Standards are good for interoperating with someone you don't know. When you are making the system others will hook into or the thing that hooks into someone's system. USB, PCI, TCP/IP(and subordinate RFC protocols) all facilitate talking to someone you've never worked with, and with a little luck it will all Just Work.
When you are working internally, or with a small group who is all in on the project, there's no need to bother unless you plan on opening up later. Standards are necessarily generic but there is optimization that can be gained from a specific solution. And after a little success, your specific solution could even become a standard.
I'm kinda dissapointed the vector engine didn't get a boost. Where I come from we like 1024-4096 bit vectors processing 128 bit extended floating point values, but Los Alamos is just odd like that.
I really think the next PPC vector engine should be 256 bit and support 64 bit floating point.
On the other hand, it may be that the world is also learning how general purpose processing is cheaper and scales better. And still, coding for vector processing is hard. (Has anyone fully utilized the PlayStation2 yet?)
There are some wonderful things you can do with computerized voting, but if all I get to do is cast the same vote for the same tired parties then I may more easily overcome laziness, but I won't affect the outcome.
If I expect that I won't affect the outcome, I become apathetic, and don't bother to vote.
I could vote for a real candidate, more interesting than the two parties, but they won't get elected because only the two parties get elected and anything else is throwing my vote away. Why bother?
Solution: Change the voting system to one that is fair for any number of candidates instead of the current one that reinforces duopoly.
Acceptance Voting or Rated Voting should be implemented as soon as possible at all levels.
See the URL in my sig http://bolson.org/voting/ (yes, this is my little holy cause)
...so I always shop for the best deal assuming I won't get around to sending in rebates.
Usually this means buying the $40 spindle of CDR instead of the $50 with a $20 rebate.
There are occasional offers worth it. Once bought a PowerBook that got me a free PCMCIA modem if I sent in the magic post card. Though, I would have bought that anyway.
If Content is the key to your MMOG experience, then perhaps the revolution will come with AI/Heuristic automatically generated Content.
There need to be more Unbounded games. I'd still be playing SimCity, with a population of billions no doubt, if I weren't restricted to that finite map. And then I'd network-portal my SimCity to yours and we'd scale up and build a distributed planet.
An unbounded game can't be exhausted of Content. Just set up the rules and the space.
I use gnuplot on MacOS X, but it didn't Just Work when I download-and-compiled it. I had to hack a couple pieces of code to disable it's broken NeXTStep drawing routines that it tries to use with MacOS X.
Once I got it working gnuplot has done everything I need. PNG output for web, eps output for embedding, ps output for ps2pdf. Don't bother with gnuplot's 'pdf' output mode. It depends on a very suckful PDF generation library (that stamps all your output with big "demo mode" watermark crap).
My dad uses IGOR and is very happy with it. Of course, his employer bought it for him.
He doesn't know what he wants. He didn't cite one specific gripe or even a trend. All he said was "new software sucks", simply because he was used to the old stuff. May as well say "change sucks". Some reputable theories of the universe say that change is the only constant.
That aside, yes, there's a personality/approach gap between those making the software and those using it. Most frustrating are the multitudes whose approach is so crippled that to their questions I deliver the pithy universal advice "try it and see".
I have to disagree. My own research (you can download the code and run the statistical simulation yourself) shows that IRV is only barely better than the current One Vote system. Any other of the alternate voting systems I tested did better.
Compares current model against several other models, one of which is being implemented in a few places, Instant Runoff Voting, and turns out to be the worst thing better than the current way!
C: supersonic M-1 tank C++: World War II battleship Java: Batman weapons M L : Alien Weapon Assembly: trench warfare Lisp (scheme): catapult launched lisp programmers perl: sawed-off shotgun with bayonet Fortran: Morning Star Pascal: play school "my first M-16" Basic: capgun visual languages : Nintendo games Postscript: drowning in paper COBOL: cursed dagger -2 ObjectiveC: alien tank csh: pipe bomb Prolog: dumbest of smart bombs Python: It's just a pile of tinker toys, whoever said it made a good weapon? But gosh darn it, it's modular and clean. perl: Pathologically Eclectic Rocket Launcher Ada: Garaunteed Military Grade Weaponry, Sir!
My dad is a computational physicist. He works with some codes that are k*100,000 lines of FORTRAN. Mostly f77 too.
Once upon a time, Cray had a really good automatically vectorizing compiler for FORTRAN and a mediocre C compiler. The Killer App stuck.
The word on the street (from my dad) is that new physics codes are being written in C/C++. So, maybe in 20 years most of the old FORTRAN will be replaced.
I've tried to get my dad to learn more C/C++, but he plans to retire before that's necessary.
Except for java, which lives nicely at java.sun.com. And sometimes I poke around perl.com but usually the included docs (perldoc, man perl*) are enough.
Otherwise, during the bulk of my time where I'm C and Objective/C++ programming, all I need is K&R (rarely anymore) and Bjarne Stroustrup's C++. That and the API documentation that came with the OS.
And really, for Java I download the.tar.gz of the API docs and the language spec and I'm pretty self contained there too.
