Merci pour le feedback (how's my French? I'm going to Calais
tomorrow... think I might be struggling!)
Short, obviously english inspired, but clearly understandable. I'm
sure your trip went well, hasn't it?
Persistence: Not sure about memory-mapped files - these lack
transactions, queries and those other good 'enterprise' features. I
think you're closer when you mention meta-programming further down.
Alot of the issues you are bringing seems to involve difficult
tradeoffs of memory usage, safety and performance. This would explain
why they are not making way into language support. This is were
metaprogramming comes in nicely : once you have picked a library which
implements the balance you are looking for, if its syntax is the
simplest and cleaness, can ensure a more regular usage across your
programming team.
Anyway, the idea is that the process state is preserved
without the kind of explicit programming involved in a workflow system
- 'business process' and program process are congruent.
Oh, I see now. You want programmable access to coredumps. mm, tricky.
OCAML: I've heard really wonderful things about this, but people tend
to emphasise the type-safety and functional programming aspects. I'd
certainly like to check out the meta-programming-like features to see
how far it could be turned into 'enterprise programming nirvana'!
The caml's meta-programming tools and called camlp4. I also got excited
about openc++ and its sister openjava.
Very interesting comment indeed. If I claim to be a bridge accross the
academia and real-life languages, I do not know of the enterprise
reality you are talking about. However, the issues you gave as example
seems to be library issue.
Commercial information tends to be persistent, not transitory. A
good language should work directly with stored data.
Is that you wish your developers used more memory-mapped files rather
than file streams?
Processes in organizations are long-lived and distributed, whereas
typical programming languages just deal with transient threads
etc. (outside workflow systems such as WebLogic Integration).
Programs represent rules, algorithms and other forms of knowledge
that end-users will want to add to (e.g. a discount formula). Not only
should the environment allow run-time modification and extension, it
should also support representations and syntaxes accessible by
non-programmers.
You would like languages that ship with an interpreter within the
library, like lisp.
Although I clearly don't understand everything involved in the rest of
the examples you give, it seems to me they can be solved with either
lamba's ("SQL-style set predicates") or meta-language extension.
I would sugest you check out ocaml, which a
natively-compiled, c++ fast, safe web enabled language with a browser
applet plugins (if thus is your focus). Very buzzword compliant, very worthy.
I have worked on the McGill university entry to Aibo-playing Robocup from 1999 to 2001. A build the vision system, the localisation system and a control custom language for decisio making. If you weant an idea of want is involved, you can read my report here, (11 pages).
The short story is, everyteam need to code a whole bunch of difficult but fairly independent modules. The worse constraints were cpu power and noise. Oh so much noise, from all the jitering jump shaking of the walking. Loads of noise the wheeling robots (in the other categories) didn't have to deal with.
Vision - Very small pictures, 80x60 pixels. Fluorenscent-color-coded object to detect : expect they fill 10 pixels, and 4 of them is glare, 3 of them is shadow. Across the teams, little creativity was seen here : training huv->class, the more or less arbitrary function to get rid of the noise. We had something baysian and iterative.
Localisation - Noisy sighting of the corners comming in, decision which way to turn to see the target goal comming out. More variety here than in the vision, with various successes. Monte carlo was too heavy and had-oc method weren't good enough. But the best teams of 2000 could track properly.
Odometry - some team could infer their movement on the field geometricaly, by traking the angle of the joints - minus all the sliping.
Walking - walking was made up by hand more or less, with various amount of creativity, of mathematic and of success from each teams.
Decision - decision making was rule based pretty much everywhere. If you don't see the ball, find it. If you see the ball then run on it. If you are close enough, then line up with the target goal. If you are lined up, run in.
In 2000 the system began being barely good enough of attempt some cooperation : if you both see the ball and a team mate, back off so to not get in his way.
Winning team were those that could execute this the most reliably, and who walked the fastest. Fastest without falling over all that is, for there was lots of nucking and elboying between the robots, and few could had vision good enough to see each other.
The big bang was an explosing of 'space'. aka, the big bang created the space into which maters lives. It is meaningless to try to locate it within its own ejecta.
