employer.
It's not just the Linux kernel maintainers. Debian's technical committee almost didn't recommend switching to systemd, precisely because many people have a... different... standard of how to do technical collaboration.
Though, it's alarming that Kay thinks he can just ignore the upstream kernel people. Lennart Poettering is trying to develop kdbus and get it into the kernel, so Linux would have a more useful RPC system. Lennart seems to be siding with Kay here, and Red Hat does maintain its own sets of patches away from the upstream Linux kernel. This doesn't bode well for collaboration.
Gentoo is already forking udev away from systemd. Since systemd is free software, perhaps the solution will be forking systemd at some point, like all those forks that GLIBC and GCC used to get until their respective leaderships changed.
This post has been deleted by its author
Sure, ideally, the kernel should be able to handle any mount of logging crap thrown at it.
But if you identify a particular userspace program that is sending it a lot of useless noise logging rather than vital debugging info that is necessary to the operation of the system, THEN it is the goddamn fault of the moron responsible for said userspace program.
If your kids are shouting at you 16 hours a day that they want a PS4 to the point where it is driving you insane, is your fix to try to improve your mental stability to where kids shouting at you 16 hours a day don't faze you, or is the real fix to tell your kids to shut the hell up about the stupid PS4?
DougS, the real fix is obviously being able to withstand 16 hours of shouting per day. Once an Achilles’ heel has been revealed (kids are smart — they’ll know perfectly well the real reason why they were told to shut the hell up), they’ll aim at it over and over and over again. To paraphrase Milton,
Of what now I sufferThey were not the prime cause, but I myself,
Who, vanquished with a peal of words, (O weakness!)
Gave up my fort of silence to my children.
"If your kids are shouting at you 16 hours a day that they want a PS4 to the point where it is driving you insane, is your fix to try to improve your mental stability to where kids shouting at you 16 hours a day don't faze you, or is the real fix to tell your kids to shut the hell up about the stupid PS4?"
nah, give them a XBox
Yes, that was my reading of the situation. In any software a subsystem in debug mode is entitled (even required) to emit as much debug info as it deems appropriate. If the host system cannot cope with that level of information then it is up to the host system (the kernel, in this case) to throttle it.
So, agreed, this appears to be a kernel bug.
Wrote :- Systemd sends excessive data to the kernel .. to the point where it either hangs or crashes, ... So it's a bug in the kernel then ... can't cope with the data it's being sent"
As I understand, it isn't "data", it's garbage. The example I saw in the bug report is just repetitive. Nor is the kernel crashing - dealing with this stream of garbage is taking all the processor power. That is not a kernel bug. Bear in mind that this is stuff at the system level where things should work together without needing kludges.
The issue is whether the kernel should be patched to recognise this stream of garbage and cut it off, or whether systemd (which is the newcomer here) should be patched to stop creating the garbage in the first place.
"The issue is whether the kernel should be patched to recognise this stream of garbage and cut it off, or whether systemd (which is the newcomer here) should be patched to stop creating the garbage in the first place."
The kernel needs to be fixed, otherwise you have a nice easy way a userland process can give the appearance of bringing down the machine.
[this doesn't mean systemd was not faulty - but the kernel shouldn't choke so easily]
Should systemd be filing the kernel debug log with garbage? No.
Are Linux and co discussing a rate limiter on the kernel debug log? Yes! And even with the very draconian limits they tested, only one program actually hit them. Guess which.
Go read the full thread. It's enlightening - it's also clear that, far from simply shouting, Linus was very restrained and entirely justified in every statement he made.
Systemd sends excessive data to the kernel logging system to the point where it either hangs or crashes, right?
So it's a bug in the kernel then.
Errr..... No. If a routine is flooding the kernel with spurious nonsense, it needs to be shut up. The kernel actually does that. Sadly, the programme that causes the problem is fundamental to the boot routines of a couple of Linux distros. The dingbat who wrote the faulty code should fix it. If he can't, he should admit his inability, and ask for help.
> "Systemd sends excessive data to the kernel logging system to the point where it either hangs or crashes, right? So it's a bug in the kernel then. "
No, this is a critical, hyper-privileged system process which works hand-in-hand wth the kernel. A misbehaving user-space process should not adversely affect the kernel, but that is not the case here.
