"I realize that getting the POWER people to accept that they have been total morons when it comes to VM for the last three decades is hard, but somebody in the POWER hardware design camp should (a) be told and (b) be really ashamed of themselves."
The Linux kernel is Linus's, doesn't matter who contributes code to it. He can say anything he damn well pleases about or concerning it right or wrong, doesn't matter. Anyone who thinks they can better his efforts and accomplishments should give it a try. If it were not for Linus you morons would not even have a Linux kernel period.
If it were not for Linus you morons would not even have a Linux kernel period.
Yes i don't have a linux kernel because i came from puff daddy bill and steve camp period.
As someone who started working on kernels with PWB Unix in 1981, I think it's fair to say that I know a thing or two about kernels. Linus is occasionally way off the mark, and he has been known to shoot from the hip*. However, I've watched as he listened, learned, and changed his mind about things once he had a better understanding, and he even does a good job of admitting he was wrong in the first place. So, even though you've probably never, ever, hit the send button when you didn't realize you didn't know what you were talking about, let's give him the consideration you'd want if you ever did make such a mistake.
* - For those who don't speak Western movie slang, that means he shoots too fast to have much chance of hitting the intended target.
I programmed POWER for many years and I really like it. It is hard though to live in a world where people think that x86 is a completely sane design just because everyone has one.
I do feel Linus' pain to some degree though, not on this issue specifically, but POWER is an odd beast if you are used to x86, I do also think he is overreacting.
Maybe he, Theo de Raadt and Ulrich Drepper should be placed in a room and the last one standing wins, like Developer Death Match .. or something.
The winner fights Stallman.
Yes Linus could be a bastard from time to time (heck he himself said so), and sometimes is too quick to piss off others, but from what I've seen for the past few years, he admits his own mistakes as quickly, if not more. I think that's a person that I can trust.
True most of the kernel code is not written by him now, but that's a good thing. He still sees the overall picture and keeps the wheel rolling. That's much more scalable than, say, seven years ago.
Before going off on a ridiculous rant about something you aren't familiar with, it'd be a good idea to do a little background check on the subject first.
"Before going off on a ridiculous rant about something you aren't familiar with, it'd be a good idea to do a little background check on the subject first."
The very worst part of this whole thing is not that Linus idn't familiar with the details of the POWER architecture, but the fact that the performance hit he was measuring was due to the implementation of the machine specific MM code for the arch in Linux.
This, from the guy that so many people trust to have a complete understanding of the OS resulting from the project that he leads. There are reasons that vendor kernels are more often than not patched heavilly; the vanilla kernel is maintained by a guy who knows very little about it; merely merging patches from the people who do the actual work.
You are familiar with all several dozen architecture Linux supports, including their arch-specific code?
Linus has a very good grasp of some parts of the kernel and a weak grasp of other parts. There is nobody who has a full and accurate grasp of the entire kernel. Too large, too many archs, too many drivers, too many sub-systems (all of which are necessary to get the job done)...
Linus' job isn't to understand the whole kernel. NO leader's job is to know everything. This is precisely why Linus delegates so much of the work to others, and that is exactly what EVERY leader should do - anyone who doesn't is an idiot.
Linus might spout off about something being dumb on occassion, but do note that (a) he doesn't block commits from the people who know more about the topic than he does, and (b) when he's told he's wrong he listens and updates his working set of knowledge, unlike most people who just get pissed and fume whenever told they're wrong.
Maybe Linus should have just asked what the issue was in more detail instead of spouting off on POWER without having a full picture. Maybe he didn't know that he lacked the full picture (it's hard to ask for the right information if you aren't aware that the information you have is wrong). It wouldn't have changed anything, though, since either way he got the correction information and everything kept moving along nice and brisk and without problems.
It has absolutely nothing to do with not knowing about the kernel, and everything to do with misunderstanding another hacker's message.
I take that you always understand messages completely, especially when they are in writing?
Read the follow-up posts... Linus doesn't have more than half a clue what he's talking about half the time, and has no problems spouting off crap like this.
It's a very good thing that 98% or so of the modern Linux kernel isn't writen by him. He's completely oblivious to a wide range of topics, and he comes across as being completely off his nut.
The sooner Linus retires, the better.
