Slashdot Mirror
Cox on Torvalds and Linux Kernel Development
sebFlyte writes "Alan Cox' speech at FOSDEM sounds like it was interesting... according to this ZDNet report on it he has some interesting views. For one, he says: 'Linus is a good developer, but is a terrible engineer.' He also has a few digs at Torvald's methods surrounding security fixes, and some other interesting insights in the kernel development process: 'Sometimes you see a fix and think "this is perfect, move my fix into the kernel tree." Later you think, "I must have been drunk. Don't apply that patch."'"
than one guy running the show?
Linus is the linux ruler.
(forgive the funny subject, I'm refering to tracking the dynamic elements of a piece of code):
I've written some code, and try to visualize how my code will run, stepping through each section in order.
The question I have is, is it still possible for these kernel gurus/hackers to effectively have the kernel and all its nuances inside their head, fully functional at a theoretical/experimental level? Or does development at this point consist of sub groups that are specialized and don't require a level of understanding to 'run the kernel in your head'?? If this is a fantasy of the past due to current complexity, when did the change occur?
-thanks
Silly Rabbit: tricks are for kids.
Calling Linus an engineer is like calling Gandhi a politician... you need to look up just a little to judge how badly you underestimated the impact of the man and his followers.
Gandhi had his moments of pettiness and just plain tom-foolishness, but the sum of his efforts changed the way people gain power back from those who would usurp it for their own.
In a different, yet no less trivial way, so did Linus (although I would not call him Mahatma).
You can't talk about Wikipedia's flaws on Wikipedia
I was going to mod, but decided to jump in, instead... I don't know Linus, I'm not a kernel-demigod, and you may know a lot more about him than I do. And while I'm a linux-enthusiast (and therefore an admirer of all the work that goes into it), I'm not a groupie who automatically jumps up to defend the Order of the Penguin. With that said, I don't see how "contemporary ideas" have anything to do with his ability to manage and guide the development of an OS. I've read correspondence about kernel issues (as I've come across them), and it always seems to me that he tries to keep it simple and direct. "Does it work?" and "Will it screw things up later?" appear to be the underlying themes...very admirable ones, in my opinion. Even more to the point: Why should anyone care if he has little or know knowledge outside his project? (And it appears to me that he has a lot of experience...but I can't/won't try to rattle off his resume. See above.) If I have to have brain surgery, I don't give a damn whether or not my surgeon knows how to do an appendectomy; he's got one job to do, and that's all I care about. Well-rounded educations, backgrounds, etc. are great when your project has to cover a wide range of issues. (Ever get involved in a government software project? It's a nightmare!) But if your needs are specific, then the more of an expert you are in that one area, the better off you'll be. To me, he's a smart guy doing a pretty good job of herding cats. 'Nuff said.
Never confuse movement with action. --Hemingway
Oh you have asked him about "OO" yourself, "PimpDawg"?
Have you ever even looked at the Linux kernel source? The VFS is a better OO abstraction than anything you will ever come up with in your lifetime.
How did this get modded above -1?
Linus isn't running the show. He's not paying anybody, he can't fire anybody, he can't make anybody drop one project or idea to work on another.
He can direct some developers to do something and they can tell him to take a hike, or they can do it because they think it's a good idea.
More often, though, there are just many ideas (patches, development threads, what have you) to choose from and Linus "rules" by choosing which goes into his kernel.
The cathedral is about direction. That isn't what Linus does -- he just selects what is best from what the bazaar has produced.
(Sure, he may also make suggestions and remarks that indicate what his selection criteria are, and that may in turn influence kernel developers, but that doesn't prevent someone from coming up with an even better idea that Linus hadn't considered before and changing Linus's mind. That doesn't happen in a cathedral -- do you think some workmen with a brilliant but different idea for St. Paul's would have been paid attention to by Christopher Wren?)
-- Alastair
I *know* this is offtopic, but as far as I know the concept of the Flawed Hero comes from the greek tragedy, and I wouldn't ever call it characteristic of contemporary literature, which is more fond of anti-heroes methinks.
O make me a mask
That's interesting. Thanks for the link!
The most interesting part to me is that he uses a Mac running Linux now. For a long time the story was that he ran SuSE (I'm assuming on x86). I'm a SuSE user, and some of the enthusiasts in SuSE-land used to use this as proof of why SuSE was "the best". I just laughed. I wonder what distro he runs now.
I don't know what other people think of this, but I think it's funny!
