You missed both the point of the article and the point of the reply.
E.g. "If you intelligently apply the rule"- what rule? What are you talking about "net"? Since when is net a persistent store? Problem original article is talking about is maintaining integrity when state is spread across multiple stores. Unfortunately a) the article does not just come out and say that and instead goes into some weird rules about inputs and outputs and b) there are (and pretty common) very legit cases where violation of this rule is required. And in those cases you should deal with any edge conditions explicitly. And no "database and log file and images and whatever else crap is the single store because it's on disk" shows that you did not understand the point of the article. "single store" is wherever you have a clear contract between yourself and the final state of the store (aka transactionality or even ACID if you are lucky). BTW, most "persistent stores" other than databases make no guarantees so even if you are using only one output you still need to enforce your own failure checks.

Rest of your response shows as much understanding. Note, that tepples is *arguing against* blunt points made in the article, not making his own. Take that in context as well.

Cheers,
---

Hear, hear!
Pretty much every point of the article can be prefaced with "if you have no clue what you are doing, ".

Overall, article seems to be pretty worthless. "don't just do for the hell of it, think first"- well, erm... a) duh; b) yeah, I know anybody actually doing this is a rarity but that does not change the fact that [see 'c']; c) duh!

Not really. My assumption is that the same issues grmoc described plague JVM implementation just as well.
Everything else from this reply to the original article itself... a) I don't know who's those "everybody" that believes that on single socket async is faster than blocking. IMO it's a pretty dumb idea and blocking is as fast as it gets for throughput. People go async not because they want better throughput but because they want to be able to use the cycles on the same thread.
b) rest is about "what's implemented worse, thread switching or selectors" and that's heavily OS dependent and changes all the time but it's also pretty well known that selectors are crippled with any significant number of sockets. epol works much better AFAIK but on the question of epol vs. thread number- I don't know.

Ultimately, I believe that neither original post or this reply make any sense and you take a number of factors into account (how you work with data, can you even do multiple threads efficiently, etc.) and not optimize early.

PS on async, just don't use selectors, period.

Sounds very odd (I'm not confident enough in my thoughts to say anything stronger, feel free to explain where I'm wrong).
Java uses system threads, i.e. it has no internal scheduler. NPTL or others are just an interface into the threading/scheduler so are fairly thin. Thus trends for performance of IO and multiple threads should be consistent between Java & C++. Unless people that implemented threading/IO for JVM implemented it a little differently (better?) that people that implemented your benchmark ;)

hear, hear.

Moot point. Yes they should provide you with hardware. In fact if they do not (and you have an email to say that they require a full time employee to use their own equipment) they might be in for some legal trouble (IANAL). They can make you a contractor that has to have tools of trade. They can do many things. Ultimately though, it's far from unreasonable to attempt to secure their network and their data. How they do this- stateless Live CD for reading emails and not saving them (and not bringing every virus on the planet into the protected network), separate partition for accessing data or requirement to encrypt HD (which is not that strong of a measure in itself since it protects only one vector), they SHOULD do this and it's GOOD (despite the inconvenience it is on poor ol' you).

Yup. Rather than you loosing a laptop with unencrypted emails with confidential patient information (oh, and that would NEVER happen and does not happen with such frequency that "oh, another million records on a lost laptop, what else is new" became the reaction to the news like that).

I just hope just encrypting disks is not enough to allow you access, since viruses and spyware don't really care about disk encryption, they will happily let you enter password for them. And then send that password along with every other byte that enters or leaves your machine to your friendly virtual neighborhood ahole that will not abuse them, he is just curious.

Indeed. I was trying to find comments about this and was surprised that yours seem to be the only one. Both original-original article and the rebuff discuss FOSS and copyright questions and barely touch on net neutrality, which is supposedly the primary topic. Weird.

Err... Well... Aghm... Yes, it's a restriction on the property holder. However I would argue that it's far from unreasonable for two reasons:
1) Broadband access is highly monopolized. There is usually at most two half-decent providers in an area (one in my area, some might have decent cable, DSL and optical but it would be a huge exception). Where there is a monopoly there is no market and there has to be oversight. There aren't that many entities that hold a power to oversee and government is one of these entities.
2) Internet has become a pretty vital information source in todays society. And if we like internet providers to phone companies, net neutrality provisions would be like preventing phone companies from reducing call quality for outside long distance providers forcing to use theirs long distance. Or blocking any other company that provides similar services (conf. calls, remote voice mail, etc.).

Ultimately, if I subscribe to "10Mbit internet" as a service I should be able to use it however I want, be that making skype calls or downloading videos. They can put use caps, limit throughput or do many other interesting things to make sure they stay afloat but IMO discriminating one service over the other as they please violates the concept of what internet is.

Not really. Only 2% of sick will go in for treatment and 95% of those will be sent home withing minutes with recommendation to take some over the counter drug (not covered by insurance).

I'm not defending original post, which is a) large ad-homonym attack (maybe well-deserved but still...) and b) does not contradict research findings. Would they happily fund a study that could save them some $$? Likely. Does this mean findings are incorrect? Not really, unless you suggest fraud, which is a big step up from just being a cheapskate.