Think of the big bang as point sized ballon getting blown up. The surface of the ballon is space, the drawings on the ballon is mater, and the center of the ballon is where the big bang happend.
The functionnality of video certainly takes away from their artful claims. I remember this entry in a art exposition, some guy had blodied himself scripting a poem right off his skin. Didn't made the it throught, deemed too self-serving.
There is a well know word for what coding is about. It's a craft, not an art : purpose full beauty and skill.
In the end, art is about wider forms of human-to human-communication; anything beside the mere straitly purposeful prose we use everyday. Under that angle, I don't see many video games thus far behind used as a medium for political statements, comments on the fatefulness of the human condition.
It time to come up with a standard encoding for streams of phenomes. A big central server should be doing the english -> phenomes conversion.
It should do it offline, using the lastest mightiest, cpu-intensive, experimental text-reading AI available. Once that's done, and it only need to be done once, you can stream the phenomes down and expect the end device to render them with the user's favorite voice tone. Neato.
I guess they get hung up on the first urdle, just like you did. The tool we are looking for here is some good old normalisation.
If I can only get my hand on a good set of numbers. I would do it myself and post the results up.
If you want to know how the CPU performs on all benchmarks, just look at all of them
Well, this is more or less what I have been doing. It gets long and tedious after a while, and it makes me post on slashdot looking for a better solution, wishing for even more competent reviewing.
Now the one thing I realy can't understand is : Why the hell don't those benchmarker people publish lower-bounds/upper-bounds & stardard deviation data accross their benchmark set. I am the only one that can read them meaningfully?
Afterall, that's what the P4 saga showed us. People not only one their processor to be fast, they also want it to be predictable. It's not ok to run most software bazingly fact and run some other like a dog.
Those are generic purpose cpus for a reason. Being able to run a wide style range of code with top performance is part of their design requirement. Why does nobody's benchmarks represent this?
Re:Give me some targetted marketing
on
Slashdot Updates
·
· Score: 1
Well, on the web people pay per showing, don't they? You just wouldn't charge the bad advertisers as much.
But this is just not a unique short-term/long-term dilema in advertising.
I know local paper computer papers that swear every month to their readership they will never sell as many adds as PC-Magazine. Even though they would make more money that month, if they did, their discriminating readers would just plainly stop reading. Restraint is the right long term strategy.
I know here an art weekly that carefuly refuses adds for shows they don't belive in. Therefore the adds somehow becomes part of the editorial content. As a side effect, the advertisers knows we will very carefully read their add, and are ready to pay extra once they make it throught the filtering.
Yes you go to one of your advertiser and say : "sorry, our readership don't belive in your product (as represented in the ads). They voted againts it and you ended without much showings. Good thing, because you would probably would have wasted much more advertisement money otherwise. At least, now you leave with a free consumer survey!"
I will agree, you would need balls and a vision to set up a system like that one - but isn't what you would expect from slashdot?
Re:Here's a concept: mod the ads
on
Slashdot Updates
·
· Score: 1
The unix way : let them be small and modular. Interestigly enough, web site could be built with the same objective. Let the advertisement moderation system be modular, user-selectable, and user implementable.
Re:Give me some targetted marketing
on
Slashdot Updates
·
· Score: 2, Interesting
Actully, lets push that idea farther. Double up with google in their quest for truly useful advertisements. Yes, let's strive for plainly useful adds, rather than merly unannoying.
We have to create incensitives not to lie, myslead readers or unnecessarly annoy. Don't you just hate being technicaly-not-quite-lied to? Let registered users vote againts particular adds. Those generaly disliked will have a lower probability of being displayed.
Hey! Moderations systems for adds!
But lets not put it on the slashdot alone to come up with it. Community effort here we come :
Start a source forge projet. Publish an probabilistic advertisement interface for slash. Let everybody implement it in any various ways they can think. Then, offer a selection of the best ones from the user preference page. "Whose add manager do you want to use today? Linus', RMS' or CmdrTaco's?"
As time goes on, you drop from the list those that generate the innaceptable click-throught.
I claim this converges to very useful advertisements, both for the advertisers and the readers.