I'm not a coder. The cosest I get is some bash scripting, which I haven't had to do in a while. But hearing that even some of the greatest coders (who aren't bound to a company policy to keep mum) sometims screw up, makes me feel good... It just cracks me up that there are those moments in life where even Alan Cox and Linus Travolds say 'What the $#@%! was I thinking?'
And the best part? It's all visable to all the other developers. Thank goodness. I'd hate to know what kind of hairballs are in other complex, closed source software... that never get looked at by more than the core developers.
GPL + True dictator = much greater chance of code fork
I think, therefore I am. I think?
Engineers make thing from whats avaiable to produce the best design for the end result Developers create new ideas and methods that seem really cool Programmers code the idea's into a working product It's even better when you can get a person who good at all 3, more likly there good at 1 and 3 or 2 and 3
I was once given some advice about interviews from my Dad, who said something along the lines of: often, the interviewer is looking at whether you know your own limitations. For example, they will often continue to probe about a particular issue, and if the interviewee keeps saying that (s)he'd keep trying to fix the problem, that a Bad Thing(tm). The Correct Answer(tm) is to say, after a while, "That's when it becomes your problem". After a certain amount of time with a problem, you should always defer it higher-up.
Seeing as GNU is mentioned zero times in the article, they probably cut it off to avoid having to explain to people what GNU was. Even though it is in Linux magazine I guess some readers may not know what GNU is, and a discussion of GNU is probably (depends on how you look at it) irrelevant to an interview with Linus Torvalds about kernel development.
HAL 7000, fewer features than the HAL 9000, but just as homicidal!
Mathematically compliment is opposite sides of an angle. if one is acute then the other must be obtuse. They compliment each other like the chinese ying and yang.
They balance one another because they are different or compliment each other. THis is different from the compliment that sings the praise of someone else.
Mahatma is a title meaning something like Great Soul or conveying a respect for their wisdom.
Ghandi's first name was Mohandas; the title was later given to him and it is a common slip to think his non-family name was Mahatma.
"The money trail will probably end up deciding the winner."
It's not a competition. It is a non-zero-sum game. With forks, everybody gets to be the winner, because everybody gets to use what they like best.
The idea that GNU/Linux somehow needs IBM is what SCO's entire lawsuit hinges on, btw.
Actually, the design of St Paul's and the surrounding area was the result of a complex interplay between Wren, the king, the church and the people:
Wren -- wanted to create baroque-style city center
King -- wanted to save money
Church -- gothic all the way, baybee
People -- wanted convenience and dense business development
In the end, the group that came closest to getting what they wanted was the people; that's why London has no great big boulevards like those of Paris (the people valued lots of living space above ease of riot control). Wren, the man with the 'brilliant but different idea' used the King's negotiating weight to slip change after change past the Church, but he couldn't change the design completely which is why it's a hybrid gothic/baroque.
Thus the design evolved by consensus, negotiation, and balancing bright ideas against established needs. Incidentally, I never used to like St. Paul's, but now that they have cleaned it all up and redeveloped the area north of it with some excellent postmodern work, it's looking really good. Weather still bad though.
Going back to the subject of software, I really don't think this whole Cathedral/Bazaar analogy was well chosen in the first place (especially as gothic cathedrals were striking examples of community efforts, as has been pointed out elsewhere). Major projects and projects that depend on a central authority can be developed in just as fluid a way as small, distributed ones -- if you have the right people. If you don't have people like Wren and the King, if you don't have people who change their minds when they find a brilliant idea, there can be problems.
Whence? Hence. Whither? Thither.
A guy who exclusively writes his weblog/homepage in Welsh calls someone difficult to work with...
Conformity is the jailer of freedom and enemy of growth. -JFK
The most probable reason for this might be code portability. If you force yourself to write code on a PPC machine then the odds are much better that you'll end up with portable code (as you already know a bazillion x86 users will be testing your code anyway.)
Microsoft did something similar with Windows NT: The x86 version was only first compiled much later in the development process. The devs all used PPC, MIPS, or Alpha machines to do their development work.
Natural != (nontoxic || beneficial)
timothy sensationalized this headline by taking Cox's words way out of context.
He shows a conflict when none exists, as others have pointed out. please RTFA to stop his apparent FUD/character assination attempt.
a good example of skewed news.
He's actually been using PPC for about a year now, IIRC; he mentioned on the mailing list that he had switched shortly after, when there were some x86 arch-specific changes that he couldn't test.
It's kind of amusing: he doesn't really get involved with the PPC arch, and continues to work on the x86 arch, just because he knows a lot of the low-level x86 stuff from before and hasn't bothered to learn the low-level PPC stuff.