I concur. Disabling a service in SysV just requires a chmod a-x on the symlink at the appropriate runlevel. Systemd puzzles me. A lot. Services sometimes turn itself back on with no intervention, and getting a desktop manager service itself to start is spooky voodoo if you install a minimal command line system first and the X components and desktop managers later. More than once I made a text only installation and later install a desktop environment, and the bloody desktop manager doesn't start. This is especially true in OpenSuSE and Sabayon where I needed to poke around and eventually managed to figure out that I need to make a symlink somewhere into systemd.
@RAMChYLD
Yes systemd is a sysadmin nightmare. I have also struggled with adding a graphics to a working minimal system. Systemd has beens the only thing to frustrate upgrading from the long-termopenSuSE-11.4 without having someone on-site to force a reboot.
I understand the attractions of the systemd approach to boot-time and daemon management, but the implementation has been a bit amateur. The megalomaniac tendency to ensnare everything it touches just does not sit well with the 'nix philosophy.
I guess one problem here is also people thinking they can do better than unix without understanding its philosophy.
So far there have been very few (if any at all) non-unixoid systems that actually work and are maintainable, while most unixoid systems were rather good successes. The Unix philosophy seems to be such a great way to reduce complexity. Most software projects fail because of overboarding complexity.
I cannot upvote you enough for this statement. I thought I was the only person left who thought along these lines.
I've been working at source level on UNIX on and off for 30+ years, and I'm finding the complexity of what is being added to Linux bewildering. I thought it was time to start thinking about retiring, but knowing that there are other people out there who think the same refreshes me.
> Go back to SysV init.
But ... but ... but ...
Then no-one could do any of that oh-so-important wily-waving about fast boot times.
Because, as we all know, boot time is *critical* to everything.
[vic@hobgoblin ~]$ uptime
13:20:55 up 257 days, 16:49, 2 users, load average: 0.70, 0.73, 0.79
Vic.
It's all in the boot script. If you code the scripts right, the difference would be negligible, down to a few milliseconds. For example, adding an ampersand in the end of the daemon invocation incantation in the Apache and Samba initscripts actually sped my boot speed up by a factor of five back in the SysV days. And to be honest, it also rings true for BSD- my OpenBSD box chews through the rc.local file as well thanks to well-placed ampersands.
You gotta fork the initscript to the background, or it will slow down as the system is waiting for the daemon to initialize before proceeding.
If you code the scripts right, the difference would be negligible, down to a few milliseconds
This is not true.
SysV is inherently synchronous. systemd is not.
systemd will therefore start a box much more quickly, but at the cost of greater complexity, because it runs the start-up in parallel.
adding an ampersand in the end of the daemon invocation incantation in the Apache and Samba initscripts actually sped my boot speed up by a factor of five back in the SysV days.
But at what cost?
SysV scripts expect all previous scripts to be finished before they start. If you're forking off into the background, this expectation is no longer met. Thus, if you have a dependency between two scripts, you have just built a race hazard into your boot system. This is not a good way to build reliable systems.
You gotta fork the initscript to the background
You must not fork the initscript into the background unless you're prepared to have boot-time failures. The minute or so you might gain simply does not warrant that sort of instability.
Vic.
Issue 1. Systemd should not be using the kernel debug setting.
Issue 2. Systemd should not issue debugs that are not useful.
Issue 3. Kernel needs to be more resilient to logger activity.
In a way systemd has done the kernel guys a favour by highlighting a deficiency in the kernel logging. That said, systemd has stuffed up, and should have admitted it right up front.
They're discussing rate-limiting incoming messages to the kernel log, and there is another display of colourful language:
>
> Could be done per-fd: put a struct ratelimit_state into struct
> devkmsg_user.
Yeah, what Andrew said. My suggestion of per-task or per-cred is
obviously moronic in comparison.
Linus "hangs head in shame" Torvalds
You should have probably linked to Linus' message instead of citing it (with bad formatting), because your intent is unclear and, as a result, it seems you were misunderstood. I guess what you wanted to do, was to post something similar to this, perhaps with a bit of irony. Didn't quite work, did it?
