Read the footnotes in Easter Seal Society for Crippled Children & Adults of Louisiana, Inc. v Playboy Enterprises, 815 F.2d 323 (CA5 1987) if you have access to the final judgement documentation.
The judge references some other cases with ludicrous names. My favourite was I-forget-who v. Satan et al which was chucked out due to lack of jurisdiction.
I agree. I used to use AutoCAD LT and could work about 8x as fast with that than I could with TurboCAD. I learnt TurboCAD before I learnt AutoCAD but even the increased familiarity couldn't compensate for the lack of command-line.
Of course, you need a _good_ CLI for this to be true. vi/vim is great, AutoCAD is great, tcsh is great. Windows cmd is mostly ok most of the time. Most of the rest I've used are crap.
I would like to moderate in this thread, but I can't let your comment pass.
Most failed commercial products never saw the light of day, so only those who worked on them will know they ever existed. If an open-source project fails (whatever that means to open-source projects) it is perforce more public than when an unreleased commercial project fails.
You cannot know the success:failure ratio of closed-source projects. And we can't discuss the success:failure ratio of open-source projects unless we all know what defines a failed open-source project.
EM64T as implemented now is...crap for 64-bit work. A lot of the 64-bit maths is done with multiple (up to 4 IIRC) passes through the 32-bit ALU.
If you just need the memory and you're happy with 32-bit integers then EM64T is fine. Most people seem to want real 64-bit CPUs rather than faked ones, which is why AMD can charge 3x as much as intel for a chip that looks to have similar spec and still outsell their capacity.
Of course, in a year's time Intel might have pulled their heads out of the sand long enough to build a 64-bit ALU for their x86 line; OTOH they might still want to coerce you into buying an Itanium machine.
Mathematica already had ports to a lot of other platforms - including some LE ones. They would at most have needed to change a small amount of their OS-X-only GUI code to do the port from OSX PPC to OSX P4. Porting something which works on nine platforms to a tenth platform is usually almost trivial; porting something which works on one platform to a second platform is much more difficult.
In this case the APIs et al are identical and the only difference is the hardware, so the porting effort required is about as much as moving from HP-UX/PA to HP-UX/ia64: nil unless you have assembler code. I had to port a project which used inlined assembler to implement spinlocks from PA to ia64 -- I sized it at a week then embarassed myself by finishing in less than 2 hours (and at least 1 hour of that was spent running c89 and ld -- +Ofaster eats a lot of cycles). This is a credit to HP's engineers, but it's probable that Apple have been able to do the same thing (after all they've been working on this for 5 years).
While we're talking about compilers: the difference between gcc -O2 (then GNU ld) and xlc -qoptimize=5 (then IBM ld -qipa=level=2) tends to be huge, but Apple have stuck with GCC so far (presumably) because xlc doesnt support Objective C. I wonder if the move to Intel processors was prompted by IBM saying "we won't make xlc understand ObjC) and Intel promising an ObjC version of icc.
Generally I would say you can do pretty much anything in O(1) time given sufficient other resources.
Consider the task of optimal path finding in a graph -- you can do that in O(1) for any journey if you have precomputed the optimal path for all possible pairs of nodes in advance. Of course this costs a lot of space and a lot of preparation time.
Factoring 1024-bit RSA keys? Prepare a table mapping all possible 1024 bit numbers to their factors and use a simple array lookup - O(1).
The problem is that this approach becomes impractical very very quickly.
I had firefox 1.0.x installed with some extensions in either/both of the browers extension dir or my profile. I uninstalled 1.0.x then installed 1.0.y (y > x), following which Firefox wouldn't start.
The same problem has afflicted thunderbird migrations for me as well before. The answer, unfortunately, is to remove the product and profile extension directories and reinstall. After reinstalling I've always been able to reacquire the problematic plugins from mozdev and use them without problems.
You may find that those 3 machines have old extension directories hanging around, in which case erasing the extension directories should sort it out.
IBM Mainframes (zSeries) cannot stop using POWER CPUs because they don't use POWER CPUs. They use zSeries CPUs.
IBM Minicomputers (iSeries) and unix (pSeries) machines use POWER CPUs.
From what I've read, Apple's processor consumption is about 1-2% of IBMs annual output: losing AAPL's business would perhaps be high-profile but in revenue terms it's insignificant.
I went back from dual-screens to a single screen because dual-screens didn't suit the way I work.
At first I tried putting the monitors together and treating it as one larger desktop, but the 2x screen-surround gap in the middle meant that didn't work for me. I was hoping that this story would be about a screen with no (or at least a 95% of my time on only one screen. I think this was mainly because I could only orient the keyboard and mouse comfortably for one screen at a time. The 5% of the time I was using the second screen was mainly out of guilt for not using it.
Like most people, I would prefer more screen space. At my previous job I had an SGI 1600SW, which was ideal, unfortunately there was only one affordable graphics board which worked with its proprietary DVI system. The closest thing I can find to that now is Apple's range of cinema displays, but they're a little too expensive for me at the moment (and the pixel density still can't touch the 1600SW).
Slashdot Mirror
Because calves are made of beef.
Well yeah, I'd have modded it interesting...
No, but the taste police can arrest you.
Useful rules for all new users should be highlighted...
Read the footnotes in Easter Seal Society for Crippled Children & Adults of Louisiana, Inc. v Playboy Enterprises, 815 F.2d 323 (CA5 1987) if you have access to the final judgement documentation.
The judge references some other cases with ludicrous names. My favourite was I-forget-who v. Satan et al which was chucked out due to lack of jurisdiction.
Phil
Ever heard of Singapore?
I for one wouldn't want to be caught and punished for graffiti there.