...to a proprietary OS. There are several good ones that make this arrangement easy and cost effective. (Yes, I work on making one, assume I'm plugging it.) You can then know exactly what's in your OS, hack on it to your heart's content, and maybe even get some good support for it.
There is a good point that HTTP over TCP doesn't work nicely when there is lag between request and response. It keeps that nasty TCP connection going the whole time, tying up resources. For higher volume, and a more flexible scheme, go to an ACK-ed UDP transport.
In TCP, the "ok..." and "got it!" phases are implicit in that TCP will tell you your message got through. Lots of overhead that the protocol doesn't really need though. In my Networks class we hear the End-to-End argument, that end to end the protocol should be designed to exibit only the state and information transfer it actually needs. Using TCP is a shortcut, and lazy. Good for getting things working fast, not optimal in the long run. Just like the STL, but that's another rant.
Sounds silly to me. By the time you get out to the 3rd level of cache, on a 1GHz core, there should be enough slow down that chip to chip interconnect will be adequately fast.
Either Intel has actually put research into this and discovered that it's a good tradeoff performancewise, or they've still got marketing driven engineering and someone said "wow! over 3 MB of on chip cache!"
Any guess on the wattage? Has Intel broken 100 Watts on their upward march of hot chips?
People have mentioned mp3 playing in the all-in-one hand held. Add Ogg-Vorbis to that. AND, and recording. If the iPod had a mic/line-in jack and could encode to mp3 or ogg-vorbis on the fly, I'd buy one. How cool would 4000 minutes of record time be? (more for mono-low-bitrate)
So, of course I'd want this feature in the Uber-PDA too.
Slashdot Mirror
Standards are good for interoperating with someone you don't know. When you are making the system others will hook into or the thing that hooks into someone's system. USB, PCI, TCP/IP(and subordinate RFC protocols) all facilitate talking to someone you've never worked with, and with a little luck it will all Just Work.
When you are working internally, or with a small group who is all in on the project, there's no need to bother unless you plan on opening up later. Standards are necessarily generic but there is optimization that can be gained from a specific solution. And after a little success, your specific solution could even become a standard.
I'm kinda dissapointed the vector engine didn't get a boost. Where I come from we like 1024-4096 bit vectors processing 128 bit extended floating point values, but Los Alamos is just odd like that.
I really think the next PPC vector engine should be 256 bit and support 64 bit floating point.
On the other hand, it may be that the world is also learning how general purpose processing is cheaper and scales better. And still, coding for vector processing is hard. (Has anyone fully utilized the PlayStation2 yet?)
It's a cult lead by a charismatic architect. If you think he's famous, purhaps it's just that you're buying into his message.
/.)
(Just what I hear from someone who lived there a couple years. Sadly she's not the type to read
macosx% gcc --version
gcc (GCC) 3.1 20020420 (prerelease)
Not going to go grab the latest GCC because I wouldn't count on it to work with ObjC, MachO, and whatever else Apple may still be patching in.
There are some wonderful things you can do with computerized voting, but if all I get to do is cast the same vote for the same tired parties then I may more easily overcome laziness, but I won't affect the outcome.
If I expect that I won't affect the outcome, I become apathetic, and don't bother to vote.
I could vote for a real candidate, more interesting than the two parties, but they won't get elected because only the two parties get elected and anything else is throwing my vote away. Why bother?
Solution: Change the voting system to one that is fair for any number of candidates instead of the current one that reinforces duopoly.
Acceptance Voting or Rated Voting should be implemented as soon as possible at all levels.
See the URL in my sig
http://bolson.org/voting/
(yes, this is my little holy cause)
...so I always shop for the best deal assuming I won't get around to sending in rebates.
Usually this means buying the $40 spindle of CDR instead of the $50 with a $20 rebate.
There are occasional offers worth it. Once bought a PowerBook that got me a free PCMCIA modem if I sent in the magic post card. Though, I would have bought that anyway.
If Content is the key to your MMOG experience, then perhaps the revolution will come with AI/Heuristic automatically generated Content.
There need to be more Unbounded games. I'd still be playing SimCity, with a population of billions no doubt, if I weren't restricted to that finite map. And then I'd network-portal my SimCity to yours and we'd scale up and build a distributed planet.
An unbounded game can't be exhausted of Content. Just set up the rules and the space.
I use gnuplot on MacOS X, but it didn't Just Work when I download-and-compiled it. I had to hack a couple pieces of code to disable it's broken NeXTStep drawing routines that it tries to use with MacOS X.
Once I got it working gnuplot has done everything I need. PNG output for web, eps output for embedding, ps output for ps2pdf. Don't bother with gnuplot's 'pdf' output mode. It depends on a very suckful PDF generation library (that stamps all your output with big "demo mode" watermark crap).
My dad uses IGOR and is very happy with it. Of course, his employer bought it for him.
He doesn't know what he wants. He didn't cite one specific gripe or even a trend. All he said was "new software sucks", simply because he was used to the old stuff. May as well say "change sucks". Some reputable theories of the universe say that change is the only constant.
That aside, yes, there's a personality/approach gap between those making the software and those using it. Most frustrating are the multitudes whose approach is so crippled that to their questions I deliver the pithy universal advice "try it and see".
