Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Comment Re:What the Idiotic Hell./ (Score 1) 353

1. More popular languages you can find less expensive developers for...(or more total developer talent for a given amount of money)

Which isn't necessarily a good thing. There's a huge difference between good and bad programmers, and the more popular 'x' is, the greater opportunity for the signal to drown in the noise. Relatively fewer good applicants.

I've for instance heard that one reason Standard Chartered Bank is happy with their Haskell usage is that the quality of the applicants is so high. There's not a lot of weed to sift through. And I'm not surprised given who learns and uses Haskell.

4. More popular languages tend to be faster. Usually a shit ton faster. Java has gone from a bloated mess to a bloated mess that is often within spitting distance of C on performance shootouts. That's from the popularity spurring further development. C is almost always king of the hill and nothing is faster. Python? Rust? Whatever n00bs. Those languages may be nice to write complex code that only gets run occasionally but if you need high end performance they aren't going to cut it.

Micro benchmarks usually doesn't tell you that much about what the efficiency of a large application is going to be. Case in point, at Ericsson we routinely beat the performance of C/C++ on large scale projects with our own Erlang, even though Erlang didn't even have a compiler and was left in the dust on micro benchmarks. But as it fit the problem so much better it in practice became an issue of "doing the right thing somewhat slower" rather than "doing the wrong thing blazingly fast".

However, for the limited tasks where languages like 'C' shine, i.e. device drivers and similar low level task, i.e. limited in scope, well defined, and making the most out of specialised hardware (specialised hardware beats 'C' on a general purpose CPU every day of the week), there's no need to go to another language than 'C', as it's already there, the most mature, and well known. So, we'd use a combination of Erlang and C. It's not an either-or proposition. Trying to write the whole thing i Java, because it's "almost as fast as C" would be a complete exercise in futility...

Comment Re: I wouldn't have (Score 1) 116

96 bytes was a lot of data in the mid-80s. On a 1200 bps connection, that's almost an entire second per packet. When I was a college student in the early 90s, we had 2400 bps modems in the dialup pool, and the entire university (~3000 students) lived on a 56k leased line. Nowadays, that's trivial. In 1984, not so much.

OTOH we didn't run SLIP (PPP wasn't invented yet) over our 1200BPS/2400BPS modems, at least I don't know of anyone who did, except as a test. We ran terminal software to login to the university computers remotely. So address space wouldn't have impacted that much. In fact, where I was, TCP/IP didn't become a thing until we were connected to the internet proper. (1988), but of course YMMV.

That said, 16 bytes for an address would probably not have flown. But 6 I can sort of see.

Comment Re:Well, that's a start. (Score 1) 106

I am no authoritarian bootlicking servant. In 99% of police interactions if you comply with their instructions, you will not be injured or worse... Are the majority of police officers reasonable human beings? Yes.

99% is far too weak a standard here (and don't even get me started on "majority"). One in a thousand is also too much (due to the base rate fallacy, look it up). One in a hundred thousand is a safe conservative estimate.

So, you're about three orders of magnitude off...

Comment Re:Commercial "education" generally fails (Score 1) 317

Well, here in Sweden that has changed as well. Since the government wants more graduates in certain areas (engineering), for example, the number of seats have "exploded" in the last ten-twenty years. When I was an undergrad you had to be as near as dammit an A-student (well, 4.7-5.0 on a 1-5 scale normally distributed with a 3.0 average) to get in. Today you'll get in with a 'C' average (but you probably won't graduate!).

And if you didn't do too well in high school there are remedial classes available. If you chose the wrong "track" in high school and hence haven't studied prerequisites, then there's an extra year at university that lets you study that as a preparation.

But of course, it's in general, sort-of-kind-of a meritocracy. For the subjects with strong professional organisations, I'm looking at you in medicin, there hasn't been an increase in the number of spots, so you still need a perfect 'A' average to be admitted. (And with the increased focus on theory at the lower levels, girls now make up a majority overall, and are at 60% in medicin. Not engineering though...)

Comment Re:Why is Windows 10 the benchmark? (Score 1) 205

Yupp. Soft realtime is where it is. Even in telecoms we don't do any hard realtime if it can be avoided. It's specialised hardware (including co-processors) where that's required, but controlling it can be left to soft realtime systems. (I.e. Linux on a board).

The main thing in that case is instead reliability, but hard real time isn't required. As you say, that's for control systems, but even those tend to isolate into specific co-processors rather than rely on much in the way of an RTOS capabilities anyway. You're not going to run the weather radar software on spare cycles from your aileron control computer anyway. It's too important a task to be meddled with.

