My g/f's windows machine (that she hates, but her clients require her to use proprietory software), has had about 250 reboots (usually in batches of half a dozen) in the time that my machine has had one:
475 days ago, there was a power cut, and its uptime was 350 days before then. 825 days ago, there was also a power cut, and its uptime was 305 days before that. 1130 days ago, I moved into my current flat.
Of course, that's not my only machine: phil@megaspaz:GEF$ uptime Unknown HZ value! (1154) Assume 1024.
19:15:26 up 380 days, 6:30, 12 users, load average: 1.00, 1.00, 1.00
The only thing that I've ever met that was more rock solid than my linux machines was a HPUX box from the stone age that had basically never power cycled.
"PowerPCs, by comparison, don't support it at all; they have separate instruction and data caches."
WHat kind of a comparison is that? Some x86 implementations have separate instruction and data caches. P4s even have caches of uops (pre-decoded instructions) instead of the actual instructions - i.e. the data cacha and instruction cache store completely different things. Others have instruction caches and data caches with different associativities, which means that the two behave differently. You've entirely failed to make a contrast between PPC and x86 here.
"If you modify code in a PowerPC, you have to flush the instruction cache or it won't work."
That's wrong because you've arbitrarily assumed that the code that you're modifying is already in the instruction cache. Congratulations, you've got several k of temporarily modification-resistant RAM, what about the other 500MB?
As a long time happy linux user, but also a kernel author (not x86 though, C80), I can't share your positive attitude towards linux on this issue.
Linus sloppily decided to avoid _almost all_ of the protection mechanisms that the 386 makes available to the system. That's why you can smash the stack for fun and profit. He chose to let CS access the same pages as DS (and SS,ES,FS,GS), whe he could have allocated some linear addresses as code-only, and others as data-only. After that you simply need to ensure that no CS ever was given access outside the executable range, and no other segment was given access in the executabe range. And you can ensure this - as you, the kernel, are entirely in charge of setting up user-space descriptors.
To do so would have added a bit more complexity to the memory management (with lower case letters) part of the kernel, but would have prevented all smash stacking and heap smashing attacks.
Linux is not _technically_ as good OS at all. It's simply _practically_ (for people like me) a good OS.
Tannenbaum is still right. (And when Tannenbaum says "run 20% slower" he means "take up 0.6% of the CPU rather than 0.5% of it, thus giving apps 99.4% of the cpu rather than 99.5%. But that's another rant.)
i386 _does_ have support for segment-level distinction between data and executables. i.e. an "execute bit".
In fact it's bit 43 of the segment descriptor. (Documented typically as 'bit 2' of the 'type' field, which is bits 41-43.) 1 means executable, 0 means data.
If your OS does not utilise all the features made available to it by the hardware, then don't claim the hardware doesn't support the feature.
Would moderators who up-modded the above misinformation please like to contribute constructively or humorously to this thread so that their moderation gets undone, please?
However, You're maybe forgetting Preda Mihailescu's proof of Catalan's conjecture and the AKS proof that PRIMES is in P.
(Catalan conjecture is that there are only a finite number of x,y,z,a,b,c \in |N, 1/a+1/b+1/c1 such that x^a+y^b=z^c. It's kind of a generalisation of FLT.)
Both the Catalan proof and this TPC purported proof resort to the use of analysis (integrals, the complex plane) for their proof. This makes them, to some mathematicians, much less elegant. (However, analysis is so powerful that it's used everywhere.)
If Carl Sagan said exactly that, then he was just plain wrong. The radio telescopes mop up loads of sunlight, including heat and ultraviolet, during the day. However, I doubt that he would have said exactly that, as 'electromatic' seems to be missing 'gne'.
I'm waiting until Granville, Odlyzko, Mihailescu, or someone similar gives it the thumbs up. However, it's not obvious tosh, and therefore if it does have flaws it may well be correctible, or at least provide new insight.
The guy certainly _was_ brilliant, but given that he started his peak in the mid-60s, there's no guarantee he's still at it.
I've used (e)links, but for a lynx alternative, my favourite is still w3m. The mouse working over an ssh session, and the images in a text mode browsers (in X) are a site to be seen.
They claim on http://www.fitaly.com/fitaly/ofkey.htm that they're trying to minimise finger movement, and yet in their numeric analysis they _only_ look at the individual letters' frequencies and _not_ digraph or trigraph frequency. If they had, then they would realise that the "h" should abutt both "t" and "e", which is possible by swapping "a" and "t", and "l" and "h", for example (IIRC - I'm doing this from memory).
2 minutes with a genetic algorithm and a modern English language corpus would have been all you'd need to come up with a near-optimal layout.
Wasn't there a Dr Dobbs competition about 10-12 years ago which included a tutorial on possible solutions to give people a jump start (as they actaully wanted working handwriting recognition out of the winner, not just something that was least crap). I'm sure that concepts like this were mentioned. That would be 3-5 years before the patent by the looks of things.