Ugh. It's like Linus Torvalds is deliberately trying to be as obnoxious as possible. His antics are receiving more coverage than his achievements. Whenever I see his name on the headlines, 10 to 10 it's about him throwing a fit. He's now officially known as that Linux guy with a shitty attitude. Why does someone so accomplished need to cultivate such toxic behavior? This is the type of stuff that ruins communities, causes forks and makes a lot of people angry - and the already grumpy programmers even grumpier. I think at this point he let the thought of being the creator of Linux go to his head and deluded himself into thinking he's royalty.
Honestly, I can't. Why antagonize the very people we need to have as allies? It's an irrational approach to problem solving. You can perfectly fine address these conflicts without being overactive, which sadly seems to be de rigueur among many programmers. It's an unnecessary acquired behavior and does nothing but create hostility in these projects. It also detracts from the issue at hand, so much that this article mainly focuses on yet another Linus' "verbal smackdown," otherwise it wouldn't be news. I applaud Lennart Poettering for addressing the issue with class and his plea for more friendliness.
"Whenever I see his name on the headlines, 10 to 10 it's about him throwing a fit."
That's part of the problem - people getting their viewpoint from misleading headlines instead of bothering to look into the full back story. Put yourself in his shoes, read up what actually happened, understand the issues (yeah, who wants to bother when it's so much more fun to bless the world with an ill-formed opinion instead?) and you might find it's a different story.
People, we have an unparalleled fantastic resource here on the internet, communications our forefathers could have barely dreamt of, with the ability to research matters and get all the information we want. Don't ruin it by commenting on a summary of a biased summary, third-hand, designed to inflame and fling FUD. I for one still admire Linus and his work.
It's getting worse. As a programmer of 20+ years the "young uns" seem to be absorbing this geek chic rubbish and think they are masters of the universe, just because they can write some crappy code.
To paraphrase Tori Amos. "So you can make some code, doesn't make you Jesus"!
What a hell of a word. Open like nice, what a miss understanding. There has never been any nice with open. Open is where you cannot hide behind your company, their lawyers, and your errors. Open is where you cannot onanate behind closed doors. A world where your imaginary penis is just imaginary. And now forgive me for this rant. Now, during my 35 years of programming I have had to understand and accept that I have never had the right to question the shit my superiors have programmed nor have I ever had the right to question mad time schedules. What a nice life, always supported bye strong dorks as long as there was nobody but me to accuse. Open source is nothing like that, it's more like honest, shit is shit, and please, you are welcome again if you skip the shit. So what was this article about, as far as I have understood there was a guy who made a silly programming error, then there was somebody who pointed it out and the first guy said shit, and then (after presumably 10000 words of softness) Torvalds said "SHIT" you damned primadonna. A rather soft expression towards a prick, or have I missed something. Anyway, you programmers of to day, If you want to live and learn software go and join the open source world and perhaps, if necessary do some work where ever, keeping you soul and beer on the table. And for primadonnas, they dwell further south, never in the north, or make Kimi a primadonna if you can, and sometimes with this northern inability to vomit lots and lots of words, "you shit" often works better and is shorter too.
That the last dev to feel the wrath of Linus also worked at Redhat.
They really are such delicate little flowers there - and I've repeatedly run into refusals to fix critical bugs as a RH customer.
The Debian crowd are right to be moving away from systemd. As long as the main devs continue to exhibit the behaviour described in the article (and comments) their code is best heavily quarantined.
It's not a major tech company it's an Open Source project that functions by collaboration.
And have you considered what would have happened if it were in a major tech company? Maybe the issue that Linus was commenting upon wouldn't have arisen in the first place, maybe the 'employee' would have been fired long ago? Who is to say?
Got any other useless comparisons?
The OS project may not be a major tech company, I do not know what it's legal status is and am too idle to research it. But any serious working place, whether a "project" (and I would suggest those days are long gone) or a firm of any size should not and can not tolerate such behaviour. Once, or even twice, perhaps: once it is clear that the perpetrator is unable to manage the common decencies and persists in aggression, he should be and, usually is, OUT. No one, not even the greatest genius, is indispensible; but a supported and loyal work force is. In this case, the genius left a great Denial of Service type hole in the operating system. Pure genius. People will make mistakes (unless they never do anything). That's why professionally written software is supposed to guard against them, particularly operating systems.
