As long as you have C code doing the heavy lifting, you might as well enjoy the pleasant Python language for setting up the desired ray-tracing.
Yes, and that's the problem: you can't use Python for everything, you can only use it for the glue. C needs to go, and unfortunately, when you use Python, you still need C.
Python is in wide use now for math stuff like matrix operations because there are good C modules for scientific math.
NumPy is great, but, come on, stop thinking within a little box: dynamic languages like Python can be compiled into fast code, if they are properly designed. The fact that you have to us C extensions with Python is a fundamental design fault in Python, and an unnecessary one. Python is at the stage where Lisp 1.5 was 40 years ago, it's depressing.
Returning to the topic, the OLPC project is writing a desktop environment in Python, and their machine has limited memory. I was wondering if Python bytecodes are reasonably compact, and if the Python is actually helping them conserve RAM.
Python probably helps them conserve RAM. OTOH, Python's lack of an efficient compiler means that they are running much more slowly than they otherwise would. And Python's lack of a decent IDE means that programming on the OLPC is much, much harder than it needs to be.
As I was saying, Python is one of the nicest scripting languages around. But there's something wrong with the concept of a scripting language itself: we need to stop programming by typing long files into text editor and running back to C whenever we need speed.
I like Python and use it a lot. It's a really nice language. I think people will continue using it for quite some time to come.
However, Python will likely remain "the scripting language that you write add-ons in", not a mainstream language for core desktop component development. There are a bunch of reasons for that, related to packaging, error checking, development environments, object model, and raw performance. I don't see those getting fixed in P3K either.
Or, in different words, Python is one of the best scripting languages, but we should move beyond "scripting".
No, it's not. But Miguel is right on this one, and he is hardly alone in this view.
The current state of GUI and application development tools on all major platforms (Windows, OS X, Linux) is pathetic. Mono is a significant step forward, although even Mono is far from ideal.
Including a minor tool for a trivial task which takes as much memory as the rest of core Gnome together is something I can't really understand
That's bullshit. Every gnome applet or application on my machine has an RSS between 5 Mbytes and 15 Mbytes, and there are dozens of them. Tomboy has an RSS size of 26 Mbytes, which is more, but not a lot more.
But unlike all those other applets and tools, you would only need a single Mono VM to run all applets and most applications safely together. If it were fully based on Mono, you could probably run the entire gnome desktop in a small fraction of the memory it's using now, and you'd laugh at the silly suggestion of writing anything in C because it's just too inefficient.
It's the only part of Gnome proper which uses mono -- so why do they bother shipping it?
Because they realize that Gnome can't survive as a desktop that's being hacked in C/C++.
I really don't know whether the Mono VM is the future for the Gnome desktop. But I do know this much: C, C++, and Python are not the future of desktops.
A decade or more of sitting through government meetings and government training camps, being paid a lousy salary, in order to maybe make a few space flights to the stinky international space station, where you can then run a few experiments?
Frankly, you're a lot better off making enough money in the private sector to buy your own flight into space.
Sun keeps trying to have it both ways, open sourcing stuff yet trying to keep control, and that's becoming a nuisance. I think if there's a future in those packages at all, we should just get it over with and fork OpenOffice and the GPL'ed version of Java to take Sun out of the loop.
However, I'm not convinced that it's even worth it. OpenOffice and Sun Java strike me as pretty messy codebases, and it would probably be better to rewrite from scratch than to keep living with them. Although a rewrite in C/C++ would be a lot of work, a rewrite in C# or D would likely be much simpler (for Sun Java, people are already creating Harmony and IKVM, of course).
Apple Inc. CEO Steve Jobs said today that it's his company's job to stymie hackers who try to unlock the iPhone -- the first time the company has officially said it would fight attempts to use the popular device on unauthorized networks.
Good. The iPhone is sold as a non-programmable, closed, proprietary, locked platform, and I think it is good if Apple makes sure people understand that.
If you want an unlocked, programmable phone without those restrictions, buy one from the companies that sell them: Nokia, Palm, and FIC.
