Ideally, tickless will obviate the need to play with HZ. I think this a good overall strategy: reduce the number of knobs, and replace them with intelligent algorithms that work for everyone.
I wonder how big a deal it would be to allow renice(2) to renice a process up to 0, or some slightly higher value, if the target process and the one calling renice() are owned by the same user and the process doing the renicing inherited some special capability bit -- that way, the window manager really could play with priority as windows come in and out of focus, but programs couldn't renice themselves to normal priority arbitrarily.
From: Ingo Molnar Date: Sun Apr 15 2007 - 11:06:49 EST
i partially replied to that point to Will already, and i'd like to make it clear again: yes, i rejected plugsched 2-3 years ago (which already drifted away from wli's original codebase) and i would still reject it today....
that's the extra complication i didnt like 3 years ago and which i still dont like today. What the current plugsched code does is that it simplifies the adding of new experimental schedulers, but it doesnt really do what i wanted: to simplify the _scheduler itself_. Personally i'm still not primarily interested in having a large selection of schedulers, i'm mainly interested in a good and maintainable scheduler that works for people.
so the rejection was on these grounds, and i still very much stand by that position here and today: i didnt want to see the Linux scheduler landscape balkanized and i saw no technological reasons for the complication that external modularization brings.
Since Linux is released under the GPL, every module must be under GPL compatible licenses, irrespective of whether they contain, or depend on, any GPL'd code.
That's not true. Non-GPLed kernel modules are "tolerated" by the Linux kernel developers, and in principle, a ZFS module could be created and loaded with no problems, assuming it doesn't rely on GPL-only symbols. AFAIK, the VFS doesn't have many of those.
What can't happen under the CDDL is the ZFS code being included in the kernel source tree the same way XFS, ext3 and so on are. That doesn't mean you can't maintain and distribute a module separately! The only reason a ZFS module doesn't exist today is that nobody's gone through the trouble of creating one.
Is it legal to run an extension cord from one's house to the curb? I live in a normal house, but there's no off-street parking. Most of the time, I'm directly in front of my house, so an extension cord would be all right.
College students (except for the rare 17-year-old) are adults, and we should never forget this. A college has no business filling any portion of a parent's role, period.
Forgive my ignorance, but how would using a SAN have helped in this situation? Are you proposing that a single SAN storage net span multiple (remote) physical locations? And with SAN, can't a disk only be used by one computer at a time anyway?
Sure, you could use RAID 1(+0) and put the mirrored halves at different locations, but I can't imagine that being acceptable from either a performance or a reliability point of view.
Wouldn't master-slave database replication be more appropriate for this kind of work?
I know it DOESN'T apply, but why SHOULDN'T it? The principle of assuming people are innocent until shown otherwise recognizes that it's much worse to punish an innocent person than to let a guilty one go free, and has been a hallmark of our society for centuries. Why would this argument apply any less strongly to a non-government, non-criminal system of trial and punishment? I think you're confusing legality and morality here.
No, you're not teaching your students the same thing at all. Before, you were teaching them how to create and manage a thread pool. Now, you're teaching them how to use some shitty prefab library to do roughly the same thing without understanding WHY or HOW the library does what it does.
As luck would have it, I stumbled across a twist on the captcha concept while registering for a site. Instead of asking the human user to correctly enter the word displayed in an image, it presented the user with a grid of images. About half of them were of cars. The other half were cats.
The site just asked the user to check off each image representing a living thing.
Simple, and brutally effective against current AI. I can think of various tricks one can use to make the comparison more difficult as well.
How long until we're using the kind of tests we saw in Blade Runner?
Fair enough. It was just an idea. How would you increase the quality of the average programmer though? Just increasing the salary in accordance with demand seems to result in an influx of programmers who don't give a damn about quality, and who are only in "computers" for the money.
You know you've played too much Total Annihilation
on
Floating Wind Turbines
·
· Score: 2, Funny
You know you've played too much Total Annihilation when your first thought upon reading the headline is "why not build underwater fusion reactors?"
Some programmers are forklift operators; others are skilled masons, and others, architects and engineers in their own right.
I think the root of this whole debate is that computer science has forked into two disciplines, and we need to recognize that. Also, it's possible for someone to be acquainted (and even skilled) in both areas simultaneously, which further confuses the debate. My friend at CMU said that the students there are almost strictly classified into "leet hacker" types or math-friendly ones that disdain actual coding. Both have value, but only the latter should be considered computer scientists.
Oh, simplification and abstraction are great things. The problem with brutish languages arises when they try to simplify problems to a greater degree than is possible. Some tasks are inherently complex, and if the language punts, then the user of the language has to pick up that complexity, often in subtle ways. Now, PHP doesn't encourage this anymore, but consider the magic_quotes_gpc feature. It's an archetypal example of that kind of thinking.
I can't comment on VB.NET, having not used it.
