Why not maintain a symbol database of every symbol exported and required by every package (sure, it might get pretty big, but most package management systems have bloat).
That way, appfoo could register as requiring symbols bar, baz and qxt. If libfred-dev and dev-libgfred both supplied those symbols, either one could meet the dependency. Kind of like what portage does with the virtual/ targets, only more fine-grained.
There would still be some problems, but that at least would solve your various-package-names issue.
Ummm... no. entry.S defines every system call in the kernel. system.map defines every symbol exported by the kernel. Your kernel exports 10104 symbols (which makes me think you compiled it with debugging turned on). Of those 10104 symbols, 190 (I think that's the magic number for 2.4) are system calls. The rest are just exported symbols.
They're just names for parts of memory that make linkers a little easier to write -- in your example, anything you could do with the symbol length_code you could do with the number c0343080.
Linus' decision-making becomes the focus when there is a "tie" (for lack of a better word) between competing visions. And so what if it is? I know many people who run -ac kernels exclusively. And it's still Linux.
"Linux" is a trademark and something is "Linux" if Linus says it is; he could just say "only my source tree is Linux; the rest of you have to say your kernels are 'Linux-based'." AFAIK that's Linus's only real control on the situation, that his kernel is what will in the end be called "Linux".
Never. I just install programs, and they -- well, just work. Look, I like Linux, but it is seriously f-ed up when it comes to clean installing of applications.
I gotta call bullshit here. I mean, I'm not saying that you're lying, just that this sounds like total bullshit to me since my experience has been the exact opposite. I have really never had a dependency problem in Linux. Ever. I keep getting them, over and over again, in Windows.
I'm the same way with drivers. I cannot get hardware to work in Windows. Take my girlfriend's computer: I've put in 3 different ethernet cards (a 3com, a SBC, and a Linksys) and Windows can't load the driver for any of them. I've reinstalled windows, and the installer can't find the driver for any of the ethernet cards. They will show up in the device manager with a question mark on top of them and refuses to locate a driver for them. The same computer, I boot with Knoppix and finds the ethernet cards fine.
Now, tons of people have the opposite problem apparently, but I've really never had problems with Linux hardware and I've always had problems with Windows hardware.
Most_ of the time Windows installs out of the box with a GUI installer that requires no command line tweaking required. Even Mandrake required some command line tweaking.
Nah. Windows' installation process is about as hard as most Linuxes: easier than Gentoo or Debian, sure, but harder than SuSE. Do you honestly think most Windows users could successfully install Windows? I'd be really surprised if they could.
A state can be a republic and a democracy. Or it can be only one of them. Or it can be neither. The questions of a state's being a republic and its being a democracy are orthogonal.
A state is a republic if its chief of state is not a monarch. Therefore, the US is a republic, as is Russia, as was Iraq under Saddam Houssein. The UK, on the other hand, is not a democracy, even though all effective power is wielded by popularly-elected MPs.
A state is a democracy if the governing authority is originates from the people. This is a much more subjective judgment than declaring if a state is a republic or not, obviously, but nothing precludes a state from A) deriving its government's authority from the people and B) having a chief of state who is not a monarch.
This is the one of the few things I agree with Badnarik on: the lesser of two evils is still evil, and None Of The Above should be on every ballot, with defaced ballots counting as a vote for NOTA.
Intelligent design/creationism/etc are NOT taught in Public schools
They are in some. And in most schools the faculty is too terrified of the fundies to teach evolution at all -- witness the fact that if you ask most high school students, they'll think evolution means "man descended from monkeys". And in fact they can't even tell you the difference between a monkey and an ape.
What bugs me is not that parents with religious conviction are trying to have a say in the education of their children -- I'm for that. What bugs me is that people who cannot define the word "allele" have the gall to spout their opinions on evolutionary biology and demand their arguments be treated as having equal weight with scientific conclusions. That would be like me going to a church and saying "you know, I've never really read Ezekial but we need to stop using it in scripture readings because I've heard it contradicts my field, comparitive linguistics."
I don't think that the simple fact that schools are too scared to teach the theory of evolution is what breaks our schools. But, it is part of a larger trend of religious conservatives fighting tooth and nail against intellectualism in general. And that is what's killing our schools.
And all I can say is, what a collosal waste of effort by this jackass. Just write the malware in C and compile it; no need to shellcode yourself like that. If I'm reading it right, that is. Too bad the lameness filters kill the disassembly.
That is why they use a signed certificate for Windows Update, in fact. And why Microsoft PGP signs their security bulletin emails. Is there a joke there I'm missing?