We haven't been back and that's just fine because we have problems at home to fix.
On the other hand, I don't believe that's entirely right thinking. It might do us good to expand on other fronts a little and do some multitasking.
I have to disagree. My own research (you can download the code and run the statistical simulation yourself) shows that IRV is only barely better than the current One Vote system. Any other of the alternate voting systems I tested did better.
See overview and some data.
I've done some simulation work on this topic. http://bolson.org/voting/
Compares current model against several other models, one of which is being implemented in a few places, Instant Runoff Voting, and turns out to be the worst thing better than the current way!
http://www-3.ibm.com/chips/news/2002/1014_powerpc. html
C: supersonic M-1 tank
C++: World War II battleship
Java: Batman weapons
M L : Alien Weapon
Assembly: trench warfare
Lisp (scheme): catapult launched lisp programmers
perl: sawed-off shotgun with bayonet
Fortran: Morning Star
Pascal: play school "my first M-16"
Basic: capgun
visual languages : Nintendo games
Postscript: drowning in paper
COBOL: cursed dagger -2
ObjectiveC: alien tank
csh: pipe bomb
Prolog: dumbest of smart bombs
Python: It's just a pile of tinker toys, whoever said it made a good weapon? But gosh darn it, it's modular and clean.
perl: Pathologically Eclectic Rocket Launcher
Ada: Garaunteed Military Grade Weaponry, Sir!
It won't read /var/mail anymore! This is for unix-clued MacOS X users out there who turned on Sendmail that's included with every MacOS X.
/Applications/Mail.app/ to save for later use.
If you haven't upgraded yet, tar up a copy of
Or at least download and compile pine. (I needed to tweak the makefiles & os-specific sources a tad, dunno if their distro patches that yet.)
My dad is a computational physicist. He works with some codes that are k*100,000 lines of FORTRAN. Mostly f77 too.
Once upon a time, Cray had a really good automatically vectorizing compiler for FORTRAN and a mediocre C compiler. The Killer App stuck.
The word on the street (from my dad) is that new physics codes are being written in C/C++. So, maybe in 20 years most of the old FORTRAN will be replaced.
I've tried to get my dad to learn more C/C++, but he plans to retire before that's necessary.
Except for java, which lives nicely at java.sun.com. And sometimes I poke around perl.com but usually the included docs (perldoc, man perl*) are enough.
.tar.gz of the API docs and the language spec and I'm pretty self contained there too.
Otherwise, during the bulk of my time where I'm C and Objective/C++ programming, all I need is K&R (rarely anymore) and Bjarne Stroustrup's C++. That and the API documentation that came with the OS.
And really, for Java I download the
Albany is the ugliest, grungiest, most depressing mat of suburban sprawl I have ever seen.
At its worst: SUNY Albany is a whole bunch of same looking grey concrete buildings.
They'd be better off in Pittsburgh.
...to a proprietary OS. There are several good ones that make this arrangement easy and cost effective. (Yes, I work on making one, assume I'm plugging it.) You can then know exactly what's in your OS, hack on it to your heart's content, and maybe even get some good support for it.
Provide a service that many of us Nerds/Geeks really need.
There is a good point that HTTP over TCP doesn't work nicely when there is lag between request and response. It keeps that nasty TCP connection going the whole time, tying up resources. For higher volume, and a more flexible scheme, go to an ACK-ed UDP transport.
Client: Hey! I want X!
Server: ok...
[time passes]
Server: (X)
Client: got it!
In TCP, the "ok..." and "got it!" phases are implicit in that TCP will tell you your message got through. Lots of overhead that the protocol doesn't really need though. In my Networks class we hear the End-to-End argument, that end to end the protocol should be designed to exibit only the state and information transfer it actually needs. Using TCP is a shortcut, and lazy. Good for getting things working fast, not optimal in the long run. Just like the STL, but that's another rant.
Sounds silly to me. By the time you get out to the 3rd level of cache, on a 1GHz core, there should be enough slow down that chip to chip interconnect will be adequately fast.
Either Intel has actually put research into this and discovered that it's a good tradeoff performancewise, or they've still got marketing driven engineering and someone said "wow! over 3 MB of on chip cache!"
Any guess on the wattage? Has Intel broken 100 Watts on their upward march of hot chips?
People have mentioned mp3 playing in the all-in-one hand held. Add Ogg-Vorbis to that. AND, and recording. If the iPod had a mic/line-in jack and could encode to mp3 or ogg-vorbis on the fly, I'd buy one. How cool would 4000 minutes of record time be? (more for mono-low-bitrate)
So, of course I'd want this feature in the Uber-PDA too.
Clustering virtual machines is about as useful as making a RAID drive accross partitions of one drive.
(Being that it sucks and is very inneficient, but lets you test the software holding it together.)
Apple quietly notes that the bigger case has a bigger battery and a rated extra hour (6 instead of 5) of run time over the smaller model.
Aside from that, I want pixels! but I suppose a few people will like seeing things a little bigger.
It'd be really cool if in 3-5 months apple rev'd this model with a new screen...