C# is not a bad language [though I'd still prefer Lisp or Python;-)] either.
As for new data structures --- in my experience, I've found that it's the newer programmers that will attempt to implement their own data structures, sorting algorithms, etc. instead of using standard language features. I think every programmer goes through a microptimization phase, during which he writes absolute crap. I know I did.:-)
About Javascript --- yes, it has some nasty gotchas, but it has closures. That's its saving grace. Also, the prototypeish object system can be used to implement a standard inheritance hierarchy; google: there are plenty of examples of how to do this. I'm not saying Javascript is perfect, but only that we should consider ourselves lucky that it isn't much, much worse.
Emacs' python-mode already handles a bunch of common cases. The problem is that often, the visual indentation is ambiguous.
You know, that would be a great technique if there were ever an international obfuscated Python coding contest --- write code that looks like it does one thing, but actually does something completely different based on indentation.
Oh, forgot to mention: teaching algorithms and data structures in Java makes it difficult to communicate some of the subtler points.
Consider a linked list. Many Java programmers, especially new ones, are confused about the difference between a reference to an object and the object itself, since the difference isn't explicit like it is in, say, C. Really learning about data structures requires keeping these fine distinctions in mind.
The timecube guy gave a talk at Stanford. Giving a talk means nothing.
He does? I didn't get that from the article. Can you site something in that review or in the book?
First, it's cite, not site. And no, I can't cite anything from the article that supports exactly that supposition. However, judging by the overall tone of the article and the reaction to it, that's what the author was getting at.
I'm not sure if O(logN) is really an algorithm either.
I'm collecting quotes from the discussion of this article, and the above just made it in. Did you read what I wrote? Do you understand what an algorithm is? Complexity is a property of an algorithm, not an algorithm itself. Some scheduling algorithms might be O(logN), others O(1), and still others O(n^2). For convenience's sake, I use the word "scheduler" to refer to the process that does the scheduling, including the algorithm it uses.
We already have an algorithmic-y (tm me) field and that's called mathematics. Why are you trying to shoehorn it into CS?
Computer Science has always been a branch of mathematics. You are trying to adulterate it by removing the mathematics and replacing it with non-mathematical, non-science pragmatic material. That material has a place, agreed, but it's not in computer science. People who want to learn only about the latest object design fads should not call themselves computer scientists.
Yeah your typical VB programmer can't, but who cares?
I, for one, care. These VB programmers are writing production software that affects not only themselves and their companies, but all of us. How many lives have been ruined as a result of identity theft caused by easily identifiable security holes? Programmers (or mechanics, for that matter) have a moral responsibility to be competent in their field, since very often their clients have no way of telling the bad apples from the talented folk.
It's Clark's third law: "any sufficiently advanced technology is indistinguishable from magic."
Your analysis is spot-on, and I see this phenomenon happing all around me. It's just sad that professions in the field are beginning to succumb to this mindset.
I used to think that tabs were better than spaces too. But after dealing with one misaligned program too many, I changed my mind.
The problem is that there are two uses for left-margin whitespace: logical indentation and aesthetic lineup, and people use a mixture of tabs and spaces for both.
If people were to follow the rule that only tabs were to be used in doing the logical indentation, and only spaces were to be used for, say, lining up function arguments, we could still all use spaces and adjust indentation how we wanted.
But people don't do that, so we're stuck with using spaces if we want our code to be consistently presentable.
Slashdot Mirror
Ideally, tickless will obviate the need to play with HZ. I think this a good overall strategy: reduce the number of knobs, and replace them with intelligent algorithms that work for everyone.
I wonder how big a deal it would be to allow renice(2) to renice a process up to 0, or some slightly higher value, if the target process and the one calling renice() are owned by the same user and the process doing the renicing inherited some special capability bit -- that way, the window manager really could play with priority as windows come in and out of focus, but programs couldn't renice themselves to normal priority arbitrarily.
Source: http://ussg.iu.edu/hypermail/linux/kernel/0704.1/
That's not true. Non-GPLed kernel modules are "tolerated" by the Linux kernel developers, and in principle, a ZFS module could be created and loaded with no problems, assuming it doesn't rely on GPL-only symbols. AFAIK, the VFS doesn't have many of those.
What can't happen under the CDDL is the ZFS code being included in the kernel source tree the same way XFS, ext3 and so on are. That doesn't mean you can't maintain and distribute a module separately! The only reason a ZFS module doesn't exist today is that nobody's gone through the trouble of creating one.
Is it legal to run an extension cord from one's house to the curb? I live in a normal house, but there's no off-street parking. Most of the time, I'm directly in front of my house, so an extension cord would be all right.
College students (except for the rare 17-year-old) are adults, and we should never forget this. A college has no business filling any portion of a parent's role, period.