I don't see a problem with gamers pushing their system to their limits. Just disallow to provide less then x% of the available cpu time to the pool. After all everybody wants to see the snazy effect you are computing (fluid dynamics maybe?)
Cheating at games is the same kind of reverse engineering problem as breaking a copy prevention scheme. After you dealt with the blatant security holes, your best and only weapons is obfuscated code. There was an article on slashdot about this some time ago.
Esotheric programming forms and programming paradims: http://www.catseye.mb.ca/esoteric
Re:Still Needs a New Killer App
on
Webpads, Anyone?
·
· Score: 1
I see cooperative work as the future for those. Picture the scenarios : I'm working on a piece of code when I need to discuss something to my pal across in the next room.
See it happend : I just wank my lightweight screen from its dock and walk over. Its just as soon as the connection the the keyboard is servered that an on-screen keyboard appear.
The greast thing is, since we both have touch-screen stylus handy, we can work togheter right there on the same screen, without having to fight for mouse control.
It's very natural, unintrusive. Very paper-like.
Did anybody think of putting a full size keyboard on the flipside of them? With proper straps you could type two-handedly where there is no table.
I'm sorry to hear that. If that make you fell better, I go around yelling at people with such experience to switch university while there is still time. My teaching at McGill university had none of such nonsence.
The three classes I had though by the head of the cs undergrad had the best rule of all : we could work togheter all we wanted. You only had to write who you worked with on your assignement and nobody would complain. Except, if you did very in the assignement and bombed the test, and in particular if you parterner did well, he treaten to call you in for an oral test - to check out how much you realy knew. From the tone in his voice, people learned fear, and cheating was few.
I have no illusions that this is theft. But, you know what? I really don't give a fuck. I trusted my employer, and they've already screwed me over with impossible demands, tortuous contracts, and farcically worthless stock options. If they make the final betrayal, I'll loot the office without hesitation then sleep very soundly in my bed, believe me
I'm with you. This might be thief, but it is also justifyable civil disobedience.
Did you say binding verbal contract? How do they enforce them? That is my big question. Do you have first or maybe second hand account of procedures? I'm very curious. Do you mind saying where is that?
Re:what is it good for?
on
2.2 GHz Xeon
·
· Score: 1
Of course, even though they knew the right number was 30, they went for 24 to save some of that expensive film roll.
ah the joys of trade-offs! On one side my movie would cost 10'000 bucks less to make, but on the other 2000 guys will trow up watching it from the jeckiness. mmm.... decisions.
I'll roll along with you. Once your neighborg has been compromised, disabling his worm could/should be seen as self defence.
Having a trobing ecology of worm as Wishus so beautifully describes will motivate keepting admins to safe guard their machines. Because those that won't will get arbitraly shutdown by by bunch of angry grey-hat worm writers trying to protect themselves.
Re:what is it good for?
on
2.2 GHz Xeon
·
· Score: 1
Everyone knows that nobody can really see the difference between 40fps and 100+fps
Not true,worst-case frame rates over 30 are how computers do motion blur. Remember, the figure 30 came from cinematography, where they get motion blur for free. In computer graphics you need to over-sample in time to reproduce the effect.
>FOR THE LAST TIME, THE ICFP CHALLENGES ARE NOT
>TAILORED FOR DECLARATIVE LANGUAGES!
I refused to judge of the declarativeness of the challenges for fear of exposing my own bias. I would say though the orginizers are doing a might good job at comming up with cool challenges.
>Do you think language researchers sit around
>thinking, "Hmm, today I will add yet another
>obscure and utility-free aspect to my carefully
>constructed-to-be-useless language"?
>These languages are designed for universal
>utility and/or to do research into better ways to
>solve problems faced by contemporary programmers
>working with C/C++/Java/... They are not niche
>languages.
Well, I've seen alot of mislead language-design going on. I saw many preaching : "computers are getting faster so we can affort our language to be slow". It's sad to see how many languages failed because of this premise; Java is certainly surfering from it everyday.
This, along with other pervasive misconceptions, the need to create niche languages to explore particular aspect of languages design, and just the raw fact that's it's pretty hard to grow a languages to full general usefulness; With all that, alot of languages don't make it, sadly enough.