I'm sorry, but you seem to be having a problem understanding the meaning of the phrase "product X is overpriced". When people say "product X is overpriced", they are obviously not disputing the fact that product X might be sold at the price at which it is sold. What they are saying is that "people who buy product X at the current price either have incomplete information, or they are acting irrationally". In fact, it is often the buyers themselves that later realize that when they get more information.
It's easy to reset that yourself; check the web. They can also just take out the disk and wipe it.
That's not to say it's entirely useless: it deters quick reboots from other devices, and if someone has reset it, you'll know. But it won't protect a stolen laptop.
As soon as anybody discovers anything wrong with scientific studies, everybody from creationists and holocaust deniers to global warming opponents comes out of the woodwork and uses it as an excuse.
While it is probably true that many scientific papers are wrong or poorly supported, what remains still allows science to make progress.
If you haven't seen the documentary, I highly recommend it. One of the key issues they point out was that Gore's graph [...]
I see: so you prefer, instead of peer reviewed journals, to advance human knowledge through the medium of "published documentaries". Science may have its flaws, but it's still better than that.
Your example illustrates another flaw: it really doesn't matter what Gore's graphs show; Gore is a politician, not a scientist. He may be lying through his teeth and it doesn't affect science or the scientific consensus on the question of global warming one iota. The only facts that matter in science are those that are published in peer reviewed journals. That process isn't perfect, but it's still better at getting at truth than any other process we have.
Does anyone have a better suggestion of what to do with someone causing a public disturbance other than just not bother trying to prevent it?
Yes: the police officers should only intervene when either there is a clear and present danger, or when asked to do so by the organizers (in which case the organizers have to assume responsibility for the actions).
The guy clearly was no danger to anybody: he had no weapon and he wasn't threatening anybody. He didn't seem insane or otherwise disturbed. He was speaking at a free discussion forum. And Kerry wanted to answer his question.
A political debate is not an opera or a movie theater, where it's clear that people can and should be removed if they spoil the enjoyment of the show by others. Political debates get heated and adversarial, and police must sit on their hands. If they can't tell the difference between a corporate event and a political debate, they have no business wearing the uniform.
I think Troll Tech is seeing the writing on the wall: Gtk+ and X11 are making big inroads on mobile Linux.
Let's not fall for Troll Tech's trap: while I don't care much either way what Troll Tech does on the desktop, since it doesn't prevent me from using other toolkits, Qt on cellphones is evil because it effectively excludes use of all other toolkits. Furthermore, it forces people to use their mobile environment which (having suffered through on several devices) is not very good in my opinion.
Troll Tech claims that they developed the embedded version of Qt because X11 is supposedly too heavy-weight for cell phones. That's, of course, completely bogus: X11 runs great on cell phones and Gtk+/X11 requires less resources and is faster than Qt/Embedded. And that really shouldn't surprise anybody given that X11 was originally developed for systems with 1/1000th the amount of memory and CPU power of today's machines.
So, please stop Qt/Embedded on cell phones; you can always use Qt/X11 and have your applications live happily side-by-side with other X11-based toolkits.
I think the majority of working scientists don't "know" this and won't believe it either.
Personally, I think it's true. Between publish-or-perish, financial conflicts of interest, and various political and social movements trying to influence science, a lot of published science is worthless.
However, there's a difference between believing that and actually showing that. Anybody wanting to fix that needs clear and convincing proof first.
Can you even find anyone who stays within their TOS at all? That's so limiting as to make their service useless.
You can't have it both ways: if you're going to be picky about their use of "unlimited", they have a right to be picky about the TOS you agreed to.
It reminds me of this quote from Ayn Rand:
I'm sure Ayn Rand would completely support Comcast's right not to do business with whoever they choose for whatever reason they choose; that's a libertarian issue, you know.
In fact, the way people may hit this bandwidth limit may be through Internet TV. And, frankly, I don't want my connection to be slow because someone else spends 24/7 downloading videos. If you want that kind of service, pay for it: Comcast has higher end plans that let you do that sort of thing, and they don't cost a lot more.
It says clearly "no P2P" and "no servers". So, anybody who runs a P2P client or any kind of server can be kicked even if they use very little bandwidth.
