If you want to technically guarantee anonymity to prevent censorship and violations of the right to free speech by technical means, there is no way to distinguish legitimate users from assholes. The software cannot do this, by definition.
Well, you're half right, but you're coming to a bogus conclusion.
Yes, I agree that it's probably infeasible for a software package alone to currently do a good job of determining whether someone is an "asshole". However, we have trust networks and other mechanisms, where software acts as a powerful tool to assist human rating. Note that Freenet is inherently something of a trust network, as its native interface is a Web browser with Web pages. You presumably link to people that are not "assholes", who link to people that *they* feel are not "assholes".
I've even seen posts saying (to paraphrase) 'everyone should have free speech except kiddie pornographers and nazis'.
Sounds pretty European to me. A lot of European countries *really* still have a lot of social crap left over from after World War II. Garmany and France, in particular, are incredibly uptight (at least from an American standpoint) about Nazi-related stuff. The UK has some kind of pedophiliaphobia. I mean, sure, nobody likes the worst-case sort of sexual content related to kids -- kids getting abducted and raped or similar. However, the British are absolutely rabid about avoiding any kind of surfacing of pedophilia. Really unusual.
Arguing that a lot of Europeans should loosen up about Nazism is probably a fun debate, but it doesn't have as much impact in the US, so I'll argue the child pornography standpoint.
Frankly, I never really saw how banning child pornography has significant social benefits (especially since posession and distribution of blood sport content *is* legal). Sure, there can be all kinds of bad things associated with child pornography -- people worry about their kids getting abducted and raped or something -- but I don't see quite how eliminating distribution of such material on Freenet does anything to avoid real life sexual abuse.
The first argument I see in favor of censorship of child pornography might be that if there is a profitable industry for producing child pornography, then children will be involved in production of that pornography (as opposed to actors/actresses appearing to be children, or CG doctoring, or pornographic animation containing depictions of underage sex). This may lead to one of two potentially bad things: first, children may be nude in videos. In the Victorian tradition, being seen nude (particularly females) somehow "degrades" or damages future social standing. I've nver seen this as an immutable -- our society happens to have a nudity taboo, but it is arguable as to whether that is at all beneficial to society. The second bad point is that children may be physically injured in the production of such content. There are laws already dealing with injuring children -- I'm not sure why a special point needs to be made for pornography-related content.
The second issue might be that distribution of child ponography tends to exacerbate pedophiliac behavior. This is certainly a decent thought, but I'm not sure how grounded in reason it is. We in the United States allow distributing movies showing images of people being shot (actually, I just posted regarding this earlier today). It doesn't seem that the violence present alone (see my complaints about cartoon violence, which are different) significantly have had an impact on the increase of violence. Why would we think that sexual content alone would drive someone to engage in a sexual act?
Anti-child pornography laws are one of the few near-global laws, and I'm a bit curious why, as they seem to be the product of fear and emotion rather than a particularly reasoned decision -- at the very least, they are inconsistent with decisions about censorship that we have made in other content areas.
I'll let others wonder why the US can get away without censoring Nazi content and yet doesn't have massive Nazi surges, yet France feels the need to prevent people from having Nazi content.
Frankly, I wish that both Phipps and ESR would stop acting like reactinary idiots. It does absolutely nothing but hurt open source. ESR blasted off a ridiculously inaccurate and amaturish letter to Sun, and decided to make it an "open" letter to piss off the maximal number of people. He managed to trade in some Open Source credibility to advance his whim-of-the-day. Phipps, instead of doing the right thing and either ignoring ESR or sending him a form letter, actually responded. Had he just ignored ESR, the whole thing would have died. Now it's alive *again*. Christ. This whole argument is so incredibily stupid.
ESR is a voice in the Open Source movement BECAUSE HE IS ONE. A lot of people interview him for things. A lot of people seek his opinions. He gets invited to speak and consult. People listen to him. He is the embodiment of the movement's very principles! If you don't like how he does "spokesperson", I suggest you go try and replace him. Put out a better implementation - yourself, or whoever.
He just did. He threw out some verbal criticism, which is what's at issue, not software.
Don't respond to stpuid letters when you dnn't need to do so. The Open Source community will not hold you responsible.
ESR has some name recognition, but you are under absolutely no obligation to respond.
If you wanted to do so, a nice "thanks, we'll take your thoughts into consideration" note would work nicely at tamping the thing down. ESR clearly didn't understand the issues involved, and made quite an ass out of himself with his letter. It would have been quite easy to simply ignore the letter or send back a form letter and let the whole thing die down and be forgotten. Now you've managed to insult him and start a potential flamewar. For Chrissake, use some common sense.