Anyway, do you have more examples of non-niche aspiring 4th generations languages? I'm somewhat familliar with the efforts of Ocaml and Haskell. Their aspiration as non-niche multi-paradims languages is fresh, difficult and brave.
Hard to say. Hard to say since any such judment of mine, if based on the values of the particular problem, would be subject to the bias of my own education.
I would attempt maybe the acm programming constest were more imperative. Of particular notice, the problems where the best pure-functional code is provably less efficient (in the big-O sence, and they exists). Those problem will tend to have massive non-local garbage collecting going on and/or degenerative lazy-hunks allocations patterns.
My consern is about seeing multipixel width lines where single pixel lines would be sufficients. I also doubt the view quality of thin anti-aliased vertical and horizontal lines that do not fall directly on grid lines.
Slashdot Mirror
Merci pour le feedback (how's my French? I'm going to Calais
tomorrow... think I might be struggling!)
Short, obviously english inspired, but clearly understandable. I'm
sure your trip went well, hasn't it?
Persistence: Not sure about memory-mapped files - these lack
transactions, queries and those other good 'enterprise' features. I
think you're closer when you mention meta-programming further down.
Alot of the issues you are bringing seems to involve difficult
tradeoffs of memory usage, safety and performance. This would explain
why they are not making way into language support. This is were
metaprogramming comes in nicely : once you have picked a library which
implements the balance you are looking for, if its syntax is the
simplest and cleaness, can ensure a more regular usage across your
programming team.
Anyway, the idea is that the process state is preserved
without the kind of explicit programming involved in a workflow system
- 'business process' and program process are congruent.
Oh, I see now. You want programmable access to coredumps. mm, tricky.
OCAML: I've heard really wonderful things about this, but people tend
to emphasise the type-safety and functional programming aspects. I'd
certainly like to check out the meta-programming-like features to see
how far it could be turned into 'enterprise programming nirvana'!
The caml's meta-programming tools and called camlp4. I also got excited
about openc++
and its sister openjava.
Commercial information tends to be persistent, not transitory. A good language should work directly with stored data.
Is that you wish your developers used more memory-mapped files rather than file streams?
Processes in organizations are long-lived and distributed, whereas typical programming languages just deal with transient threads etc. (outside workflow systems such as WebLogic Integration).
Check out channels threading primitive. Although I'm not sure what you mean by transient threads.
Programs represent rules, algorithms and other forms of knowledge that end-users will want to add to (e.g. a discount formula). Not only should the environment allow run-time modification and extension, it should also support representations and syntaxes accessible by non-programmers.
You would like languages that ship with an interpreter within the library, like lisp.
Although I clearly don't understand everything involved in the rest of the examples you give, it seems to me they can be solved with either lamba's ("SQL-style set predicates") or meta-language extension.
I would sugest you check out ocaml, which a natively-compiled, c++ fast, safe web enabled language with a browser applet plugins (if thus is your focus). Very buzzword compliant, very worthy.
Serialization is called Marshalling in caml. Check it out.
The short story is, everyteam need to code a whole bunch of difficult but fairly independent modules. The worse constraints were cpu power and noise. Oh so much noise, from all the jitering jump shaking of the walking. Loads of noise the wheeling robots (in the other categories) didn't have to deal with.
- Vision - Very small pictures, 80x60 pixels. Fluorenscent-color-coded object to detect : expect they fill 10 pixels, and 4 of them is glare, 3 of them is shadow. Across the teams, little creativity was seen here : training huv->class, the more or less arbitrary function to get rid of the noise. We had something baysian and iterative.
- Localisation - Noisy sighting of the corners comming in, decision which way to turn to see the target goal comming out. More variety here than in the vision, with various successes. Monte carlo was too heavy and had-oc method weren't good enough. But the best teams of 2000 could track properly.
- Odometry - some team could infer their movement on the field geometricaly, by traking the angle of the joints - minus all the sliping.
- Walking - walking was made up by hand more or less, with various amount of creativity, of mathematic and of success from each teams.
- Decision - decision making was rule based pretty much everywhere. If you don't see the ball, find it. If you see the ball then run on it. If you are close enough, then line up with the target goal. If you are lined up, run in.
