How many hands is this new iPod going to go into? How many of them will start really doing mobile web browsing as a day-to-day activity? Mobile facebook is digital crack, that you can't leave at your home or in your laptop now.
The only code you're running in kernel space is the JVM or code generated by it.
While I'm posting. I'd like to point out that you can DTrace all the way from inside your kernel up through your shared libraries, up through your JVM, and into your java code. And it's safe to do this on a production system*.
* Not that I distrust solaris on this one, but I'd try and keep it light on critical servers, just in case;-)
Nope. Flops are worthless when you can't transmit them fast enough. What's the point of doing a couple-dozen GFlops if you have to wait really long to transmit your results? Hurry up and wait, hurry up and wait.
FTA: It's a "MIPS-like ISA with a few important and peculiar features"
I'll be interested to see what they're going to do about making it easier to program. Wire delay's going to be exposed as hops on the on-chip network. IMHO, the toolchain side's far more interesting to me than shoving a bunch of cores together on an on-die network....
Assuming they did anything interesting on the toolchain side.
I think that when they let their employees have laptops full of my (unencrypted) personal data, which subsequently gets lost or stolen, that they should bear the responsibility.
For phishing sites, etc. There are technological solutions to this sort of problem. Just require better verification than 'the domain name matches the SSL certificate'.
Carbon's what you use if you're doing C++. It's got a pretty solid API, and is in *no* way being deprecated.
Admittedly, it's a lot easier to develop GUIs in Cocoa, just b/c of the language features. But, if you want to ever ship on more than one platform, you're already doing a C++ backend.
Try out Basecamp. It's pretty straightforward, and it's more than a cheap wrapper around a database. The views are well thought out, just the right amount of AJAX to make your day go easier, and good collaboration support.
Start off free and pay if you need more than the free account provides. It doesn't get very expensive.
If that's not your thing, put down another recommendation for trac.
Because it's a terrible, terrible idea. The Mac software stack is *large*, with API compatibility going back 20 years. 3 full-size APIs are supported (bsd, carbon, and cocoa), and they're all constantly being improved by apple. Any such project, while also being an absurd waste of time, could never catch up. Not to mention all the GPU stuff they're doing these days, integrated into the window server (Quartz Extreme, CoreAnimation, etc.). Feel like extending X11 to get decent performance? I don't, and neither does anyone else.
You've already gone past the hurdle that keeps most from using Linux: buying a Mac. If you want all the linux software, just download port from http://www.macports.org/ and let it download prebuilt binaries of traditionally linux applications for your mac. The website is crap but the tool's good and the repository is active and well maintained. They run just like the linux side, only you don't have to start hating your life by using Linux as your desktop OS. Switching back from OS X to Linux is about as painful as shoving a screwdriver in your eye. There's no point.
Some corrections:
* Parallels/VMWare aren't emulating anything. They're using newish x86 instructions to let the system run 2 OS's simultaneously
* Ever consider recompiling? I mean, it's called open source for a reason.
Also, if you're gonna tinker, consider Solaris. It's free and Parallels supports it with nice X11 extensions for mouse sharing, etc. Also, it's BrandZ lets you run Linux binaries.
Frankly, it's like any other technological solution.
Right now, it's riddled with more trouble than it's worth. It'll have lots of public failures. But, each failure will lead to an improvement, and eventually it's pretty decent.
The problem most have with electronic ballots is the threat of indetectable corruption. But how much corruption do we have in paper ballot systems? Considering the cost of the counting process, we can't really use the paper trail very often. Instead, we're stuck with believing what we're told by a bunch of typically very-partisan political appointees or politicians.
It's not news when someone hacks into a box full of paper slips -- any jackass can do it. That bothers me a lot more than a team of researchers with a ballot box at home with full source code.
Electronic voting could let us all download an anonymized list of votes, and we can verify them ourselves. Cryptography could prove very useful here -- let each citizen verify that the national tally includes their vote correctly, and implement some safeguards* to make sure that there aren't any fake votes.
Computers could be used to *secure* democracy. Probably better than any other time in human history.
