Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?
DEAL: For $25 - Add A Second Phone Number To Your Smartphone for life! Use promo code SLASHDOT25. Also, Slashdot's Facebook page has a chat bot now. Message it for stories and more. Check out the new SourceForge HTML5 Internet speed test! ×

Comment Re:Ghostbusters (Score 1) 487

I quite liked the new Ghostbusters. It wasn't the best film ever, but it felt like an actual Ghostbusters film with the same kind of zany humour and atmosphere, unlike e.g. the Crystal Skull Indiana Jones abomination. And I say that as someone who saw the first two films as a child on the big screen.

Also, if you look at the content of the negative comments for Ghostbusters on IMDB, plenty of them do seem to be from people that somehow took offence to it.

Comment Re:Industry Shill (Score 3, Insightful) 154

It's not that surprising since the whole "drain the swamp" mantra was just something someone in Trump's campaign team proposed as a slogan, but that Trump didn't like. He then tried it out at a meeting, discovered it caught on, and kept using it. That's what Trump himself said after the elections anyway.

Comment Re:Performance? (Score 1) 85

Having an intermediate format that you statically translate into the target architecture is definitely useful (like Android is now doing with ART), but keep in mind that LLVM IR is not architecture-independent most of the time. E.g., when LLVM IR is generated from C, then this C code will at least have been compiled based on a certain pointer size, size of long, size of long long, alignments for struct fields, etc. CIL is better in this regard.

However, you should see this as a solution that will be used in an addition to running native apps, not as something you promote to do instead. It probably mainly because it seems unlikely that all, or even many, developers will start shipping tons of Windows 10 AArch64 applications right away. It's a bit like how Apple shipped Rosetta with the first Mac OS X versions for Intel Macs so they could emulate PowerPC apps (and an m68k emulator with the PowerPC versions of classic MacOS), while at the same time encouraging developers to start creating native Intel apps.

Comment Re: Performance? (Score 1) 85

Back then it was still Dynamo. And they only managed to do that on a particular HP PA-RISC architecture, because it was very sensitive to instruction cache missers (or had a bad branch predictor?) so that creating linear traces of code was very performant. They later tried to reproduce it on x86 and failed horribly (just like I did during my master's thesis; the best I got was a 20% slowdown for gzip, I think the best they got was no performance loss with some benchmarks).

Comment Re:Performance? (Score 4, Interesting) 85

The takeaway is that simply emulating the x86 instruction set results in about a 100x slowdown for an equivalent clock rate.

Emulation definitely results in slowdowns, but it's generally much less than 100x. In particular since any emulator that focuses even slightly on performance uses dynamic compilation: it translates the code once from x86 to the host architecture and from then on runs this translation. The translated version will probably be less efficient than the original code, but by no means 100x slower. 2x to 5x seems more realistic on average, although there are certainly outliers (e.g. code that intensively mucks with system registers or that triggers context switches will be slower, while some straightforward calculation loops may actually become just as fast as or even faster than the original code depending on the target architecture's nature).

Comment Re:Formal verification is worthless IRL. (Score 1) 531

I love that people can and do undertake this type of thing. It's good for the development and improvement of the state of the art, continued research is valuable purely as research and I don't really want it to stop.

It's also so far removed from pragmatic non-academic reality that it's insane.

SYSGO is a pretty pragmatic non-academic company that has existed in reality since 1991.

Three years, and found two bugs?

Keep in mind that PikeOS was already used in automotive and avionic systems, so it was already rigorously developed and tested. It is indeed quite a testament to the quality of the product.

Let alone the obvious risks around the complexity, completeness and accuracy of a model that took so long to develop.

This model was developed exactly because existing models and theories could not be applied to real-life systems such as the PikeOS microkernel. The papers I linked earlier explicitly discuss managing the complexity.

The EAL link is interesting though, hadn't come across that before.

I hope you are not working on safety-critical systems then.

I suspect most of the systems I work on would qualify at EAL4 but the cost of demonstrating it would kill their commercial value; we'd discontinue the products rather than waste the money.

In the industries where this currently matters, it is the same: you either get your certification (this, or another one depending on the field) or you discontinue the product since you are not allowed to sell it at all (or rather: your prospective customers are not allowed to use it).