To be honest, Microsoft's "trust this software?" dialog is pretty good: hard to fake and lets you view the signing certificate if you want to. The "Always trust software from these people" option kind of bugs me but I guess it's not much different from setting a key's trust level in GPG.
Crackers are an ingenious lot, and security holes are security holes are security holes. They WILL be exploited in linux sooner or later.
Will be exploited? Download the metasploit framework sometime; there are more exploits for Linux than for Solaris or Windows. But this is where the guy's point becomes important: because of how Windows deals with security tokens (here is a good place to start if you're curious), any exploit that gains access can probably execute code in the SYSTEM context.
So, of the Linux exploits that are trivially available to exploit, none can reliably execute arbitrary system code, while all of the Windows exploits can. That's not this one guy's opinion, that's just how the operating systems work.
I think Graham's ANSI Common Lisp did it pretty well, because he didn't bother trying to port the Algol-based concepts of graphs, trees, etc. but put it in terms of building s-expressions. I think you're right that until you grok the way that the data structures are your program and vice-versa, reading that statement won't help you much.
I still like Lisp as a first language because it's not pedantic (C shares this virtue but is a little too machine-focused for my tastes). Learning $OO_FLAVOR_OF_THE_MONTH first forces you to think of everything in an OO-fashion -- which is really only a good idea for a certain type of problem. For all the other problems, you're left with this hammer that makes you think everything is a nail.
By those criteria, I'd say good first languages are Lisp, C, Forth, and Perl (I'd probably add Python if I knew it). Each of them is going to shut a few doors for you, but on the whole they all lead to a pretty balanced view of what you can do and how you can think as a programmer.
I started with Logo, but that was in the 3rd grade. Really useful, to be honest. But not really as appealing to adults as it might be to kids. I'd say learn something like Lisp or Haskell first, personally.
Slashdot Mirror
Why not maintain a symbol database of every symbol exported and required by every package (sure, it might get pretty big, but most package management systems have bloat).
That way, appfoo could register as requiring symbols bar, baz and qxt. If libfred-dev and dev-libgfred both supplied those symbols, either one could meet the dependency. Kind of like what portage does with the virtual/ targets, only more fine-grained.
There would still be some problems, but that at least would solve your various-package-names issue.
Ummm... no. entry.S defines every system call in the kernel. system.map defines every symbol exported by the kernel. Your kernel exports 10104 symbols (which makes me think you compiled it with debugging turned on). Of those 10104 symbols, 190 (I think that's the magic number for 2.4) are system calls. The rest are just exported symbols.
They're just names for parts of memory that make linkers a little easier to write -- in your example, anything you could do with the symbol length_code you could do with the number c0343080.
Ummm... first off, that whole post wasn't comprehensible.
Secondly, there aren't "1000 or more" system calls. I quote /usr/src/linux/arch/i386/entry.S:
So, there are 284 syscalls. Hardly "1000 or more".Polio? Has it really been that long?
"Linux" is a trademark and something is "Linux" if Linus says it is; he could just say "only my source tree is Linux; the rest of you have to say your kernels are 'Linux-based'." AFAIK that's Linus's only real control on the situation, that his kernel is what will in the end be called "Linux".
I gotta call bullshit here. I mean, I'm not saying that you're lying, just that this sounds like total bullshit to me since my experience has been the exact opposite. I have really never had a dependency problem in Linux. Ever. I keep getting them, over and over again, in Windows.
I'm the same way with drivers. I cannot get hardware to work in Windows. Take my girlfriend's computer: I've put in 3 different ethernet cards (a 3com, a SBC, and a Linksys) and Windows can't load the driver for any of them. I've reinstalled windows, and the installer can't find the driver for any of the ethernet cards. They will show up in the device manager with a question mark on top of them and refuses to locate a driver for them. The same computer, I boot with Knoppix and finds the ethernet cards fine.
Now, tons of people have the opposite problem apparently, but I've really never had problems with Linux hardware and I've always had problems with Windows hardware.
Nah. Windows' installation process is about as hard as most Linuxes: easier than Gentoo or Debian, sure, but harder than SuSE. Do you honestly think most Windows users could successfully install Windows? I'd be really surprised if they could.
A state can be a republic and a democracy. Or it can be only one of them. Or it can be neither. The questions of a state's being a republic and its being a democracy are orthogonal.
A state is a republic if its chief of state is not a monarch. Therefore, the US is a republic, as is Russia, as was Iraq under Saddam Houssein. The UK, on the other hand, is not a democracy, even though all effective power is wielded by popularly-elected MPs.