Now, if you can find someone that gets cut-off while staying completely within their TOS, let us all know.
That's probably around one Ubuntu DVD per day, or a couple of movies per day, neither of which seems totally unreasonable.
Or, to put it differently, if you are watching a high-quality video camera that you're watching (e.g., keeping an eye on your business), you'll exceed that with just a few hours per day.
I could take the FreeBSD distro and distribute it under any license I choose,
You could; the BSD license doesn't keep you from doing that. However, since the BSD code itself is available under the BSD license, you can't keep people from using that under the BSD license, no matter what license you put on it. The only thing you can effectively put additional restrictions on is modifications and additions you make. It doesn't take much, but it takes something.
That seems to work out to a volume limit of maybe 100-200G/month, or maybe about $0.20/Gbyte at their rates. That doesn't strike me as intrinsically unreasonable.
But why not just state the volume limit and give people warnings when they get close to it?
No, I'm afraid it's not reliable for most material, because there is no assurance of available seeds.
That's not a technological issue. Technologically, USENET feeds and Torrent seeds have the same availability: they are available when someone makes them available.
Go to many of the Bittorrent repositories, and they list a huge percentage of their torrents that simply aren't available and will never be available again.
The equivalent situation has existed on USENET: articles being unavailable, sites not forwarding them, entire newsgroups being boycotted or blacklisted, etc.
Slashdot Mirror
The language is irrelevant.
Language is quite relevant: it can make a huge difference in productivity, training times, performance, and frequency of errors.
People can start doing things in Python, VisualBasic, shell much faster than in C or C++.
C and C++ are a b*tch to refactor, while for Java, Smalltalk, and C#, there are tools that do it automatically.
Almost all the buffer overrun problems in the world, and the security problems and crashes related to them, are in C and C++ programs.
And the same people who claim that "language is irrelevant" often turn around and say that they must use C or C++ "for speed".
Anybody who claims that "language is irrelevant" just doesn't know what they are talking about.
As long as you have C code doing the heavy lifting, you might as well enjoy the pleasant Python language for setting up the desired ray-tracing.
Yes, and that's the problem: you can't use Python for everything, you can only use it for the glue. C needs to go, and unfortunately, when you use Python, you still need C.
Python is in wide use now for math stuff like matrix operations because there are good C modules for scientific math.
NumPy is great, but, come on, stop thinking within a little box: dynamic languages like Python can be compiled into fast code, if they are properly designed. The fact that you have to us C extensions with Python is a fundamental design fault in Python, and an unnecessary one. Python is at the stage where Lisp 1.5 was 40 years ago, it's depressing.
Returning to the topic, the OLPC project is writing a desktop environment in Python, and their machine has limited memory. I was wondering if Python bytecodes are reasonably compact, and if the Python is actually helping them conserve RAM.
Python probably helps them conserve RAM. OTOH, Python's lack of an efficient compiler means that they are running much more slowly than they otherwise would. And Python's lack of a decent IDE means that programming on the OLPC is much, much harder than it needs to be.
As I was saying, Python is one of the nicest scripting languages around. But there's something wrong with the concept of a scripting language itself: we need to stop programming by typing long files into text editor and running back to C whenever we need speed.
I like Python and use it a lot. It's a really nice language. I think people will continue using it for quite some time to come.
However, Python will likely remain "the scripting language that you write add-ons in", not a mainstream language for core desktop component development. There are a bunch of reasons for that, related to packaging, error checking, development environments, object model, and raw performance. I don't see those getting fixed in P3K either.
Or, in different words, Python is one of the best scripting languages, but we should move beyond "scripting".
No, it's not. But Miguel is right on this one, and he is hardly alone in this view.
The current state of GUI and application development tools on all major platforms (Windows, OS X, Linux) is pathetic. Mono is a significant step forward, although even Mono is far from ideal.
Including a minor tool for a trivial task which takes as much memory as the rest of core Gnome together is something I can't really understand
That's bullshit. Every gnome applet or application on my machine has an RSS between 5 Mbytes and 15 Mbytes, and there are dozens of them. Tomboy has an RSS size of 26 Mbytes, which is more, but not a lot more.