As for the writer of the code in question: good leaders and employers know not to simply sack people who, on the whole do satisfactory work, because of even quite a serious mistake: this demotivates the others, loses a generally useful worker and is really a sign that the quality and testing systems are at fault. It is better to learn from the mistake.
You know, work, profit etc. are not here for their own sakes. People are not here for their sake. Work, profit - these are their to provide people with the wherewithal to live. Even Linux is not more important than people (though some of its devotees make me consider revising that idea at times).
You miss the point. Linus, and the other kernel developers, were not railing against a colleague or employee. They were expressing displeasure at a supplier (systemd) for repeatedly delivering a shoddy product that impacted on their work, and for largely disregarding customer (kernel developers and distribution admins) feedback.
I think ms is looking for literate devs and since this guy made a rookie mistake, he should consider the alternative in Redmond.
As for Linus, come one, that was the prime example if a rookie mistake, untested, blind checkin - on a central daemon like that, shit, get your act together, Kay (Entschuldige) - Linus was absolutely right. Why should the kernel team spend time on making the kernel idiot-proof?
And, Window cleaners, please keep off these discussions you cannot understand - there are those comments on many windows flaws that need your attention more. If you administrate windows server, sql server, or use ie, your opinion dies not count!
That's exactly what a kernel must be. Users, tired developers (even experienced ones) will make mistakes or just not foresee a risk. At one time, the kernel was so vulnerable one could write a recursive for-loop in shell script and halt it. Of course the kernel, as any code, must allow for unfortunate or malicious code at any level. That's what distinguishes good code from bad.
Now, Mr Perfection Hans 1: are you honestly telling me that you never made a newbie mistake despite being, in your opinion, experienced and clever? Did you never find a bug in your code that was so basic, you just could not have put it there? Did you never do the daily drive to work, that you do every day for years, and not do a mistake that a novice driver would have been ashamed to admit?
I suspect you are due for a most embarrassing pratt fall. I look forward to you telling us about it and apologising for your arrogance.
Oh, your mis-use of English grammar in your missive shows the "rookie" mistakes you make in a language that, presumably, you use every day, typing on a keyboard that, I assume, you are not using for the first time.
Ok, I have changed my opinion somewhat after reading the systemd bug:
https://bugs.freedesktop.org/show_bug.cgi?id=76935
So, the issue is, there was a bug in systemd that had been already fixed (at the time the above bug was created) and that caused the kernel log buffer to fill up, preventing the system to boot.
Now, the discussion was, that systemd should not interpret debug on command line, only systemd.debug. I think this is wrong. If you specify debug on the kernel command line, you want to debug everything, if you want to debug systemd or kernel, you prefix debug with systemd. or kernel., respectively.
I am not saying Kay has good communication skills. It seems obvious to me that kernel developers probably want to use kernel.debug if they only want kernel debug output.
Imagine you have a problem to boot the system and you do not know where the problem lies, specifying debug on the kernel command line should debug everything. As I said, Kay committed crap code that broke the most fundamental thing, however, as usual, there are two sides of the story and it took me a while to gather all required info.
I still think Linus was right, though, because he has probably made Kay think about the quality of his code, overconfidence, or both. I remember breaking something with a typo in a logging statement I did not test before committing, silly, happens, especially when it was the last little thing I added...
If you specify debug on the kernel command line, you want to debug everything,
Not at all.
If you specify the long-standing command-line parameter to debug the kernel, that's what you're asking for. systemd hijacked that parameter - which predates systemd by some while - to force systemd into debug as well. This is incorrect.
If you want to debug "everything", you need to say so. I wouldn't expect my apache and sendmail processes to go into debug mode just because I've told the kernel to; they would have their own parameters to do such. It is exactly the same with systemd - it should not have hihjacked the kernel parameter.
Imagine you have a problem to boot the system and you do not know where the problem lies, specifying debug on the kernel command line should debug everything
Nope. That means you only have coarse-grained control over your debug; aside from the probability of that becoming intrusive, thereby destroying your debugging attempt anyway, it also makes finding the correct log entry much harder - there is far more chaff.