In 2000 the system began being barely good enough of attempt some cooperation : if you both see the ball and a team mate, back off so to not get in his way.
Winning team were those that could execute this the most reliably, and who walked the fastest. Fastest without falling over all that is, for there was lots of nucking and elboying between the robots, and few could had vision good enough to see each other.The big bang was an explosing of 'space'. aka, the big bang created the space into which maters lives. It is meaningless to try to locate it within its own ejecta.
Think of the big bang as point sized ballon getting blown up. The surface of the ballon is space, the drawings on the ballon is mater, and the center of the ballon is where the big bang happend.
The functionnality of video certainly takes away from their artful claims. I remember this entry in a art exposition, some guy had blodied himself scripting a poem right off his skin. Didn't made the it throught, deemed too self-serving.
There is a well know word for what coding is about. It's a craft, not an art : purpose full beauty and skill.
In the end, art is about wider forms of human-to human-communication; anything beside the mere straitly purposeful prose we use everyday. Under that angle, I don't see many video games thus far behind used as a medium for political statements, comments on the fatefulness of the human condition.
It time to come up with a standard encoding for streams of phenomes. A big central server should be doing the english -> phenomes conversion.
It should do it offline, using the lastest mightiest, cpu-intensive, experimental text-reading AI available. Once that's done, and it only need to be done once, you can stream the phenomes down and expect the end device to render them with the user's favorite voice tone. Neato.
If I can only get my hand on a good set of numbers. I would do it myself and post the results up.
If you want to know how the CPU performs on all benchmarks, just look at all of them
Well, this is more or less what I have been doing. It gets long and tedious after a while, and it makes me post on slashdot looking for a better solution, wishing for even more competent reviewing.
Now the one thing I realy can't understand is : Why the hell don't those benchmarker people publish lower-bounds/upper-bounds & stardard deviation data accross their benchmark set. I am the only one that can read them meaningfully?
Afterall, that's what the P4 saga showed us. People not only one their processor to be fast, they also want it to be predictable. It's not ok to run most software bazingly fact and run some other like a dog.
Those are generic purpose cpus for a reason. Being able to run a wide style range of code with top performance is part of their design requirement. Why does nobody's benchmarks represent this?
Well, on the web people pay per showing, don't they? You just wouldn't charge the bad advertisers as much.
But this is just not a unique short-term/long-term dilema in advertising.
I know local paper computer papers that swear every month to their readership they will never sell as many adds as PC-Magazine. Even though they would make more money that month, if they did, their discriminating readers would just plainly stop reading. Restraint is the right long term strategy.
I know here an art weekly that carefuly refuses adds for shows they don't belive in. Therefore the adds somehow becomes part of the editorial content. As a side effect, the advertisers knows we will very carefully read their add, and are ready to pay extra once they make it throught the filtering.
Yes you go to one of your advertiser and say : "sorry, our readership don't belive in your product (as represented in the ads). They voted againts it and you ended without much showings. Good thing, because you would probably would have wasted much more advertisement money otherwise. At least, now you leave with a free consumer survey!"
I will agree, you would need balls and a vision to set up a system like that one - but isn't what you would expect from slashdot?
Dam, You beat me to the karma
The unix way : let them be small and modular. Interestigly enough, web site could be built with the same objective. Let the advertisement moderation system be modular, user-selectable, and user implementable.
Actully, lets push that idea farther. Double up with google in their quest for truly useful advertisements. Yes, let's strive for plainly useful adds, rather than merly unannoying.
We have to create incensitives not to lie, myslead readers or unnecessarly annoy. Don't you just hate being technicaly-not-quite-lied to? Let registered users vote againts particular adds. Those generaly disliked will have a lower probability of being displayed.
Hey! Moderations systems for adds!
But lets not put it on the slashdot alone to come up with it. Community effort here we come :
Start a source forge projet. Publish an probabilistic advertisement interface for slash. Let everybody implement it in any various ways they can think. Then, offer a selection of the best ones from the user preference page. "Whose add manager do you want to use today? Linus', RMS' or CmdrTaco's?"