The idea of (even parental) censorship is not without value, but I think that it's overvalued.
The rationales for parental censorship that I can see go something like this (w/ my responses):
Issue: If I expose my child to this scary content, he/she is not old enough to have mental constructs or required knowledge in place to prevent him/her from being overwhelmed with irrational fear.
Response: I don't think I can agree. The mental constructs issue is, I think, not a convincing argument. The way people seem to develop contructs to deal with frightening-but-not-dangerous content is by being exposed to exactly that content. You can wait until they're twenty or start when they're six. I think that, ultimately, kids are going to have to have a few frightening-but-not-dangerous-experiences to learn how to deal with fright. My own parents put strict limits on what frightening movies I could watch as a child. The Thing was the first R-rated movie I saw. When I finally saw it, I was quite frightened. My friends, who had been watching frightening movies earlier than I was, were decidedly unintimidated. Today, I shrug off The Thing and similar movies. It took exposure to a good amount of frightening content, though, to build up that state of mond. I don't think that age is a factor so much as experience.
Knowledge base is a bit more convincing. I really think that a large factor is knowledge directly related to frightening content. In this case, parents can pretty easily talk about it. For example, we really don't have any reason to think that Jason from Friday the 13th could exist. It's just as reasonable to think that there's a horde of guardian angels running around overhead, a decidedly less frightening prospect. Furthermore, Jason is some guy in a (not all that great, frequently) costume. If a child is frightened by a movie, a parent can do a bit of explaining to avoid most of it.
I think that there *is* something to the argument that there is non-movie-specific knowledge that must be had to help deal with fear, however. Generally, if a movie represents something that isn't a possible danger, there's no reason to be frightened of it. A lot of things can be ruled out as possible dangers with a knowledge base -- i.e. it's pretty unlikely that an A.I. is going to fall in love with one of its creators and kill off the creator's spouse, because such behavior is not exactly easy to impart, is probably not easily accidently evolved in the kind of environment an A.I. exists in, and is pretty complex. There's not a lot of way to ensure that a child has a wide enough knowledge base to produce explanations and assign those explanations a high enough "convincing factor" (for lack of a better term). Furthermore, risk aversion is a common mental irrationality that humans are prone to that children seem to be even more vulnerable to. If shown something that is a real, potential danger, such as dying from a rattlesnake bite in California, people do not deal well with grappling with the very small degrees of probability involved, and will assign too much danger to that bite, even though car crashes pose a far greater risk to them. That's something that I think it takes a significant amount of experience to overcome. It is *possible* that a small child simply has not taken enough life risks and reasoned about them to be able to assign slight dangers a proper amount of risk.
Issue: Marketers design their campigns specifically to appeal to humans. I don't want them to be exposed to tobacco/alcohol, and let those marketers to get their hooks into my child.
Response: I really feel that a better way to deal with marketing of potentially harmful products to a child is to attempt to innoculate him to that product, rather than isolate him from it. You simply are not going to be able to shield a child forever from a product, and I think that it would be better to simply ensure that they can make an intelligent decision early on. Explaining to a child why cigarettes are b
IBM is a huge company. The people making hard drives probably have very little to do with the people making Linux other than the same style of standardized HR forms. IBM in particular seems to operate divisions with a good deal of modularity, from what I've heard from people working there.
For Chrissake, man, IBM knows relationships. IBM has been around for, what, 50 years now?
Your complaint is true of some tech companies (the.com boom is probably a good example), but you chose IBM and Intel for examples of companies that are "bad...at relationships"? Come *on*!
iChat is a perfect example in os x, where aim popup windows are initially transparent and out of the way. You are probably aware of how windows AIM handles this. it's obnoxious, gets in your way, and overlaps whatever you working on. what if that window appeared transparently on top of your current window allowing you to first finish your task then answer the message? this is just one of many examples i could give where true transparency is not only "eye candy" but genuinely useful.
Sure. The problem is that in most scenerios where translucency would work, I generally see simply effective window-switching as being a better solution, because then the same pixels aren't having a bunch of stuff on top of it interfering with visual recognition. In your iChat example, I'd probably do what some IM clients have done before, and simply bring up a small (translucent, if you like) indicator that a new message has arrived. Tapping a button or a mouse acknowledges that message.