Comment Re:Totally. (Score 1) 122

In short it's a trade off between your resources and the supposed resources of the attacker. Even though SO14 escorts "heads of state" they don't escort the US first lady, etc. they bring their own apparatus.

So, since SO14/Britain couldn't do what the US does, because it would be too disruptive and cost too much given the level of threat and resources, the US secret service can. They have both a higher threat to deal with, and the resource to do so.

It's the same as in the military. If you just want to know what's going on, sneaking a four man recon patrol in, using maximum stealth is the way to go. However, if you expect the enemy to be too thick on the ground for that (problem with small patrols is that they go missing all the time, and then what?), an armoured reconnaissance battalion is the way to go. So either go very small, or sufficiently large. If you can't cordon off the area and man the whole route, then go small and stealthy, trying not to give away anything. If you can, then you go whole hog, complete with barriers and "x" number of officers per yard of road travelled.

This of course also depends on your objective. If you check the SO14 film, it's a couple of cars. The presidential motorcade OTOH isn't that far off in size from an armoured reconnaissance battalion (well, company at least), so they have to behave like one. You couldn't bring that kind of mass of vehicles and people through on a contingency basis anyway, all traffic would grind to a halt, and the smallest fender bender ahead of you would make you a static target. So, you're basically forced to go the whole nine yards anyway, and hence you have to do it properly.

Comment Re:What about Slashdot? (Score 1) 382

I guess once you get to certain levels of government the "CYA" is deleting all of your e-mails.

Of course it is. At lower levels you need to preserve the evidence that you were actually told to do something by the higher-ups. This to ensure that you have their support when the shit hits the fan, rather than being an expedient way of getting the problem to go away, by being lead out in front of the bus.

Said higher-ups OTOH are the ones that actually give said orders. It's not in their interests to leave traces of them left, right and center as possible ammunition for your peers, i.e. the ones at the same level as you, that are out to weaken you, so they can be strengthened.

Hence, to be able to attack the those above you, you have to ally yourself with their enemies. There's really no other way. No-one else has the power to have your back. (Which of course explains why Snowden is in Russia. Where else on earth could he be?) Which is why such organisations value loyalty above all else. They couldn't function if they didn't.

Comment Re:Oh dear, poor SpaceX. (Score 1) 55

I seriously doubt that. Lauching might become cheaper but thats all.

And that's basically everything that has ever happened that took something from esoteric parlour trick to technology we can't live without.

E.g. it wasn't Gutenberg that ushered in the era of the printed word, it was the steam printing press. That enabled a whole new medium where words could be printed cheaply enough to be thrown away after just one reading. Whole new ball game.

Space launch is pretty much the same. What's holding us back today is cost. With lower cost many interesting things will happen. Whether cost can ever be brought down to levels that will enable a paradigm shift, that's of course still undecided, but 50% here and 50% there, before you know it, launch could become cheap.

But if you're not willing to settle for anything less than Star Trek (reactionless drive is pretty much that), then of course Musk won't be able to hold a candle to that.

Comment Re:Single payer system would avoid this problem (Score 1) 326

Untrue. In most countries the government is in charge of health care and they have a VERY easy way to regulate price gouging such as this. In any single payer system the national health service basically sets the price they are willing to pay and that's what it costs. End of story.

Yes, since I have several laying about I just looked up how much EpiPens cost here in Sweden. They're $100 for two (you can't buy them in singles). But medication for children is free (i.e. paid through the single payer health insurance system), and if you're an adult your maximum cost is capped at $250/year, with a 100% co-pay for your first purchases, which then gradually decreases until they reach zero at $250. (Cost of living is generally higher in Sweden than the US).

This is for the exact same drug, from the exact same manufacturer. And they seem to make money here as well. But since there's only one buyer that is free to import from the country where they can get the best price (and buy generics if they feel like it), they of course have some clout.

Comment Re:Failure modes (Score 1) 326

Who is calculating the dose? The problem is that person. It can be REALLY easy to get that wrong. Even trained medical professionals get dosing wrong sometimes. It does NOT eliminate the possibility of an overdose because it does not eliminate the possibility of an error in calculating the correct dose. The assumption that the maximum safe dose will always be properly calculated is not a safe assumption.

While that it a problem with other medication, it's not much with an "epipen" style device. They only come in two doses: Child and adult. So the doses are heavily standardized. And its not considered more dangerous than "if the first shot doesn't have the desired effect within 5-15 minutes, then administer another one." And that's from a body weight of 30kg and up.