Ah yes, "war", you've hit the nail on the head there.
Whoever wrote that article was under the bizarre impression that IE got the largest "market share", when in fact the concept of there being a "market" is a myth. Noone chose their OS+browser bundle because of the browser -- i.e. actaully chose the browser.
Wrong. On non-hosted environments, it's perfectly valid for main to return void. Of course, 99.9% of people work with hosted environments, but those of us who code kernels for mobile phones _don't have an OS to return to_ on leaving main (and we don't ever even want to leave main), and therefore void main has been perfectly standard. Notice in the standard that they always says things like "[in|to] the hosted environment", which of course only applies to hosted environents.
Of course, pedantically, almost anything, including BASIC, FORTRAN, and Pascal, is standards-conforming C. That's because the standard's definition of conforming programs is almost entirely meaningless.
See numerous threads on comp.lang.c(.moderated) on the topic.
Slashdot Mirror
Even xterm's bloaty -
bash-2.05b$ time for i in 0 1 2 3 4 5 6 7 8 9 ; do xterm -e date; done
real 0m1.783s
user 0m0.520s
sys 0m0.230s
bash-2.05b$ time for i in 0 1 2 3 4 5 6 7 8 9 ; do rxvt -e date; done
real 0m1.370s
user 0m0.320s
sys 0m0.080s
(heavily loaded Duron/900)
I thik that 'xvt' might even be lighter weight than rxvt, but don't have it installed.
FP.
Debian Stable. Testing is OK too.
FP.
My g/f's windows machine (that she hates, but her clients require her to use proprietory software), has had about 250 reboots (usually in batches of half a dozen) in the time that my machine has had one:
phil@kilospaz:tmp$ uptime
19:19:47 up 475 days, 9:24, 7 users, load average: 1.02, 1.02, 1.00
475 days ago, there was a power cut, and its uptime was 350 days before then. 825 days ago, there was also a power cut, and its uptime was 305 days before that. 1130 days ago, I moved into my current flat.
Of course, that's not my only machine:
phil@megaspaz:GEF$ uptime
Unknown HZ value! (1154) Assume 1024.
19:15:26 up 380 days, 6:30, 12 users, load average: 1.00, 1.00, 1.00
The only thing that I've ever met that was more rock solid than my linux machines was a HPUX box from the stone age that had basically never power cycled.
FP.
$ w3m slashdot.org/
Cursor keys to navigate, no need to constantly move the hand on and off the mouse. Unless you want to use the mouse, in which case you can, of course.
FP.
"PowerPCs, by comparison, don't support it at all; they have separate instruction and data caches."
WHat kind of a comparison is that? Some x86 implementations have separate instruction and data caches. P4s even have caches of uops (pre-decoded instructions) instead of the actual instructions - i.e. the data cacha and instruction cache store completely different things. Others have instruction caches and data caches with different associativities, which means that the two behave differently. You've entirely failed to make a contrast between PPC and x86 here.
"If you modify code in a PowerPC, you have to flush the instruction cache or it won't work."
That's wrong because you've arbitrarily assumed that the code that you're modifying is already in the instruction cache. Congratulations, you've got several k of temporarily modification-resistant RAM, what about the other 500MB?
FP.
As a long time happy linux user, but also a kernel author (not x86 though, C80), I can't share your positive attitude towards linux on this issue.
Linus sloppily decided to avoid _almost all_ of the protection mechanisms that the 386 makes available to the system. That's why you can smash the stack for fun and profit. He chose to let CS access the same pages as DS (and SS,ES,FS,GS), whe he could have allocated some linear addresses as code-only, and others as data-only. After that you simply need to ensure that no CS ever was given access outside the executable range, and no other segment was given access in the executabe range.
And you can ensure this - as you, the kernel, are entirely in charge of setting up user-space descriptors.
To do so would have added a bit more complexity to the memory management (with lower case letters) part of the kernel, but would have prevented all smash stacking and heap smashing attacks.
Linux is not _technically_ as good OS at all. It's simply _practically_ (for people like me) a good OS.
Tannenbaum is still right. (And when Tannenbaum says "run 20% slower" he means "take up 0.6% of the CPU rather than 0.5% of it, thus giving apps 99.4% of the cpu rather than 99.5%. But that's another rant.)
FP.
Bollocks.
i386 _does_ have support for segment-level distinction between data and executables. i.e. an "execute bit".
In fact it's bit 43 of the segment descriptor. (Documented typically as 'bit 2' of the 'type' field, which is bits 41-43.) 1 means executable, 0 means data.
If your OS does not utilise all the features made available to it by the hardware, then don't claim the hardware doesn't support the feature.
Would moderators who up-modded the above misinformation please like to contribute constructively or humorously to this thread so that their moderation gets undone, please?