Transparency tends to be more usable if at least one of the two (or more, heaven forbid) layers involved is fairly simple and large, which is why I mentioned the track list information as being a valid use of transparency. Here, there is only one line of text, and even against a complext background, one does not generally run into difficulty or slowdowns dealing with two superimposed images. I'm not saying that transparency doesn't have a place, just that it's been oversold as a HCI element thus far, and barring new ideas in user interface (perhaps the concept of some sort of large general status indicators being placed on the screen in very transparent, muted colors), I think it's going to stay there.
Your other two suggestions -- the "genie effect" (and I believe that you're right, it can't be done efficiently as it relies on true transparency -- though it's not a vector graphic effect, rather mutation of a raster image) and a 3d rotate transition effect when switching desktops (similar issue, relies on dumping a screen as far as I know) are not bad in any way, but they are certainly not more significant than eye candy. The "genie effect" is functionally nothing more than a fancy form of the Finder ZoomRects that classic Mac OS had in at least System 6, and probably well before. The transition effect is pretty and currently novel, but ultimately adds nothing to the functionality of the system.
From what I can tell from the discussion here, XFree does not contain GPLed code. As a matter of fact, if the old XFree86 license and the GPL are actually incompatible, it had *darn* well better not.
There are a few people saying that code that was submitted to the Linux kernel was also submitted to XFree86 -- all that means is that the author did an implicit dual license.
The issue is (probably) not GPL incompatibility. While Red Hat pushes hard to be as Open Source as possible, (and harder than most of the major vendors, especially our favorite fallen angel, Caldera). They've shipped closed-source software before (Netscape Navigator) when there was no usable open source alternative, and they ship many GPLed packages.
The problem may be that they have to put this attribution snippet on any ads for their distro if they include XFree86 in a feature list.
Frankly, I really don't like the whole advertising clause. I like it as little as I like Stallman's stupid "GNU" prefixes. If I want to know who wrote software, there are many places to find out. The AUTHORS/CREDITS file, an about box in the program, docs...it's there *if you want to look it up*. On the other hand, XFree86 hijacks ad space on everyone else's ads, and the same goes for Stallman (well, in his ideal world.) While it certainly is someone's right to release their own work under whatever license they want, it would be nicer to let people get information if they want it, but not constantly throw it in their face.
I suspect that the politics and feelings involved in XFree86 are a lot more complicated and have been going on for longer than Slashdot's random and recent reporting on the subject.
For all I know, Dawes could be an awful person, but (from a wholely ignorant outside viewer) it seems that the Slashdot story submitters have a habit of slagging on him a bit much.
I think that, no matter what, some kind of fork is going to happen. This really sucks, because if Dawes and the other core folks are willing to take all the crap they get (XFree86 and X11 may be two of the most unjustly maligned projects I know of) and can handle such a large package (there are *verY* few projects the size of X11 that constantly run as root, directly access hardware, and must remain stable and responsive)
I do hope that there isn't any bad blood over it. The Samba team managed to work things out without pain in much the same way the gcc team did. Basically, the folks that felt that the package needed to be stable just sat on their version and maintained it, only slowly improving it. The other folks took a version and ran with it, throwing tons of features in and going absolutely crazy. And in both cases, the projects were eventually re-merged (once the new fork was stable). They became simply long-term unstable branches under different leadership.
The simple truth is XFree86 is not capable of the features one should expect from a modern display system. Take one look at Mac OS X's Quartz Extreme in a CompUSA to get a good example as to why XFree86 (NOT X11, that isn't the problem) needs to shape up it's act.
Really? Specifically what problems do you have with XFree86, or are you just talking about the most common themes in XFree86 desktop environments?
Let the ludites running 486's keep their XFree86, and let us get on with our lives using a modern X11 implementation with real features like true transparency
You know, one of the things that people like about Linux is that it doesn't have crazy hardware requirements. Unlike OS X, or, to a lesser extent, Windows.
Transparency can be nice, but honestly, it adds very little functionality to a desktop environment. Antialiased text was a different story -- it allows a user to be given more data, by using gray levels. Plain old window transparency isn't good for a lot other than eye candy. And that eye candy is largely novelty ("look, I have transparent windows!"), and not necessary a long-term draw. I've tried working with transparent windows, and never been too impressed. Generally, interfaces are fairly modal at the window level -- I'm working with a single widget, and don't need to see what's behind it, and I'd rather devote the pixels composing that widget to making the widget easily recognizable, instead of giving some information about what's behind it. It just makes it harder to see what's being worked on. The reason windows are draggable is so that you can drag them into a configuration where you can see both windows that you're working with for the rare occasions when you need to have multiple windows visible at once.