So, in this case, the therapeutic range is rather large, and dosage isn't too critical. (In fact, the only time I had to give one in anger to a child, I had to use two, since the first didn't have the proper effect. No problem.)

No, misfiring, i.e. giving too low a dose is a problem of course. We've had one of the other ones prescribed that was withdrawn since it failed to give the correct (large) dose. However, if you've now managed to make them so expensive in the US that people go without, taking the chance of not getting enough is of course preferable to going completely without. Which is the certain result if you're completely without one.

Comment Re:Why do I think (Score 3, Informative) 74

Rules are for thee, not for me. Back to work, subject.

That's a bit harsh IMHO. Impersonating a federal agent is a crime that has been on the books for a long time and with a substantial body of case law to support it. In this case, as a search was performed it's clearly illegal. (Interestingly it's not the impersonation as such that's a problem, otherwise it'd be difficult to make the X-files, but how you use it. Gaining anything of value, or performing a search are clearly out of bounds.)

Note that this rule is there (mainly) to protect the public from fraudsters, not to protect the government.

Journalists are, however, not a protected group in this sense. There are no laws on the books, and that's probably also for the best, lest every blogger in the land be hauled before the magistrate for "impersonating a journalist". So the lack of protection is arguably to the benefit of the freedom of the press.

Now, impersonating a journalist for law enforcement purposes may be ill advised, no argument there, but but clearly not illegal. It's also noteworthy that the rules have since changed to make this practice less available.

Comment Re:Universities aren't completely honest either (Score 1) 420

Yes, there's that. The grading that is. I don't think its as difficult that we often make it out to be, but since engineering studies require a lot from the students, the notion that "we have to make sure that no-one can get away with coasting or cheating" seems quite spread (in the west at least, which is the only academic culture I can speak of).

So since collaboration invites questions about who did what, and that could lead to "cheating", we're institutionally weary of it.

Comment Re:Universities aren't completely honest either (Score 2) 420

If I had a dollar for every time I heard "the nation's largest tech companies are demanding these skills" (group work, collaboration, well-rounded, etc.) I could retire and start my own school.

Well, being a university professor I've hear the same things, and you know, the companies are absolutely right. I think you're a bit too cynical. The reason they say these things ("we need engineers who can write and present and work well i groups") is that they do feel a real need, and also feel current graduates lacking in these respects.

However, that's always been the case. It's a pretty stable criticism. So then you have to figure out why. And the answer is very simple and I always point it out when I meet industry representatives who lament the current state of teaching: "We could do that. We could turn out engineers who are much, much better at writing, speaking and what have you. But if we do, we have to cut something else. And we already have trouble cramming the vital subject matter into the education as it is. We've already, during my lifetime added a whole year to the master's engineering degree (from four to five), and you still have to spend too much effort to train them until they can become really productive. So do you really want us to spend more time on writing/presentation, at the cost of less time teaching programming and whatnot?"

The answer is always, "Well, not not really..." and "Well, when you put it that way..."

So, if you ask people what they want, its always always something else than what they would actually spend money on, and pay for. Schools as a general rule don't understand this (an affliction they share with many others in the public sector), the opt for asking and delivering that, instead of realising that wishing and actually putting your money where your mouth is, are two very different things.

Comment Re:Machinists (Score 1) 420

I've worked a plant with 50 turret lathes that date from around WWII - still going strong today. No CNC machine is going to drive them out any time soon for economic reasons if nothing else because CNC is expensive.

Yes, but in a sense they're worse for employment than CNC are. With a turret lathe you'll get an even lower cost per item than with a CNC machine, if you have the steady, large scale, and stable-over-time, demand to keep them running.

The demand scale is still: If you need a couple, do them manually. If you need more, CNC. If you need even more than that, build (or adapt) a specialised solution (aka turret lathe).

I remember when I was at SKF, the production engineers could tell me to the cent how large a run of what part made the most sense to produce with what tools available. (Of course I was at the research laboratory where we had exactly one really skilled machinist, with two or three manual machines, as we only did one-offs and not enough of them to keep more than one guy busy.

Back in the day though, you didn't have CNC, so either you used manual, or a turret lathe. The majority of machinist were busy doing the production work "in the middle" that CNC now has dominated. That's where all the jobs went, and that's why CNC was the killer it was/is. Not that we still don't people of all skills. (Even though they're getting harder to find.)

Slashdot Top Deals

If all else fails, lower your standards.

Working...