But unlike all those other applets and tools, you would only need a single Mono VM to run all applets and most applications safely together. If it were fully based on Mono, you could probably run the entire gnome desktop in a small fraction of the memory it's using now, and you'd laugh at the silly suggestion of writing anything in C because it's just too inefficient.
It's the only part of Gnome proper which uses mono -- so why do they bother shipping it?
Because they realize that Gnome can't survive as a desktop that's being hacked in C/C++.
I really don't know whether the Mono VM is the future for the Gnome desktop. But I do know this much: C, C++, and Python are not the future of desktops.
A decade or more of sitting through government meetings and government training camps, being paid a lousy salary, in order to maybe make a few space flights to the stinky international space station, where you can then run a few experiments?
Frankly, you're a lot better off making enough money in the private sector to buy your own flight into space.
Sun keeps trying to have it both ways, open sourcing stuff yet trying to keep control, and that's becoming a nuisance. I think if there's a future in those packages at all, we should just get it over with and fork OpenOffice and the GPL'ed version of Java to take Sun out of the loop.
However, I'm not convinced that it's even worth it. OpenOffice and Sun Java strike me as pretty messy codebases, and it would probably be better to rewrite from scratch than to keep living with them. Although a rewrite in C/C++ would be a lot of work, a rewrite in C# or D would likely be much simpler (for Sun Java, people are already creating Harmony and IKVM, of course).
Apple Inc. CEO Steve Jobs said today that it's his company's job to stymie hackers who try to unlock the iPhone -- the first time the company has officially said it would fight attempts to use the popular device on unauthorized networks.
Good. The iPhone is sold as a non-programmable, closed, proprietary, locked platform, and I think it is good if Apple makes sure people understand that.
If you want an unlocked, programmable phone without those restrictions, buy one from the companies that sell them: Nokia, Palm, and FIC.
I'm sorry, but you seem to be having a problem understanding the meaning of the phrase "product X is overpriced". When people say "product X is overpriced", they are obviously not disputing the fact that product X might be sold at the price at which it is sold. What they are saying is that "people who buy product X at the current price either have incomplete information, or they are acting irrationally". In fact, it is often the buyers themselves that later realize that when they get more information.
I see the Troll Tech fanboys are out in full force again, suppressing anything that might be critical of their favorite company.
Well, yet more proof that Qt developers are realizing the technical and legal problems with their favorite platform.
Except having a light weight windowing system
There is nothing "light weight" about Qt/Embedded.
made for a framebuffer
It's not really "made for a framebuffer", it's an adaptation of a desktop toolkit.
allows Qtopia to do things like window composition effects without fancy hradware
So does X11.
and the X Composite Extension.
Indeed. And that again illustrates how Qt is designed: as a big, monolithic C++ library.
X11's protocol and extension architecture is a good thing, and it's a major deficiency that Qt isn't designed like that.
Not if it's a properly configured MacBook laptop.
It's easy to reset that yourself; check the web. They can also just take out the disk and wipe it.
That's not to say it's entirely useless: it deters quick reboots from other devices, and if someone has reset it, you'll know. But it won't protect a stolen laptop.
One down, three more to go.
As soon as anybody discovers anything wrong with scientific studies, everybody from creationists and holocaust deniers to global warming opponents comes out of the woodwork and uses it as an excuse.
While it is probably true that many scientific papers are wrong or poorly supported, what remains still allows science to make progress.
If you haven't seen the documentary, I highly recommend it. One of the key issues they point out was that Gore's graph [...]
I see: so you prefer, instead of peer reviewed journals, to advance human knowledge through the medium of "published documentaries". Science may have its flaws, but it's still better than that.
Your example illustrates another flaw: it really doesn't matter what Gore's graphs show; Gore is a politician, not a scientist. He may be lying through his teeth and it doesn't affect science or the scientific consensus on the question of global warming one iota. The only facts that matter in science are those that are published in peer reviewed journals. That process isn't perfect, but it's still better at getting at truth than any other process we have.
was there any black oil or flashes of bright light involved?