There are a few cases for transparency. It's nice for onscreen display type elements -- if someone wants to display song titles from their player, for example, they might be into displaying it transluencly. Frankly, though, the desktop metaphor is not a transparency-oriented one, and I've yet to see good improvements suggested to it that require translucency.
vector scaling
XFree86 can do vector graphics via OpenGL.
and GPU acceleration.
XFree86 has extensive support for both 2d and 3d acceleration.
The rationale behind avoiding monoculture is that not all members have the same weaknesses, so an attack will not destroy the entire population. While this is a valid point for biological populations, there are some issues with it as apply to computer security. We are not dealing with "members" getting "killed" -- we are dealing with "computers" being "compromised".
The first issue is that many elements of the whole in some computer systems have the same degree of access. Perhaps half of the workstations at a company run Linux and half Windows. If all of them have roughly the same tasks (as opposed to devoting Windows to web browsing and Linux to email reading), then a compromise of *any* of them allows a compromise of all the important data. Many security systems are weakest-link -- if one element can be compromised, the whole system falls. In this case, all having a polyculture does is expose more weaknesses, reducing the security of the system as a whole.
The second element is somewhat similar -- most computer networks have some degree of trust relationship between members. It may be something explicit, like having IP-based rsh auth (though that's a bit of an old problem) or allowing access to various intranet Web pages to any internal computers. It may be just allowing a compromised computer to sniff a network that other computers pass traffic over. In this case, a compromise of one member of the network provides an attack vector against the other members of the network. Again, a polyculture exposes more weaknesses, weakening the security of the system as a whole.
Third, there are security management issues. Most medium or large computer networks have someone or some group with some degree of responsibliity for computer security. That group usually has finite resources and budget. Much of their effort can generally be replicated across similar members -- for example, securing a plaintext authentication in Windows means a fix that just has to be replicated across all members in the network. If their time and money must be spread across multiple types of members, they are less able to spend resources on any one group, and each type of member may be less well managed.
Fourth, most networks do not follow a "Russian doll" approach, where a potential cracker must compromise first one computer, then another computer, then another computer to get in to the network proper from the outside. In such a scenerio, making each of the dolls different does improve security, since a cracker must compromise all, rather than just one, system. It's pretty common to just have a NATted network with all hosts inside at roughly the same level of internal access, however.
Overall, I *do* think that it's a good idea to move away from "Microsoft only" on computer networks. Competition tends to improve products, and Microsoft has a poor security track record (and doesn't focus on security very well). However, if an CIO has the sole goal of improving security, and has the choice of rolling out Linux or rolling out Kerberos on existing Windows boxes, I'd have to say that rolling out Kerberos is probably going to do more for security.
There are a number of reasons that CS majors generally don't need particularly powerful machines. For most work in CS, you simply aren't doing masses of computation. If you're writing something that does, you can generally run it on a powerful machine somewhere. Contrast this with, say, a mechanical engineering student, that may want to do stress analysis on their home computer to avoid having to go to a lab to do so.
Compiling code isn't a real-time, interactive task, and generally doesn't take all that long. We have pretty good systems to cache built components of software.
Furthermore, I've found that a lot of CS students know how to poke their system to run pretty efficiently, so they don't need a really jacked system.
If I had to recommend a college system...hmm.
First, laptops are popular right now, as in they're selling really strongly. Laptops can be nice, but neither are they necessary -- they tend to be more fragile, more easily lost or stolen, not very upgradable, and less comfortable to use for extended periods of time. I don't know a lot of people that do serious note-taking with laptops, though I have seen people playing games in classes with laptops. Oh, and they don't tend to be as nice for game-playing, and college dorms are a fun place to play multiplayer games. They don't seem to be a particularly necessary item. On the other hand, they do easily let you move to a lounge or library to work with someone else, they let you take advantage of more and more common wireless compus networks, and they're much easier to take home with you during the holidays.
I think that most people are going to want to upgrade their computer. They're probably going to want to upgrade at least hard drive and memory, some time in college. Since you're likely to do some upgrading, buying the fanciest system you can afford right at the start seems like a bit of a waste.
I'd recommend a quiet keyboard. Some roommates can be irritated by constant clicking. The same goes for a good pair of headphones with a long cord. College dorms are a place where you want to play speakers, but you're surrounded on all sides by people and likely have a roommate. You don't have to worry about rules about playing music too late, and it's much easier to afford really good quality headphones than really good quiality speakers.