Forgive my ignorance, but how would using a SAN have helped in this situation? Are you proposing that a single SAN storage net span multiple (remote) physical locations? And with SAN, can't a disk only be used by one computer at a time anyway?
Sure, you could use RAID 1(+0) and put the mirrored halves at different locations, but I can't imagine that being acceptable from either a performance or a reliability point of view.
Wouldn't master-slave database replication be more appropriate for this kind of work?
I know it DOESN'T apply, but why SHOULDN'T it? The principle of assuming people are innocent until shown otherwise recognizes that it's much worse to punish an innocent person than to let a guilty one go free, and has been a hallmark of our society for centuries. Why would this argument apply any less strongly to a non-government, non-criminal system of trial and punishment? I think you're confusing legality and morality here.
No, you're not teaching your students the same thing at all. Before, you were teaching them how to create and manage a thread pool. Now, you're teaching them how to use some shitty prefab library to do roughly the same thing without understanding WHY or HOW the library does what it does.
Why would any employer disallow internet radio? I listen to it all the time at work, and I'm more productive for it.
*sigh*
Every single damn patent on that list has abundant prior art.
Would you consider assert()s to be sufficient checking for impossible conditions?
Citation?
As luck would have it, I stumbled across a twist on the captcha concept while registering for a site. Instead of asking the human user to correctly enter the word displayed in an image, it presented the user with a grid of images. About half of them were of cars. The other half were cats.
The site just asked the user to check off each image representing a living thing.
Simple, and brutally effective against current AI. I can think of various tricks one can use to make the comparison more difficult as well.
How long until we're using the kind of tests we saw in Blade Runner?
Fair enough. It was just an idea. How would you increase the quality of the average programmer though? Just increasing the salary in accordance with demand seems to result in an influx of programmers who don't give a damn about quality, and who are only in "computers" for the money.
You know you've played too much Total Annihilation when your first thought upon reading the headline is "why not build underwater fusion reactors?"
Some programmers are forklift operators; others are skilled masons, and others, architects and engineers in their own right.
I think the root of this whole debate is that computer science has forked into two disciplines, and we need to recognize that. Also, it's possible for someone to be acquainted (and even skilled) in both areas simultaneously, which further confuses the debate. My friend at CMU said that the students there are almost strictly classified into "leet hacker" types or math-friendly ones that disdain actual coding. Both have value, but only the latter should be considered computer scientists.
Oh, simplification and abstraction are great things. The problem with brutish languages arises when they try to simplify problems to a greater degree than is possible. Some tasks are inherently complex, and if the language punts, then the user of the language has to pick up that complexity, often in subtle ways. Now, PHP doesn't encourage this anymore, but consider the magic_quotes_gpc feature. It's an archetypal example of that kind of thinking.
;-)] either.
:-)
I can't comment on VB.NET, having not used it.
C# is not a bad language [though I'd still prefer Lisp or Python
As for new data structures --- in my experience, I've found that it's the newer programmers that will attempt to implement their own data structures, sorting algorithms, etc. instead of using standard language features. I think every programmer goes through a microptimization phase, during which he writes absolute crap. I know I did.
About Javascript --- yes, it has some nasty gotchas, but it has closures. That's its saving grace. Also, the prototypeish object system can be used to implement a standard inheritance hierarchy; google: there are plenty of examples of how to do this. I'm not saying Javascript is perfect, but only that we should consider ourselves lucky that it isn't much, much worse.
Case in point: PHP's hybrid list/dictionary structure makes many otherwise-common cases tricky.
/ php/#section_5
See http://www.ukuug.org/events/linux2002/papers/html
Emacs' python-mode already handles a bunch of common cases. The problem is that often, the visual indentation is ambiguous.
You know, that would be a great technique if there were ever an international obfuscated Python coding contest --- write code that looks like it does one thing, but actually does something completely different based on indentation.
If you don't care about my opinions, why are you reading them?
Oh, forgot to mention: teaching algorithms and data structures in Java makes it difficult to communicate some of the subtler points.
Consider a linked list. Many Java programmers, especially new ones, are confused about the difference between a reference to an object and the object itself, since the difference isn't explicit like it is in, say, C. Really learning about data structures requires keeping these fine distinctions in mind.
It's Clark's third law: "any sufficiently advanced technology is indistinguishable from magic."
Your analysis is spot-on, and I see this phenomenon happing all around me. It's just sad that professions in the field are beginning to succumb to this mindset.
I used to think that tabs were better than spaces too. But after dealing with one misaligned program too many, I changed my mind.
The problem is that there are two uses for left-margin whitespace: logical indentation and aesthetic lineup, and people use a mixture of tabs and spaces for both.
If people were to follow the rule that only tabs were to be used in doing the logical indentation, and only spaces were to be used for, say, lining up function arguments, we could still all use spaces and adjust indentation how we wanted.
But people don't do that, so we're stuck with using spaces if we want our code to be consistently presentable.