Leaders tend to be a little obnoxious sometimes. Expect it, accept it, move on...
You realise that Linus is speaking of HARDWARE and you are answering about a SOFTWARE problem, didn't you? POWER is (one of) the PPC microprocessor in this case.
Now having a extremely complex virtual memory system with up to 4 level of indirection
in the page table is the oppposite of any KISS principle, calling for bugs and random
efficiency behaviour; and there is not a lot of people on earth able to write some
acceptable software to manage it - not even speaking about optimisation or adaptation
to a special workload.
Ohhh, so misunderstanding another person's post is now enough reason for being called "He's completely oblivious to a wide range of topics"? Linus goes off quickly (Like most central OSS people does, I guess it comes with the job).
How much I love these quotes, I think that maybe they're a bad idea. People read "internal" messages, and get the wrong ideas. Most of these posts are send off after 10 seconds of thought, it's partly because of that speed that things get resolved so quickly in the kernel.
They really shouldn't have to moderate their "just speaking my mind" messages, to avoid getting jumped by a bunch of morons that have no idea what they are talking about.
You ready to take up his job then? Because unless you can do anything close to what he does please keep your comments to yourself. I'm no fanboi but Linus deserves a lot of respect and appreciation for the work he has done and continues to do. If you don't like it, I hear there's this young upstart out of Redmond that's making some serious ground. Micro Soft or something like that...
Well, Linus flames others.
So he needs to accept to getting flamed too.
And RMS trolls a lot of half-truths as well. So he deserves to get flamed as well.
But we need to base flames with solid reasons.
You must have missed Paul Mackerras's reply. Linus totally missed the boat here.
Not only that, but as paul pointed out, the 21% overhead is in the LINUX KERNEL, not the CPU. So in fact in Linus' own words:
I realize that getting the Linux people to accept that they have been
total morons when it comes to VM for the last two decades is hard, but
somebody in the Linux VM design camp should (a) be told and (b) be
really ashamed of themselves.
Is this a Linux 2.6 kernel or what? Becasue 21% overhead from TLB handling on
something like gcc shows that some piece of software is absolute crap.
May I suggest people working on Linux try to fix this some day, and in the
meantime people outside should probably continue to use Windows or AIX
until the others can get their act together.
"You ready to take up his job then?"
I also have the ability to badmouth people with nothing to back up what I'm accusing them of! I also have the ability to merge other people's work into my own git tree just like Linus does.
The only thing I can't duplicate is the releasing of a non-functional, simplistic Unix-like kernel in 1991. Lack of a time machine you see.
Modern Linux is mostly written by people OTHER than Linus. Linus is a mouthy git that I wouldn't trust managing a hot-dog stand, let alone one of the biggest and most widely used open source projects.
Sign me up dude.
So it's a funny thing that after all these years, people are still following his leadership. And not yours. Whoever-you-are.
How many people are following your fork?
Wow, you're an impressive example of "delusions of grandeur".
Maybe you need help:
Erm, you're one of those simple folks aren't you?
Why does someone who doesn't respect other intelligent people with more informed opinions deserve our respect? He frequently spouts off at the mouth and is incorrect on technical issues. Linus is little more than a figurehead. Anyone who is very close to actual Linux development knows as much.
And you my friend are obviously NOT close to kernel development. Linus has this insane ability to remember code line by line that he hasn't looked at in a year or so. If anyone understands the big picture of how everything fits together it is Linus. You have no clue what you're talking about Anonymous. On the other hand, he is quite opinionated but it is his baby. He is perfectly entitled to tell them it sucks.
Here is a nice image about you: http://craphound.com/images/xkcdwrongoninternet.jpg
Except he's frequently wrong and disrespectful. Is he entitled to tell people they suck when he's wrong?
The point isn't whether he knows linux well or not. It's that he's an ass.
You can't be right all the time, but he's got a sharp mind and he debates his ideas very energetically. What's wrong with that?
Do you understand the amount of work necessary to handle patches coming like snowball from everywhere?
Could you manage that workload and still be coding like crazy?
Linus has done a great job in attracting top developers to the Linux kernel and his contribution to it, both in code and management, cannot be forgotten.
btw, click here for data regarding contributions to the kernel.