EAL7? Comically that's more likely - but only for key algorithms we develop and use. A whole system? Could take down the company..

Another part of the project was about formalising the MILS (multiple independent levels of security) architecture, which enables you to compose components certified at different levels (EAL or otherwise) in a way that does not bring everything down to the level of the least certified component.

And indeed, you will almost never certify an entire system at EAL7.

Comment Re:Formal verification is worthless IRL. (Score 1) 531

constructed a formal model of the PikeOS separation kernel

How long did this take?

The EU project lasted 3 years. Before they could model the kernel, the formal modelling guys had to extend their modelling environment. There were two guys from SYSGO (developers of PikeOS) working (very) part-time on this, two senior researchers from universities, and then some graduate students (I don't think more than two).

Did it actually add any value?

They found 1.5 bugs in PikeOS with it: 1 real bug, and one theoretical bug that could not occur in practice due to the limitations of the API in which it was found (but it was fixed anyway, since you never know whether someone might want to extend it at some point).

Additionally, formal modelling is also a requirement for higher levels of certification, which the PikeOS developers want to reach.

Would it have been quicker to just rewrite PikeOS?

Which language and what development methodology would automatically have the same results as the existing working product and its associated formal model?

Comment Re:Formal verification is worthless IRL. (Score 4, Informative) 531

When you write a program that needs to print the primes up to a certain number, you can easily create a formal proof that your program program is correct.

But when your program is say "apache", that needs to interact with many different browsers on one side, and interpret PHP scripts that interact with databases, this formal proof becomes impossible. Similarly, you cannot write a formal spec for the interaction with the user in for example, a web browser.

While things like the halting problem obviously prevent fully formally proving the correctness of programs, you can go much farther than we generally go today. For example, I participated in an EU project where they constructed a formal model of the PikeOS separation kernel (kind of like an embedded real-time hypervisor). They also generalised this model, which includes support for things like interrupts and context switches.

Comment Re:I'll tell you what could go wrong... (Score 5, Insightful) 144

The general population should just shut their fucking mouths when they feel like spewing an "opinion" about something.

Then why don't you starting by setting a good example?

Science is a process and is hard.

That's why it's perfectly fine for people to voice concerns when they start experimenting in the wild. It's not like we have a perfect overview of how significantly reducing these mosquito populations will affect all other animals that feed on them (and the animals that feed on those, plants that depend on their excrement etc), just to name one potential unintended side effect. Or how it may allow other animals to largely expand their population due to reduced competition for habitats or food sources (mosquitos generally don't survive on blood, that's just what they need for procreation). Or conversely, certain nutrients no longer getting sufficiently removed from the water by mosquito larvae, resulting on too high concentrations of certain substances that then start killing other animals or plants.

TL;DR: Voice educated questions about scientific stuff. Do not broadcast uninformed opinions derived from your safe spot.

If anything is anti-science, it's trying to pre-emptively paint any debate by the general public as uninformed hipster trash. Because that is how you create luddites: by telling people they don't have a say, can't possibly understand anything about the ramifications, and should shut up and defer to some abstract scientists in ivory towers on the authority of some anonymous coward throwing a tantrum.

Comment Re:Seems mostly like a left wing echo chamber (Score 1) 106

Oh, the article obviously very biassed, no question about it. But I like the fact that the author doesn't try to pretend she's not biased or writing in a "balanced" way. It makes it all the more believable to me. But again: it could also be completely made up for all I know, although she's dropping a lot of names of people that were present and specifics that should make it easy to debunk things in that case.

Comment Re: Seems mostly like a left wing echo chamber (Score 2) 106

If the way that Milo guy is portrayed in the article is in any way accurate, he doesn't appear to espouse any viewpoints at all. He just enjoys trolling, including trolling people into believing that what he writes are actually his viewpoints.

Of course Twitter may, like Facebook, help/hurt/hide/quash certain trending topics etc, but I don't think this person's case a particularly good example of how it's ruled by some kind of elite that does not like any dissent.

Slashdot Top Deals

"They that can give up essential liberty to obtain a little temporary saftey deserve neither liberty not saftey." -- Benjamin Franklin, 1759