I agree. I used to use AutoCAD LT and could work about 8x as fast with that than I could with TurboCAD. I learnt TurboCAD before I learnt AutoCAD but even the increased familiarity couldn't compensate for the lack of command-line.
Of course, you need a _good_ CLI for this to be true. vi/vim is great, AutoCAD is great, tcsh is great. Windows cmd is mostly ok most of the time. Most of the rest I've used are crap.
I would like to moderate in this thread, but I can't let your comment pass.
Most failed commercial products never saw the light of day, so only those who worked on them will know they ever existed. If an open-source project fails (whatever that means to open-source projects) it is perforce more public than when an unreleased commercial project fails.
You cannot know the success:failure ratio of closed-source projects. And we can't discuss the success:failure ratio of open-source projects unless we all know what defines a failed open-source project.
I actually laughed at that one.
Of course, under new hate speech legislation that joke may well be illegal in the UK soon.
2 goes, and you still didn't get "renowned" or "heist" right.
The unix copy command.
GP probably meant PCP, aka angel dust.
indeed.
It would on both counts.
At some point you did to[sic]
Surely everyone knows their 5-times-table?! Read 5x=20 as 5 ___s are 20 and you _know_ ___ is 4.
EM64T as implemented now is...crap for 64-bit work. A lot of the 64-bit maths is done with multiple (up to 4 IIRC) passes through the 32-bit ALU.
If you just need the memory and you're happy with 32-bit integers then EM64T is fine. Most people seem to want real 64-bit CPUs rather than faked ones, which is why AMD can charge 3x as much as intel for a chip that looks to have similar spec and still outsell their capacity.
Of course, in a year's time Intel might have pulled their heads out of the sand long enough to build a 64-bit ALU for their x86 line; OTOH they might still want to coerce you into buying an Itanium machine.
Mathematica already had ports to a lot of other platforms - including some LE ones. They would at most have needed to change a small amount of their OS-X-only GUI code to do the port from OSX PPC to OSX P4. Porting something which works on nine platforms to a tenth platform is usually almost trivial; porting something which works on one platform to a second platform is much more difficult.
In this case the APIs et al are identical and the only difference is the hardware, so the porting effort required is about as much as moving from HP-UX/PA to HP-UX/ia64: nil unless you have assembler code. I had to port a project which used inlined assembler to implement spinlocks from PA to ia64 -- I sized it at a week then embarassed myself by finishing in less than 2 hours (and at least 1 hour of that was spent running c89 and ld -- +Ofaster eats a lot of cycles). This is a credit to HP's engineers, but it's probable that Apple have been able to do the same thing (after all they've been working on this for 5 years).
While we're talking about compilers: the difference between gcc -O2 (then GNU ld) and xlc -qoptimize=5 (then IBM ld -qipa=level=2) tends to be huge, but Apple have stuck with GCC so far (presumably) because xlc doesnt support Objective C. I wonder if the move to Intel processors was prompted by IBM saying "we won't make xlc understand ObjC) and Intel promising an ObjC version of icc.
I hated those tests, as it really punishes the people who can look at the question and see the answer without doing any "working".
0.25 == 1/4. I do not now, nor have I ever needed a calculator or a method for working this out.
/combdw
:0s/combinatorial//
-or-
-or (vim only)-
V:s/combinatorial//
Generally I would say you can do pretty much anything in O(1) time given sufficient other resources.
Consider the task of optimal path finding in a graph -- you can do that in O(1) for any journey if you have precomputed the optimal path for all possible pairs of nodes in advance. Of course this costs a lot of space and a lot of preparation time.
Factoring 1024-bit RSA keys? Prepare a table mapping all possible 1024 bit numbers to their factors and use a simple array lookup - O(1).
The problem is that this approach becomes impractical very very quickly.
I've seen this too:
I had firefox 1.0.x installed with some extensions in either/both of the browers extension dir or my profile. I uninstalled 1.0.x then installed 1.0.y (y > x), following which Firefox wouldn't start.
The same problem has afflicted thunderbird migrations for me as well before. The answer, unfortunately, is to remove the product and profile extension directories and reinstall. After reinstalling I've always been able to reacquire the problematic plugins from mozdev and use them without problems.
You may find that those 3 machines have old extension directories hanging around, in which case erasing the extension directories should sort it out.
YMMV
IBM Mainframes (zSeries) cannot stop using POWER CPUs because they don't use POWER CPUs. They use zSeries CPUs.
IBM Minicomputers (iSeries) and unix (pSeries) machines use POWER CPUs.
From what I've read, Apple's processor consumption is about 1-2% of IBMs annual output: losing AAPL's business would perhaps be high-profile but in revenue terms it's insignificant.
I went back from dual-screens to a single screen because dual-screens didn't suit the way I work.
At first I tried putting the monitors together and treating it as one larger desktop, but the 2x screen-surround gap in the middle meant that didn't work for me. I was hoping that this story would be about a screen with no (or at least a 95% of my time on only one screen. I think this was mainly because I could only orient the keyboard and mouse comfortably for one screen at a time. The 5% of the time I was using the second screen was mainly out of guilt for not using it.
Like most people, I would prefer more screen space. At my previous job I had an SGI 1600SW, which was ideal, unfortunately there was only one affordable graphics board which worked with its proprietary DVI system. The closest thing I can find to that now is Apple's range of cinema displays, but they're a little too expensive for me at the moment (and the pixel density still can't touch the 1600SW).
Yes, I did know that. I've even heard that Hitler planned to make Oxford the capital of Europe.
Come the finish, of course, I'm pretty sure the anti-ness worked both ways.
Hitler's Nazi party was somewhat anti-British.
Happy now?
The paper "How to 0wn the internet in your spare time" has a good overview of rapid-spreading techniques:
http://www.icir.org/vern/papers/cdc-usenix-sec02/