As for article... HUGE red flag goes off when an article about a research that goes against "status quo" and could be a groundbreaking finding appears in "The Atlantic Online" and not in Nature or a respected medical journal. In science one of the biggest safeguards against bull**it is peer review and publishing a research in a blog IMO speaks volumes.

You'd actually be surprise how much they have in common, especially as far as common faults.
* Assumption that an approach provides absolute security- is common in both realms and is as flawed is either.
* Plugging 'obvious' holes without comprehensive analysis- if it's easier to break the wall than open the door, somebody will break the wall
* Assuming that expertise is over-hyped and that anybody can just (install super-duper IDS | buy security system for 899.99 and install it) without understanding of attack vectors and stuff.
etc. The two definitely have as many dissimilarities but in the context of the original point ("security does not exist") IMO it was appropriate to illustrate that security is not about prevention but about delay and resources attacker would have to spend and it's also true for either realm.

Mmmm... Your statement is in parts wrong but more interestingly I don't get what you are trying to say. Could you clarify what part of my post you are refuting? I never said that physical and virtual security are the same thing (although some similar principles do apply). Are you saying that being under attack close to 100% translates into guarantee of being compromised? If that's the case, you are wrong. 15 minutes of thinking about your security (whatever you are protecting) will alleviate 99.9% of those attacks. From that point on it gets to be a battle of how much time/$$ you are willing to spend to protect vs. how much time/$$ attacker is willing to spend to breach. At some point one of you looses and I'd argue that for majority of services at the point cost of defending is much lower than cost of breaching- sad part is, very few invest enough time/thought (it's almost never about money) into security.

Your statement, that's a myth, one of many. Sure, there is no ABSOLUTE security, but nobody claims that. There is no absolute physical security either- with enough resources anything can be stolen and anybody can be killed. It's the understanding of how secure you are in any given situation and how to improve your chances of staying safe (in virtual or real worlds) is what defines security and surely, that exists.

The issue at hand is if Kindle 2 indeed has decent text-2-speech, it will be a market changer. If an author expects to get $X from a book and a publisher expects to get $Y from it, they lay out fees according to planned sales (different for each media). eBooks are chap to produce, inconvenient to read and a 'niche' market, thus licensing for those is cheap. Audio books get more and more popular (with mp3 players and services like Audible), have larger production costs (although pretty manageable) and expected income from those is much higher. Thus point he *should* have made is that if ebook eats into audio book sales, there should either be a separate royalty skew for "ebook with right to vocalize" or ebooks altogether should be skewed differently.
Bottom line, there is clearly no violation today and if authors believe they are not adequately compensated, they should change licensing terms for future titles.

OK, I may be a clueless bastard that believes fairy tales and does not comprehend the dangers and law of unintended stuffs and so on, but...
1) Storage costs for ISPs. GIVE ME A BREAK! Lets say a record is 100 bytes per event (and it won't be more than that since they are only asking about DHCP records), lets say an ISP has a million users and they all get new IPs twice a day (lets pretend no cable modem is ON 24/7 for months and same IP is not reassigned to the same MAC at reconnects). Such ISP would end up with horrid 73G file a year. It would likely compress to 10% of that size with zip but lets say we keep it uncompressed. Puhlease, ISP that can't handle THAT doesn't deserve to live.
2) Public AP. Same math- different budget or providers but I can guarantee that simple firmware update for the router and retention policy "download once a week, burn on a 30c CD and put it into a safe" or something like that can be easily implemented.
3) Home users... Plain stupidity, don't even want to go there but router developers should start taking better care about securing this stuff by default, like forcing first password change and defaulting to WPA2.

There were few other points floated: text files are editable, MACs are spoofable, etc. All valid, but nobody said that logs alone are supposed to carry burden of proof. Making part of evidence chain might be enough.

Finally, privacy concerns. First, bull about "ISP would have to store mount of PII" is just that, bull, unless bill says "and will not issue such IP before driver's license or alternate form of ID is provided". Just MACIP mapping. If whoever asks for these logs can find machine with that MAC- they have something. ISPs can also (and IMO should) include customer ID number or something. Now, there is (IMO) a very legit worry about censorship and about authorities using this to inquire "how went to this Arabic newspaper site" or "who googled 'how to make a nuke from household materials'". I think this is a HUGE concern and we might want to concentrate on that and how it will play with Patriot act and such. But rest is noise (again, IMO, please correct me)

Please don't confuse person asking the question with the question itself. Even though question was asked by Dell it wasn't phrased around PC hardware supply, it was centered around science and technology sector and this is a very interesting question with several sides:
1) Is Russia technologically or scientifically weak today? Hell no, I don't like Putin for a second but he is right, Russia does have very strong technological sector.
2) Will science survive long in Russia? Not without significant foreign interest. Science is a government venture there and government fails to finance it. Technology boom is there only because of cheap quality work force. With rising salaries in CS sector and economy crisis everywhere else technology will likely start suffering soon. So, unless there is a door opened for foreigners to invest in science and unless there is a new wave of outsourcing/investing in technology both sectors will suffer dearly and soon.
3) Does this mean that Russia needs help with this? Again, as much as I hate it I agree with mr. Putin. Russia is not in a position to beg (at least not yet). Russia would benefit from foreign investors and needs them but so would those investors. Russia needs entrepreneurs, not elves.