Follow Slashdot stories on Twitter


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

Comment Re:512TB? Why? (Score 1) 125

If you have a small address space then you need to write code that manually pages / caches the working set for an algorithm from storage. If you have a large address space then you use an interface similar to mmap and address the large dataset directly. It makes the code easier to write, and means that the paging / caching can be handled in hardware, where there are opportunities to speed it up.

Comment Re:Absolutely not as cool or fun, but not boring (Score 1) 449

PhDs are ten-a-penny in these parts. Most of the students that I knew back in grad-school spent half their lives on slashdot, a lot of the audience here is similar. If somebody claims to have a PhD in nuclear physics - sure why not, just like assholes, everybody's got one - but I'd still want to know what they think about the challenges in scaling stellerators up to commercial production.

Your argument about honest / open / productive doesn't really work. I can take it at face value that you personally behave that way. But this forum is not the result of a single poster, however much some people may try. It is the aggregate of many different posters so we have to talk statistically about the ouput of ACs in general. People who claim they never read AC posts probably don't do that much moderation. I seem to have points whenever I log in these days - probably the effects of c2 metamod because I do read ACs and mod up interesting posts. But they rarely turn into interesting discussions because a drive-by post by an AC doesn't generate reply notifications and if they don't care enough to register they probably don't care enough to come back and have a discussion.

Comment Re:GNU+Linux is better (Score 4, Interesting) 102

It has some features by default that really need to get added into Windows 10.

If I set Windows Defender to be off - I want it off - do not re-enable it on a timer.
If I do not want the machine to update - do not forcibly reboot it.
If I have edited the registry to avoid it rebooting after an update - do not edit it back in behind my back.

When I am running a machine in a closed environment and I only want it to change / update / reboot at desired times I don't expect Microsoft to "know better" and do it anyway. Also, it would be nice if SLI worked properly instead of over-reporting VRAM to games and causing crashes because of "memory leaks"...

Comment Re:So many things (Score 1) 181

So - some small-scale engineering issues that amount to "don't write shit code". None of which is advanced crypto.

1. Fixed IP address - don't care if it gets spoofed as long as the signature verification prevents the new code being executed.
2. Protected against buffer overflows is not advanced crypto - it is basic programming.
3. Secret keys needs to remain secret - this is not advanced crypto, this is logistics.
4. Implementing the signature check correctly is why there is a spec.
5. Side-channels - yes this does require some advanced crypto, but there are two mitigating factors:

A. Nothing leaked from the target device will compromise it if the signature scheme is public key.
B. There are no side-channel defences that work anyway - nothing currently protects against properly done DPA or fault injection.

Designing and proving the correctness of a MAC is advanced crypto. Implementing the code to run one is not.

(BTW, I was the crypto guy for many years).

Comment Re:Main reasons. (Score 1) 181

What is the advanced crypto for?

1. Check a known IP for updates.
2. If a new one is found download it.
3. Verify the signature against the public key stored in the device.

This only requires public-key crypto (from a standard library) and a basic signature scheme from a standard. Why is anything advanced required?

Comment Re: CS curriculum (Score 1) 295

It's a good article, but he does not make the argument that you say. He says that teaching C first filters out a lot of bad programmers, not makes good programmers. These are not quite the same thing.

If a school wants to make good programmers then they shouldn't teach C first as it is a horrifically bad language for learning. Joel makes the argument that it is good for testing ability. Python is actually a good first language for learning. Maybe C as a second language to learn about memory and pointers. Probably Haskell as a third language to open their minds a little.

Slashdot Top Deals

You see but you do not observe. Sir Arthur Conan Doyle, in "The Memoirs of Sherlock Holmes"