Does anyone have a better suggestion of what to do with someone causing a public disturbance other than just not bother trying to prevent it?
Yes: the police officers should only intervene when either there is a clear and present danger, or when asked to do so by the organizers (in which case the organizers have to assume responsibility for the actions).
The guy clearly was no danger to anybody: he had no weapon and he wasn't threatening anybody. He didn't seem insane or otherwise disturbed. He was speaking at a free discussion forum. And Kerry wanted to answer his question.
A political debate is not an opera or a movie theater, where it's clear that people can and should be removed if they spoil the enjoyment of the show by others. Political debates get heated and adversarial, and police must sit on their hands. If they can't tell the difference between a corporate event and a political debate, they have no business wearing the uniform.
I think Troll Tech is seeing the writing on the wall: Gtk+ and X11 are making big inroads on mobile Linux.
Let's not fall for Troll Tech's trap: while I don't care much either way what Troll Tech does on the desktop, since it doesn't prevent me from using other toolkits, Qt on cellphones is evil because it effectively excludes use of all other toolkits. Furthermore, it forces people to use their mobile environment which (having suffered through on several devices) is not very good in my opinion.
Troll Tech claims that they developed the embedded version of Qt because X11 is supposedly too heavy-weight for cell phones. That's, of course, completely bogus: X11 runs great on cell phones and Gtk+/X11 requires less resources and is faster than Qt/Embedded. And that really shouldn't surprise anybody given that X11 was originally developed for systems with 1/1000th the amount of memory and CPU power of today's machines.
So, please stop Qt/Embedded on cell phones; you can always use Qt/X11 and have your applications live happily side-by-side with other X11-based toolkits.
I think the majority of working scientists don't "know" this and won't believe it either.
Personally, I think it's true. Between publish-or-perish, financial conflicts of interest, and various political and social movements trying to influence science, a lot of published science is worthless.
However, there's a difference between believing that and actually showing that. Anybody wanting to fix that needs clear and convincing proof first.
Can you even find anyone who stays within their TOS at all? That's so limiting as to make their service useless.
You can't have it both ways: if you're going to be picky about their use of "unlimited", they have a right to be picky about the TOS you agreed to.
It reminds me of this quote from Ayn Rand:
I'm sure Ayn Rand would completely support Comcast's right not to do business with whoever they choose for whatever reason they choose; that's a libertarian issue, you know.
In fact, the way people may hit this bandwidth limit may be through Internet TV. And, frankly, I don't want my connection to be slow because someone else spends 24/7 downloading videos. If you want that kind of service, pay for it: Comcast has higher end plans that let you do that sort of thing, and they don't cost a lot more.
It says clearly "no P2P" and "no servers". So, anybody who runs a P2P client or any kind of server can be kicked even if they use very little bandwidth.
Now, if you can find someone that gets cut-off while staying completely within their TOS, let us all know.
That's probably around one Ubuntu DVD per day, or a couple of movies per day, neither of which seems totally unreasonable.
Or, to put it differently, if you are watching a high-quality video camera that you're watching (e.g., keeping an eye on your business), you'll exceed that with just a few hours per day.
I could take the FreeBSD distro and distribute it under any license I choose,
You could; the BSD license doesn't keep you from doing that. However, since the BSD code itself is available under the BSD license, you can't keep people from using that under the BSD license, no matter what license you put on it. The only thing you can effectively put additional restrictions on is modifications and additions you make. It doesn't take much, but it takes something.
That seems to work out to a volume limit of maybe 100-200G/month, or maybe about $0.20/Gbyte at their rates. That doesn't strike me as intrinsically unreasonable.
But why not just state the volume limit and give people warnings when they get close to it?
No, I'm afraid it's not reliable for most material, because there is no assurance of available seeds.
That's not a technological issue. Technologically, USENET feeds and Torrent seeds have the same availability: they are available when someone makes them available.
Go to many of the Bittorrent repositories, and they list a huge percentage of their torrents that simply aren't available and will never be available again.
The equivalent situation has existed on USENET: articles being unavailable, sites not forwarding them, entire newsgroups being boycotted or blacklisted, etc.