Have Americans (I assume he's American) finally done "leverage" to death?
Is "force-multiplier" the mot-du-jour?
FP.
Look up "ideals", in the context of Group Theory or Ring Theory.
The view isn't practically useful, but enables a new and useful view on many problems. It was part of the birth of modern mathematics.
FP.
Look up "admissible" "k-tuplets" on Professor Caldwell's Prime Pages at http://primepages.org/
FP.
Perhaps, if it pans out.
However, You're maybe forgetting Preda Mihailescu's proof of Catalan's conjecture and the AKS proof that PRIMES is in P.
(Catalan conjecture is that there are only a finite number of x,y,z,a,b,c \in |N, 1/a+1/b+1/c1 such that x^a+y^b=z^c. It's kind of a generalisation of FLT.)
Both the Catalan proof and this TPC purported proof resort to the use of analysis (integrals, the complex plane) for their proof. This makes them, to some mathematicians, much less elegant. (However, analysis is so powerful that it's used everywhere.)
FP.
If Carl Sagan said exactly that, then he was just plain wrong.
The radio telescopes mop up loads of sunlight, including heat and ultraviolet, during the day.
However, I doubt that he would have said exactly that, as 'electromatic' seems to be missing 'gne'.
FP.
Nope, you're missing concepts.
The definition of such densities is the limit of the density as the range being looked at tends to infinity.
Density(primes in the integers, |N) = 0
Density(non-1000-smooth numbers in |N) = 1
That's why he's looking at limits, it's because that's what one does.
FP.
"woah, settle down. no need to launch a tirade against me"
Sorry, it wasn't intended to be personal.
It was aimed entirely at those who would never read it, I guess, which is a waste.
Phil
"Hopefully this new paper will have some good cryptographic applications"
It won't. Sorry. Just like AKS, this is something that's entirely in the realm of the theoretical.
FP.
Slow down!
It's not been reviewed yet.
I'm waiting until Granville, Odlyzko, Mihailescu, or someone similar gives it the thumbs up.
However, it's not obvious tosh, and therefore if it does have flaws it may well be correctible, or at least provide new insight.
The guy certainly _was_ brilliant, but given that he started his peak in the mid-60s, there's no guarantee he's still at it.
FP.
But without MR's infantile and obnoxious behaviour this case would not have reached court.
MR is the Howerd Stern of IT.
As an old-school Debian user, I feel Lindows as an OS degrades the image of linux and of Debian, but I am still glad that MR is doing what he's doing.
FP.
"especially with current budget cutbacks in the US"
What's the US's current 'go fuck up foreign countries, and torture, rape and kill the prisoners' budget presently?
It costs far more to have a soldier on foreign soil than it does to wake up a judge.
FP.
I've used (e)links, but for a lynx alternative, my favourite is still w3m. The mouse working over an ssh session, and the images in a text mode browsers (in X) are a site to be seen.
FP.
Insert reference to Three Dead Trolls in a Baggie's "Every OS sucks".
I wish Vic20 was still alive...
FP.
They claim on http://www.fitaly.com/fitaly/ofkey.htm that they're trying to minimise finger movement, and yet in their numeric analysis they _only_ look at the individual letters' frequencies and _not_ digraph or trigraph frequency.
If they had, then they would realise that the "h" should abutt both "t" and "e", which is possible by swapping "a" and "t", and "l" and "h", for example (IIRC - I'm doing this from memory).
2 minutes with a genetic algorithm and a modern English language corpus would have been all you'd need to come up with a near-optimal layout.
FP.
Wasn't there a Dr Dobbs competition about 10-12 years ago which included a tutorial on possible solutions to give people a jump start (as they actaully wanted working handwriting recognition out of the winner, not just something that was least crap). I'm sure that concepts like this were mentioned. That would be 3-5 years before the patent by the looks of things.
FP.
Ah yes, "war", you've hit the nail on the head there.
Whoever wrote that article was under the bizarre impression that IE got the largest "market share", when in fact the concept of there being a "market" is a myth. Noone chose their OS+browser bundle because of the browser -- i.e. actaully chose the browser.
It was carpet bombing more than marketting.
FP.
Wrong. On non-hosted environments, it's perfectly valid for main to return void. Of course, 99.9% of people work with hosted environments, but those of us who code kernels for mobile phones _don't have an OS to return to_ on leaving main (and we don't ever even want to leave main), and therefore void main has been perfectly standard. Notice in the standard that they always says things like "[in|to] the hosted environment", which of course only applies to hosted environents.
Of course, pedantically, almost anything, including BASIC, FORTRAN, and Pascal, is standards-conforming C. That's because the standard's definition of conforming programs is almost entirely meaningless.
See numerous threads on comp.lang.c(.moderated) on the topic.
FP.
C has no concept of a stack.
Get off your high horse.
FP.