The funnier part is that someone checking gmail would actually be safe from this sort of wifi data collection, because gmail by default uses TLS. If all sites which exchange sensitive information with the user used TLS, this particular open-wifi-scanning thing wouldn't be an issue at all.

I can relate to the grand-parent poster... the issue isn't about how much it's swapping when it can choose whether to swap or not, it's about what happens when you do run out of ram.

This happened to me when I tried to convert an svg of a microchip design to a png using inkscape (generated from a cadence export with millions of rectangles). When linux really runs out of memory and begins running things off of a gig of swap on a rotational disk, you have serious problems. It took 30 minutes for me to open a terminal and kill it.

Amusingly enough, the execution slows down so much that it can take hours for the runaway process to finally exhaust all of swap as well, and finally trigger the OOM, and until then the entire computer is useless. This is what motivates people to get rid of their swap partition, so the OOM will kill a runaway process when it happens, instead of after your computer has been frozen for a couple of hours.

I've heard that havings some swap makes the vm happier under normal circumstances, so I use a 256 MB swap partition.

I suppose I shouldn't mention that I once drove 3,100 miles from Delaware to California in 3 days, on my own... start one morning, drive through the day and night and next day, take a 10 hour nap at a motel the second night, finish it in another 12 hours. It was a "well crap, it's going to suck, let's get it over with all at once" sort of thing. Nevertheless, it happens ;)

If you're not tripping over yourself trying to make it "easy" for the "common user", then making computer systems very secure is not too difficult or expensive. Further, everyone should already be doing it anyway! It's not like governments have the only active hackers on the internet.

I would hope that official hacking threats would help change the "get the hacker" mentality back to "fix the vulnerability" or even better, "use more secure software from the outset".

Please god, no.

We have things called operating systems. They can run multiple applications reliably, securely, and fast (well, those not made by microsoft), and they can be extended by anyone with new applications without getting any permission. I could go on but you get what I mean.

Every time I have to use adobe flash or acrobat reader, I get very very angry. These are horribly made applications! With no 64 bit versions for linux systems which otherwise have had no need for 32 bit support for 5 years! Which infect windows systems with background updating services and a reliably horrific stream of security vulnerabilities, and the worst performance available! Luckily, excellent alternatives to acrobat are readily available for use with proper PDFs on all operating systems.

I don't want computers to "do everything in one go". And the people who do are already getting their asses handed to them by malware and hackers on a daily basis, while running antivirus and a plethora of independent updaters which renders their super-fast system no better than one 8 years old. Frankly, they're going to need to use Acrobat as their new operating system if they want the next decade's computers to run as slowly and unreliably as current ones.

uh... are you referring to threads? I mean, when you say the MMU could trivially enforce something... in all modern operating systems, the MMU already forces complete separation of all processes, and any interaction between them is through system calls to the kernel (or shared memory, which is set up by system calls...).

My point is, one way or the other, the OS has to decide what processes are allowed to make what system calls (with what arguments). Operating systems already have mechanisms that allow parent processes to drop some privileges for their child processes.

I would agree, however, that these mechanisms could probably be improved, expecially with regards to dropping some privileges for some threads, which might be impractical because those threads can always mess with the memory of more privileged threads...

I have to disagree about address space. OS kernels tend to use a large chunk of it; on windows xp 32 bit, only 2GiB are left for the application's address space, unless you set a boot-time flag which changes the split so applications get 3GiB. Some video games which were not "large address space aware" hit the vm ceiling after a while (on super high settings) on windows xp 32 bit (and thus crashed), as long as two years ago.

I agree that no well made program should use so much address space, but the way technology progresses, I can't agree that a 4GiB address space is enough for even the short term. When you factor in things like ASLR, bounce buffers in the linux kernel... 4GiB works, but it's getting really old really fast. Kinda like ipv4 I guess...

Perhaps you underestimate the value of not having to deal with flash?

(bitter long-time 64-bit linux user here)

I'm pretty much the same thing and do pretty much the same thing.

But I have to disagree about updates, I think you should generally apply them ASAP, otherwise it becomes too easy / likely that some service (most of which I disable, but anyway) or tool enables your box to be remotely and automatically compromised, without you manually running anything.

So, the above, but with updates, and your chances of having your system or accounts compromised is more than close enough to zero. In my opinion and experience.

So, I'm only getting to reading this some days late, but...
flightgear was updated to 2.0 in the archlinux repositories on the same day you made this comment:
http://www.archlinux.org/packages/?q=flightgear

Personally I don't mind that linux doesn't "move forward" quite the same way windows and mac does. On linux, I can get whatever I want right now, or if I don't care that much I can wait a while and get it automatically. On windows and mac, I can't really get exactly what I want either way. Well I guess I could, but hacking closed-source and non-configurable binaries is way more work than I want to get into (with props to vlite / nlite / osx86 project).

"Background services? What are those to the average user?"

What are the opinions of average users to me, as a user? They probably just think "I have a slow computer" even though its multiple times faster than $3000 computers were 6 years ago.

Maybe iTunes is pretty fast for you. But I have an ultraportable with a 1GHz core 2 duo and 1GiB of RAM, 2lbs, 8 hours battery life, good screen. And my computer is very snappy! But I had to put a custom stripped down ubuntu on it to get it that way.

So... typical users don't know and don't care... but that doesn't mean I can't complain about my use cases, which are satisfied by reasonable software, including rockbox firmware on my (old) ipod.

Your undergrad simple MIPS processor was an in-order design. Part of the point of TFA is that modern processors work totally differently - they have a register renaming unit, and tagged microcoded instructions taking multiple paths of various lengths through the pipeline, and an instruction retirement unit, oh and the branch predictor and the pipeline flush and rollback stuff... in fact real modern processors are much more complicated than even that. So there is no single physical flags register in there, but there are versioned copies of it floating around...

Anyone feel free to correct me if you have more specific knowledge of how the flags are actually handled these days.

No, the funny thing is that Slackware is the _opposite_ of what the grandparent implies... All packages are included on the CDs and users are encouraged to install the majority of them, because Slackware doesn't have a dependency-resolving package manager, and thus installing a new package isn't a single terminal command...

1) Why did / does Microsoft outsource their core competency, desktop and server software development, to contractors?

2) Why didn't Microsoft develop this in house and save time in this case by using the open source software that was used to accelerate the development of this tool, and then release the source along with it? The core functionality was already open source, and their additional work mostly only increases the value of their closed-source cash-cow Windows, anyway. It sounds like they actually haven't figured out this open-source thing yet.

Go ahead and call me an Ayn Rand Asshole, but I think employers should be able to fire employees, and conversely, employees should be able to quit.

It sounds to me like your problem is that your employer can be sued for keeping you hired for a longer term, due to some sort of "fairness" issue. You have no need for a union to get you a multi-year contract if your employer likes what you do. Engineers with no long-term-employment contract often work for one company for tens upon tens of years at a time! I have no problem with people unionizing in general, and they should be free to do it. But I wouldn't do it, because often these unions create more of these "fairness" issues that disrupt mutually beneficial arrangements between employers and employees.

PS I actually no nothing about Ayn Rand.