Worth mentioning that in ancient Rome, elections were a huge deal. Politicians hired slaves to copy pamphlets to be disseminated to the voters. It was an expensive undertaking, but you know, once you won, the money came pouring in, so it was worth it.
So......there were only elections in 18 countries last year? Is that it?
Seriously, the only reason Trump is president is because he was better at social engineering than Clinton. And I have to hand it to Obama, he pushed that "Hope and Change" meme better than anyone before him ever has. It's all been done before, though.
Might add that the job industry can become specialized and segmented ad-infinitum. Right now there are people who specialize, who dedicate their career to drawing fire, or cutting fingernails. There are people who spend their lives making timer circuits. The programmer industry is divided between front-end and back-end, and could easily segment even more.
Number of deaths per year from coal, considering deaths from mining, transporting, and respiratory and other issues related to burning / Number of Deaths attributible to a Chernobyl like incident (and again, with gen 3+ designs, a chernobyl per decade is a hysterical over estimate)
ok, that's a reasonable start. Do you have numbers?
You do realize that formal verification on software is somewhere between a lot harder to impossible, right?
There have been a lot of tools created in the last decade that make formal verification easier.
At a minimum, Intel needs to use basic security practices like code review, which in many cases they are not.
Also, you missed the ABI stability part. 20 years ago it would have been hard to support loadable modules compiled separately from the main kernel if the kernel was C++. Today, that would be much less of a problem.
Is it less of a problem? Last time i checked the output, gcc for ARM has a vastly different ABI than gcc for Intel, for example.
You already spend a byte on the null terminator. One more would give you 256*256 = 65k length strings or 32k with a high bit to indicate you should read more length bytes first.
If you're obsessing that much about bytes in small strings, you don't even need to use that much. Use a single byte for length. Then if the length is equal to 255, look for another two bytes at that point.
Even Haskell, probably the most pure language that is used in practice, has a bunch of "idioms" to make up for the fact that it doesn't have mutable state. Almost any other language would consider that a hack.
Ah, this is the best thing I've read all week, thanks.:)
Do you really think the new standard would make much difference (for his uses)? They're unlikely to import Boost into the kernel, for example. OSX uses C++ but it's a subset of C++ that is mainly 20 years old. So is there really anything in the newer standards that would affect his use cases?
But, yea, keep on beating the dead horse. Or, you know, recommend that IME be updated (with something better, perhaps?) like every security ridden nightmare that is the modern PC. Push for Intel to document it a lot better?
Intel should use formal verification. They already do it for a lot of their hardware. Also, they should at least have code review, because current evidence shows that really really dumb stuff is getting through (for example, empty password always accepted).
Slashdot Mirror
Worth mentioning that in ancient Rome, elections were a huge deal. Politicians hired slaves to copy pamphlets to be disseminated to the voters. It was an expensive undertaking, but you know, once you won, the money came pouring in, so it was worth it.
So......there were only elections in 18 countries last year? Is that it?
Seriously, the only reason Trump is president is because he was better at social engineering than Clinton. And I have to hand it to Obama, he pushed that "Hope and Change" meme better than anyone before him ever has. It's all been done before, though.
Well that definitely includes you. A smart person is just someone who says something you don't know.
Did you learn anything interesting worth reporting on?
This might be a noob question, but why do you need a smart bulb separate from a smart switch?
I'd say 'the bad' is that you never really know if every flaw is patched
No, you know the answer. The answer is No, they're not patched.
Note that if you record the audio without their permission, it is illegal and you could go to jail for it.
...Spain, where they have clearly defined legal definitions of what constitutes work place bullying.
That's cool. What definition do they use?
I didn't realize I was defending a doctoral thesis here..
But wouldn't it be nice if everyone on Slashdot had links to back up their assertions? Nicely done.
Might add that the job industry can become specialized and segmented ad-infinitum. Right now there are people who specialize, who dedicate their career to drawing fire, or cutting fingernails. There are people who spend their lives making timer circuits. The programmer industry is divided between front-end and back-end, and could easily segment even more.
Number of deaths per year from coal, considering deaths from mining, transporting, and respiratory and other issues related to burning / Number of Deaths attributible to a Chernobyl like incident (and again, with gen 3+ designs, a chernobyl per decade is a hysterical over estimate)
ok, that's a reasonable start. Do you have numbers?
It takes several years of testing to ensure that the code is correct. The cost isn't just in writing the code.
"Is bland food eaten without salt? Is there any taste in the white of an egg?" -Job 6:6
When you talk about the taste of food, it is really easy to relate to people from 4000 years ago. Biologically they were just like us.
Even if we had another chernobyl ever decade (and there is no reason why we should), but even if, it would still be a net benefit...
How can you possibly measure that? I would love to see your numbers, or even a back-of-page estimate.
You do realize that formal verification on software is somewhere between a lot harder to impossible, right?
There have been a lot of tools created in the last decade that make formal verification easier.
At a minimum, Intel needs to use basic security practices like code review, which in many cases they are not.
Cool, thanks for the informative post.
The problem was that different compilers, and even different versions of the same compiler, had different ABIs on the same architecture.
Is that not a problem anymore? I seem to remember visual studio stuff not linking to gcc stuff, although I haven't checked for a while.
Bizarre. You can overload new and delete on a per-class basis.
That's exactly what makes it unpredictable.
Also, you missed the ABI stability part. 20 years ago it would have been hard to support loadable modules compiled separately from the main kernel if the kernel was C++. Today, that would be much less of a problem.
Is it less of a problem? Last time i checked the output, gcc for ARM has a vastly different ABI than gcc for Intel, for example.
There are no flaws in C that can't be fixed by becoming a better programmer.
Nice quote.
You already spend a byte on the null terminator. One more would give you 256*256 = 65k length strings or 32k with a high bit to indicate you should read more length bytes first.
If you're obsessing that much about bytes in small strings, you don't even need to use that much. Use a single byte for length. Then if the length is equal to 255, look for another two bytes at that point.
Even Haskell, probably the most pure language that is used in practice, has a bunch of "idioms" to make up for the fact that it doesn't have mutable state. Almost any other language would consider that a hack.
Ah, this is the best thing I've read all week, thanks. :)
Where can you go to write microcode if you're not working for Intel or Atmel?
Do you really think the new standard would make much difference (for his uses)? They're unlikely to import Boost into the kernel, for example. OSX uses C++ but it's a subset of C++ that is mainly 20 years old. So is there really anything in the newer standards that would affect his use cases?
But, yea, keep on beating the dead horse. Or, you know, recommend that IME be updated (with something better, perhaps?) like every security ridden nightmare that is the modern PC. Push for Intel to document it a lot better?
Intel should use formal verification. They already do it for a lot of their hardware. Also, they should at least have code review, because current evidence shows that really really dumb stuff is getting through (for example, empty password always accepted).