Now it could be argued that "debug" should now be ignored, with each piece of code having its own flag - so we explicitly use kernel.debug and systemd.debug - but I wouldn't support that; "debug" has been a kernel command-line parameter for a long time, and this change would serve purely to counter to fact that systemd thinks it can re-use that parameter; that;s quite a bit of change to everyone's debugging methods just to sort out one errant coder.
Vic.
i apologise if the following is incorrect: i am unfamiliar with the exact details of systemd, but i assume that
1. the kernel starts systemd, like init, in kernel space,
2. the kernel "moves" (part of) systemd into user space, and
3. all subsequent processes derive from (user space) systemd, live in user space, and can only access kernel space via system calls to the kernel.
could someone more informed correct me if i am wrong.
the kernel log is not the system log. the kernel log is solely for logging from kernel space, not user space. it is impossible for any user space process other than init/systemd to write to the kernel log. thus, the kernel log cannot provide a denial-of-service attack vector to any process other than the kernel, any kernel module and the init/systemd process. thus, code to protect the kernel log is an inefficient superfluity if the kernel, each kernel module, and the init/systemd process do not abuse it. it is more efficient to rewrite the systemd code to not flood the kernel log than to rewrite the kernel code to protect the kernel log from "friendly fire".
note also that the kernel log is written to a buffer, not the file system, because the kernel logs immediately after booting starts and before any file system is mounted. once a file system is mounted, and a system logger started, the kernel logger can (and usually does) flush and write the kernel log to the system log. systemd fills the kernel log so early in the boot sequence that no system log is available, and any attempts to extend the kernel log to accomodate systemd will ultimately be limited by the ram available.
finally, linus torvalds may have a colourful vernacular but i am reluctant to judge the man by his portrayal in the media: i think the image of a ranting torvalds has become something of a cliche. more indicative is that the linux project has never forked. many extremely good developers contribute to linux, and if linus torvalds were as insufferable as many paint him, it would not be beyond the abilities of any of them to execute git clone.
@Tom 13
"the kernel needs to protect itself from this kind of idiocy"
As has been explained elsewhere in these comments, there is no problem with the kernel. It worked just fine. It was systemd (before it eventually fixed this bug) that got itself into an infinite loop and failed to complete the system startup. Spewing out endless garbage to the kernel log was more of a symptom than the cause of the failure.
We used to use the Unix sysv init. This(sort of) loads a shell, mounts the root filesystem then uses a bunch of scripts to start the initial processes in the right order. The idea of systmd is that once it is running, you can just start and stop processes at will. Systemd is supposed to sort out process dependencies -- e.g. making sure that the network is up before starting ntpd or sshd. The strong promotion of systemd by Red Hat employees has meant that important/vital sub-systems, such as udev, have been rewritten to accomadate systemd. This has made it increasingly more onerous for distributions not to switch fron init to systemd. Either systemmmd will mature, and get developer tools and a workflow, such that it can be maintained without screwing other projects, or it will cause so much pain that it has to be replaced. In any case I hope something structurally less arcane can be introduced that fulfils the auto process dependency advantage of systemd.
After a very long time I again feel the need to add a few words here.
First, Linux is totally relevant as the core of Android. It provides all the stability and reliability both developers and users expect. Android is bound to be THE dominating OS.
Secondly, all the stability and reliability directly comes from Linus Thorvals being an asshole then and now. If he were a "shrewd businessman" he would not alienate developers contributing mediocre stuff. The Linux kernel would accumulate as much features as possible in order to generate more dollars. But he is an asshole to people who want to add cruft. Rightfully so - think of it as Quality Assurance Measure.
Finally, to generalize this - great engineers MUST be assholes. Otherwise their great machines/systems will be screwed up by all sorts of powerful, rich and nobility-descending idiots.
Torvalds is notoriously reluctant to make changes to the Linux kernel to address bugs found in other software
Too bloody right! I hate to think what's going to happen when Linus calls it a day and Linux goes to over to more conventional governance. I really resent having to code around a load of shit provided by A. N. Other piece of software purely because nobody's got the balls to put their foot down and tell ${CEO's_fave_project_this_week} that they need to put their underpants back inside their trousers and fix some bugs before the next round of bells and whistles gets piled onto it.