As time goes on, you drop from the list those that generate the innaceptable click-throught.
I claim this converges to very useful advertisements, both for the advertisers and the readers.
I don't see a problem with gamers pushing their system to their limits. Just disallow to provide less then x% of the available cpu time to the pool. After all everybody wants to see the snazy effect you are computing (fluid dynamics maybe?)
Cheating at games is the same kind of reverse engineering problem as breaking a copy prevention scheme. After you dealt with the blatant security holes, your best and only weapons is obfuscated code. There was an article on slashdot about this some time ago.
Esotheric programming forms and programming paradims: http://www.catseye.mb.ca/esoteric
See it happend : I just wank my lightweight screen from its dock and walk over. Its just as soon as the connection the the keyboard is servered that an on-screen keyboard appear.
The greast thing is, since we both have touch-screen stylus handy, we can work togheter right there on the same screen, without having to fight for mouse control.
It's very natural, unintrusive. Very paper-like.
Did anybody think of putting a full size keyboard on the flipside of them? With proper straps you could type two-handedly where there is no table.
The three classes I had though by the head of the cs undergrad had the best rule of all : we could work togheter all we wanted. You only had to write who you worked with on your assignement and nobody would complain. Except, if you did very in the assignement and bombed the test, and in particular if you parterner did well, he treaten to call you in for an oral test - to check out how much you realy knew. From the tone in his voice, people learned fear, and cheating was few.
oh, I forget. emacs doesn't need new versions.
I'm with you. This might be thief, but it is also justifyable civil disobedience.
Did you say binding verbal contract? How do they enforce them? That is my big question. Do you have first or maybe second hand account of procedures? I'm very curious. Do you mind saying where is that?
ah the joys of trade-offs! On one side my movie would cost 10'000 bucks less to make, but on the other 2000 guys will trow up watching it from the jeckiness. mmm.... decisions.
Having a trobing ecology of worm as Wishus so beautifully describes will motivate keepting admins to safe guard their machines. Because those that won't will get arbitraly shutdown by by bunch of angry grey-hat worm writers trying to protect themselves.
sudo apt-get update && sudo apt-get update
Not true,worst-case frame rates over 30 are how computers do motion blur. Remember, the figure 30 came from cinematography, where they get motion blur for free. In computer graphics you need to over-sample in time to reproduce the effect.
>FOR THE LAST TIME, THE ICFP CHALLENGES ARE NOT
>TAILORED FOR DECLARATIVE LANGUAGES!
I refused to judge of the declarativeness of the challenges for fear of exposing my own bias. I would say though the orginizers are doing a might good job at comming up with cool challenges.
>Do you think language researchers sit around
>thinking, "Hmm, today I will add yet another
>obscure and utility-free aspect to my carefully
>constructed-to-be-useless language"?
>These languages are designed for universal
>utility and/or to do research into better ways to
>solve problems faced by contemporary programmers
>working with C/C++/Java/... They are not niche
>languages.
Well, I've seen alot of mislead language-design going on. I saw many preaching : "computers are getting faster so we can affort our language to be slow". It's sad to see how many languages failed because of this premise; Java is certainly surfering from it everyday.
This, along with other pervasive misconceptions, the need to create niche languages to explore particular aspect of languages design, and just the raw fact that's it's pretty hard to grow a languages to full general usefulness; With all that, alot of languages don't make it, sadly enough.
Anyway, do you have more examples of non-niche aspiring 4th generations languages? I'm somewhat familliar with the efforts of Ocaml and Haskell. Their aspiration as non-niche multi-paradims languages is fresh, difficult and brave.
Hard to say. Hard to say since any such judment of mine, if based on the values of the particular problem, would be subject to the bias of my own education.
I would attempt maybe the acm programming constest were more imperative. Of particular notice, the problems where the best pure-functional code is provably less efficient (in the big-O sence, and they exists). Those problem will tend to have massive non-local garbage collecting going on and/or degenerative lazy-hunks allocations patterns.
My consern is about seeing multipixel width lines where single pixel lines would be sufficients. I also doubt the view quality of thin anti-aliased vertical and horizontal lines that do not fall directly on grid lines.