A state is a democracy if the governing authority is originates from the people. This is a much more subjective judgment than declaring if a state is a republic or not, obviously, but nothing precludes a state from A) deriving its government's authority from the people and B) having a chief of state who is not a monarch.
So, the US is both a Republic and a Democracy.
Jesus...
Kyoto doesn't apply to developing nations like India and China.
Maybe you could start learning about the subject before you form an opinion?
...my own guess is he's more worried about al Zarqawi's inroads among Iraqis and Jordanians and wants to nip that rival in the bud, if possible.
This is the one of the few things I agree with Badnarik on: the lesser of two evils is still evil, and None Of The Above should be on every ballot, with defaced ballots counting as a vote for NOTA.
They are in some. And in most schools the faculty is too terrified of the fundies to teach evolution at all -- witness the fact that if you ask most high school students, they'll think evolution means "man descended from monkeys". And in fact they can't even tell you the difference between a monkey and an ape.
What bugs me is not that parents with religious conviction are trying to have a say in the education of their children -- I'm for that. What bugs me is that people who cannot define the word "allele" have the gall to spout their opinions on evolutionary biology and demand their arguments be treated as having equal weight with scientific conclusions. That would be like me going to a church and saying "you know, I've never really read Ezekial but we need to stop using it in scripture readings because I've heard it contradicts my field, comparitive linguistics."
I don't think that the simple fact that schools are too scared to teach the theory of evolution is what breaks our schools. But, it is part of a larger trend of religious conservatives fighting tooth and nail against intellectualism in general. And that is what's killing our schools.
Eh? .arpa is reserved as the PTR namespace. So unless you've ported http to Chaos or something...
Blame the news cycle; as of 1230 EDT it was on the front of Yahoo! news.
People love to talk about .NET but I almost never see it used.
It seems damned obvious but nobody wants to do it for some reason:
- Voter selects candidates by touching the screen.
- Computer prints out ballot -- no mismarks, no chad; a computer-printed vote.
- Voter inspects ballot to verify it.
- If the voter is satisfied, he drops the ballot in the ballot box. If not, he has a poll worker destroy the ballot and votes again.
- At the end of the day, the computer-printed paper ballots are scanned and counted by computer.
Why don't people want this? Why do they insist on keeping the numbers in memory?Well, there was an attempt to port COM and OLE to *n*x but it ended up being Gnome. Go figure.
And all I can say is, what a collosal waste of effort by this jackass. Just write the malware in C and compile it; no need to shellcode yourself like that. If I'm reading it right, that is. Too bad the lameness filters kill the disassembly.
That is why they use a signed certificate for Windows Update, in fact. And why Microsoft PGP signs their security bulletin emails. Is there a joke there I'm missing?
To be honest, Microsoft's "trust this software?" dialog is pretty good: hard to fake and lets you view the signing certificate if you want to. The "Always trust software from these people" option kind of bugs me but I guess it's not much different from setting a key's trust level in GPG.
Not exactly the box most likely to get pwned by somebody.
However, the IP block clearly belongs to Yahoo, whois 66.218.75.0 lists contact point netblockadmin@yahoo-inc.com
Anybody feel like dropping them a line to tell them they're hosting trojaners?
Will be exploited? Download the metasploit framework sometime; there are more exploits for Linux than for Solaris or Windows. But this is where the guy's point becomes important: because of how Windows deals with security tokens (here is a good place to start if you're curious), any exploit that gains access can probably execute code in the SYSTEM context.
So, of the Linux exploits that are trivially available to exploit, none can reliably execute arbitrary system code, while all of the Windows exploits can. That's not this one guy's opinion, that's just how the operating systems work.
I think Graham's ANSI Common Lisp did it pretty well, because he didn't bother trying to port the Algol-based concepts of graphs, trees, etc. but put it in terms of building s-expressions. I think you're right that until you grok the way that the data structures are your program and vice-versa, reading that statement won't help you much.
I still like Lisp as a first language because it's not pedantic (C shares this virtue but is a little too machine-focused for my tastes). Learning $OO_FLAVOR_OF_THE_MONTH first forces you to think of everything in an OO-fashion -- which is really only a good idea for a certain type of problem. For all the other problems, you're left with this hammer that makes you think everything is a nail.
By those criteria, I'd say good first languages are Lisp, C, Forth, and Perl (I'd probably add Python if I knew it). Each of them is going to shut a few doors for you, but on the whole they all lead to a pretty balanced view of what you can do and how you can think as a programmer.
I started with Logo, but that was in the 3rd grade. Really useful, to be honest. But not really as appealing to adults as it might be to kids. I'd say learn something like Lisp or Haskell first, personally.