That 95% of the commentors on this thread have read ONLY the Reg Article. C'mon people. This is the Reg. It's renowned for it attention grabbing headlines. Do you really believe its exactly as they say?
Having actually read up on this, it's clear LinusT is pretty much in the right here. Kay has a reputation for producing buggy code then telling everyone else it's their code that is buggy. That is an untenable situation when it comes to the kernel. In this case, Kay used a kernel specific command line option, 'debug' and hijacked it for systemd. So if you turn on kernel debugging, it also turned on systemd debugging, flooding the *kernel* logging, and stopping booting in some circumstances. This debug option has been in the kernel for years. He is the first person to break it through clear misuse.
He also immediately claimed it wasn't his problem. Clearly, he didn't use the systemd namespace (i.e. systemd.debug) that is ALREADY used elsewhere for other command line parameters. So, yes, it IS his problem.
That said, the kernel devs are now looking in to some sort of rate limiting for kernel messages to protect against this sort of moronic behaviour.
I thought the whole _point_ of systemd was to cause as much pain for users and other projects as possible. I don't even use systemd and I've had to several changes to all my Linux installations as a result of dictates from systemd developers.
The Register (among others) seem to be drawn towards highlighting Linus' reaction to this particular incident and the short term consequences of it - "systemd have issued a patch and the world is OK" - implying (not for the first time) that Linus has overreacted.
Some basic research on Linux discussion boards (including the bug entry on the systemd) would highlight that this issue doesnt just include the 2 people involved - but teams of developers which have had a less than rosy relationship - particularly when integration issues crop up - and the affronted parties have been "controversial" in their system changes on a number of occasions.
Is the Register peeved because Linus may actually be enforcing some sort of hierarchy?
There's missing context here.
Sievers' treatment of systemd bug reports is poor, usually closing them or pointing the finger elsewhere. For example the journal logging bug, which flooded messages to the syslogger, locking up systems upon reboot; or the shutdown bug, where shutting down a system whilst shutting it down prevented future logins after the reboot.
In both of those cases people where left with non-functioning systems and repeated bug reports being closed until it was undeniable that the fault was with systemd. This behaviour would so delay bug finding that users were left with unusable Fedora installations for weeks.
With the kernel issue he's simply struck a community which, informed by those previous issues, put its foot down promptly and firmly.
To be fair, Kay has a habit of not fixing his own bugs. And this problem WAS his bug.
From what I understand of the bug, the system deadlocked - systemd is the init process, and it is SUPPOSED to record and empty the log buffers... But it was spewing so many messages that it never got around to start the recording before the system deadlocked.
Linus doesn't have problems with programmers doing stupid things once. Even twice.
But that third time, the programmer should have learned - or should get out of the field.
On the one hand I admire the guys passion, and that he still has that passion after all this time despite collaborating with others (which when you are passionate can be hard) but on the other hand it will come across as a childish rant to outsiders and reflect badly on the Linux developer community.
This post has been deleted by its author
Interesting that the techies think you have to swear to make it clear in a short, written, English sentence that the sentence poses a rhetorical question.
For me it's a matter of style. Just as in programming there are styles or writing, and elegance is always better than clumsiness. For me an interpolated curse or "lol" is an explicit GOTO used to terminate a loop.
Curse in person. Learn to write like Terry Pratchett to make a telling point in print without the need for bleepicons.
And yes. The pun was intended (for those that caught it).
Used to work with Linus at a startup. Linus was smart, funny, respectful, and seemed somewhat shy. Among his co-workers, he was well respected and liked. Never detected any ego.
If he dropped the F-bomb on this guy, this guy had it coming.
Anecdote: I was hosting Sony software engineers from Japan in a meeting in our Santa Clara HQ. One of the engineers wanted to have a T-shirt autographed by Linus. During a break, I went to Linus and made the request and asked if he could come by and say "hello" to the Sony guys. Linus dropped in later and brought the signed T-shirt and other gifts for the Sony guys. It made their week. Impressed me too!