* E.g. have certainty in the number of votes in each area, or randomly audit votes (meaning ask that those voters affirm that those were their selections), etc.
Connectix used to do this (in v3 or so) for the mac. Emulating an x86 CPU on PPC. Basically, they just provided a pass-through OpenGL driver that hit the native driver & hardware.
For native CPU & a pass-through OpenGL stack, it should be pretty close to native speed. Only concerns are: 1. Direct3D/DirectX (what's it called these days?) -- emulating that or converting it to the native graphics driver isn't trivial. Or even a direct mapping. 2. Feature differences between implementations of drivers between the mac & windows. My guess is that most of the big boys use common code in between (especially now) with build setups & wrappers for each platform. But, who knows.
It's often not about not knowing how to set up access control at the wireless router, but getting some utterly broken wireless drivers to work with them. For compatibility's sake, I've had to turn off AC more than once. Only thing that works so far is MAC-based controls, but that requires getting MAC addresses for all the machines you want on the network, and letting anyone with a promiscuous wifi card read whatever you're doing on the net. (WEP may not be really secure, but it still takes a bit of work to crack).
Slashdot Mirror
It's always costed them ~$250 to make.
Now we'll start seeing a real mobile web.
How many hands is this new iPod going to go into? How many of them will start really doing mobile web browsing as a day-to-day activity? Mobile facebook is digital crack, that you can't leave at your home or in your laptop now.
Man I wish these things had GPS...
The only code you're running in kernel space is the JVM or code generated by it.
;-)
While I'm posting. I'd like to point out that you can DTrace all the way from inside your kernel up through your shared libraries, up through your JVM, and into your java code. And it's safe to do this on a production system*.
* Not that I distrust solaris on this one, but I'd try and keep it light on critical servers, just in case
It's 2007 and nobody cares what the term meant a long time ago to a small number of unimportant people.
Hacker = supergenius who writes virii, breaks into systems, and terrorizes the entire country from a moving tractor-trailer.
Cracker = pejorative term for white people.
Any other definitions have been obsolesced. Geez, this ranting's been going on since the late 90's, please *everyone* get over it.
It is purely bidirectional. If we don't protect the State from Church A, then it'll start bearing down on Church B.
Why do people keep talking about Church in singular terms? Freedom of Religion is per person, not per state.
Nope. Flops are worthless when you can't transmit them fast enough. What's the point of doing a couple-dozen GFlops if you have to wait really long to transmit your results? Hurry up and wait, hurry up and wait.
They hypervisor just keeps the Cell from talking to the graphics chip. To keep the game developers happy. The CPU+6 SPEs is yours.
FTA: It's a "MIPS-like ISA with a few important and peculiar features"
I'll be interested to see what they're going to do about making it easier to program. Wire delay's going to be exposed as hops on the on-chip network. IMHO, the toolchain side's far more interesting to me than shoving a bunch of cores together on an on-die network....
Assuming they did anything interesting on the toolchain side.
I think that when they let their employees have laptops full of my (unencrypted) personal data, which subsequently gets lost or stolen, that they should bear the responsibility.
For phishing sites, etc. There are technological solutions to this sort of problem. Just require better verification than 'the domain name matches the SSL certificate'.
I think the reduced customization options helps keep the reliability up. More things that could go wrong.
Unless you just mean more choices on HDDs or other commodity components. I think that's just a cost-control thing.
Check online for hints. Took me no time at all to install. Runs wonderfully -- I run it on a second monitor full-screen.
This way the techs will have to upgrade the system earlier in 2010, to make sure they can have Xmas off.
Carbon's what you use if you're doing C++. It's got a pretty solid API, and is in *no* way being deprecated.
Admittedly, it's a lot easier to develop GUIs in Cocoa, just b/c of the language features. But, if you want to ever ship on more than one platform, you're already doing a C++ backend.
Try out Basecamp. It's pretty straightforward, and it's more than a cheap wrapper around a database. The views are well thought out, just the right amount of AJAX to make your day go easier, and good collaboration support. Start off free and pay if you need more than the free account provides. It doesn't get very expensive. If that's not your thing, put down another recommendation for trac.
Because it's a terrible, terrible idea. The Mac software stack is *large*, with API compatibility going back 20 years. 3 full-size APIs are supported (bsd, carbon, and cocoa), and they're all constantly being improved by apple. Any such project, while also being an absurd waste of time, could never catch up. Not to mention all the GPU stuff they're doing these days, integrated into the window server (Quartz Extreme, CoreAnimation, etc.). Feel like extending X11 to get decent performance? I don't, and neither does anyone else.
You've already gone past the hurdle that keeps most from using Linux: buying a Mac. If you want all the linux software, just download port from http://www.macports.org/ and let it download prebuilt binaries of traditionally linux applications for your mac. The website is crap but the tool's good and the repository is active and well maintained. They run just like the linux side, only you don't have to start hating your life by using Linux as your desktop OS. Switching back from OS X to Linux is about as painful as shoving a screwdriver in your eye. There's no point.
Some corrections:
* Parallels/VMWare aren't emulating anything. They're using newish x86 instructions to let the system run 2 OS's simultaneously
* Ever consider recompiling? I mean, it's called open source for a reason.
Also, if you're gonna tinker, consider Solaris. It's free and Parallels supports it with nice X11 extensions for mouse sharing, etc. Also, it's BrandZ lets you run Linux binaries.
Frankly, a lot of people wouldn't feel as safe there, nor would the conference seem as legitimate. Justified or not.
I just want the ability to auto-download package dependencies.
Download the free developer edition. It's included.
Frankly, it's like any other technological solution.
Right now, it's riddled with more trouble than it's worth. It'll have lots of public failures. But, each failure will lead to an improvement, and eventually it's pretty decent.
The problem most have with electronic ballots is the threat of indetectable corruption. But how much corruption do we have in paper ballot systems? Considering the cost of the counting process, we can't really use the paper trail very often. Instead, we're stuck with believing what we're told by a bunch of typically very-partisan political appointees or politicians.
It's not news when someone hacks into a box full of paper slips -- any jackass can do it. That bothers me a lot more than a team of researchers with a ballot box at home with full source code.
Electronic voting could let us all download an anonymized list of votes, and we can verify them ourselves. Cryptography could prove very useful here -- let each citizen verify that the national tally includes their vote correctly, and implement some safeguards* to make sure that there aren't any fake votes.
Computers could be used to *secure* democracy. Probably better than any other time in human history.
* E.g. have certainty in the number of votes in each area, or randomly audit votes (meaning ask that those voters affirm that those were their selections), etc.
Remind me to avoid AT&T -- how unstable is their network that an iPhone could take it down?
Puh-leez. I've been using linux since 1995 (Slackware 95, kernel 1.0.8!), and it _never_ supported my hardware.
If Linux supports all your hardware, your computer's obsolete.
Go MacOS.
But free software is free. You don't get your money back b/c you didn't pay them any :-P
Connectix used to do this (in v3 or so) for the mac. Emulating an x86 CPU on PPC. Basically, they just provided a pass-through OpenGL driver that hit the native driver & hardware.
For native CPU & a pass-through OpenGL stack, it should be pretty close to native speed. Only concerns are:
1. Direct3D/DirectX (what's it called these days?) -- emulating that or converting it to the native graphics driver isn't trivial. Or even a direct mapping.
2. Feature differences between implementations of drivers between the mac & windows. My guess is that most of the big boys use common code in between (especially now) with build setups & wrappers for each platform. But, who knows.
It's often not about not knowing how to set up access control at the wireless router, but getting some utterly broken wireless drivers to work with them. For compatibility's sake, I've had to turn off AC more than once. Only thing that works so far is MAC-based controls, but that requires getting MAC addresses for all the machines you want on the network, and letting anyone with a promiscuous wifi card read whatever you're doing on the net. (WEP may not be really secure, but it still takes a bit of work to crack).
Of course for backup reasons that point's moot -- multiple blu-ray discs have a much lower probability of all failing than a single 2.4 TB drive.