I'm dubious about the story submitter's intention to get a tablet PC. If they're sure that that's what they want, well and good, but if they just really like the idea, I dunno if they'll be that happy with them. Tablets are expensive, less powerful than similarly-priced computers, aren't upgradeable, and really haven't caught on because folks don't seem to like them that much. The only real reason I can think of for getting a tablet is if you really want to use drawing input as a primary form of input. Frankly, for almost everything, drawing is a lousy form of input. It's slower than just about anything else. If you're taking art classes, a stand-alone drawing tablet (Wacom or similar) is inexpensive and mature.
In this very thread I covered the fact that 'massive' does not mean 'omnipresent'. There will be packages that are not covered, as you indeed go on to talk about.
[shrug] I know people that use only packaged software. But suppose we do install, say, ten unpackaged software packages -- maybe some special emacs mode and some driver for some oddball hardware. I do know that it's a hell of a lot easier to manage a RH system with packages than with not -- 990 packages are being managed for me, and I only have to worry about ten. Handling unmanaged systems without automatic updating is a pain in the ass that I don't really want to ever go back to, having escaped it.
You can do that on Mac systems too, using mkbom and the developer tools for package management.
The entire point was that I could create a package and have that package updated and handled automatically once the software package is packaged by my vendor. Apple doesn't provide any such mechanism.
And me. Most of my systems management experience is in Solaris, and Linux. I hate SysV init. You've never fully appreciated overkill until you observe that their are eleven states you can put your server into.
Well, I don't have enough time on BSD systems to reasonably complain one way or another. I think that the fact that an overwhelming number of Linux distributions, given the choice between BSD init and SysV init, went with SysV init is a pretty strong argument, though.
Well, in SysV, you've got an interesting race condition on your hands. The BSD signal handler introduce ways in which signals could be vetoed or suspended over blocks of code.
Wow...I had to look at the GNU signal man page to even figure out what you were talking about. This issue hasn't been seen on Linux since glibc 2 was introduced -- what is that, seven years ago?
Yes it was, although Classic MacOS didn't have fat binaries. Todays topic: NeXT Computer, Inc. (later NeXT Software, Inc.)
Classic Mac OS certainly *did* have fat 680x0/PPC binaries. Fat binaries were a common way of shipping software for years after the 6100/7100/8100 debuted. Look at old entries on the Info-Mac or Merit archives, and you'll see a ton of fat binaries. For a while it was very common to ship fat binaries. The PPC code went in the data fork, and the 680x0 code in CODE resource entries in the resource fork.
To be honest, I'm not sure why you'd want fat binaries on Mac OS X, given that it only runs on PPC, and only will do so for the forseeable future.
No it's not. But that's only because you're using a system that doesn't support it. Think of this: I can develop an application on my black slab, then bundle it up and distribute that one bundle to people using SPARC, M68k, Intel and HP-PA. This saves me a lot of work. Have a look too at the Darwin installation CD available from Apple. That one CD will install on both PowerPC and IA32 architectures. Let's say I decided that I didn't like any of the systems available today, and went out to create uber-UNIX for IA32, IA64, PowerPC, Itanium, SPARC, M68k, ARM, HP-PA, PDP-7 and Alpha. Wouldn't it be great if I only had one distribution set to maintain? Wouldn't it be great if all of my developers could compile for all supported platforms, without the tedium involved with cross-compiling on less aware systems?
No, it wouldn't be great, because it would make every software release ten times as large (assuming all the architectures you listed above) as it would normally be. That's *wasteful* -- why would I want to download forty CDs worth of software when I could download four CDs worth of software (a typical Linux distribution)? It's not really hard to just download a PowerPC package, you know?
I'm sorry, but the article you linked to doesn't give enough information for those benchmarks to be supported. For instance, when comparing file system latencies, why does the benchmarker not tell us which filesystem he is us
That's funny. I know an engineering student at VT, and they have to use a Windows box.
now Apple has taken the OS to where nearly every Computer Science PHD speculated in the late 80s when they imagined: "What do we *really* want from an OS."
If by that you mean "slow, making poor use of RAM, and full of useless pixmaps".
Windows does not add $500 to the price tag of an OEM system. From what I've been able to figure out, big OEM costs are about $30 a machine. If anything, the Mac OS probably adds at *least* $30/machine in OS development costs, given the worse economies of scale.
OS X (and OS 9, as well, if that's your cup of tea) wakes from sleep in less than a second (to displaying the desktop),
Just to clear you up on this point, laptops today do this by dumping the video image to disk before sleeping, since usually people don't hit a key for a couple secs or so. The OS design has very little to do with percieved wakeup speed.
Slashdot Mirror
If you want to technically guarantee anonymity to prevent censorship and violations of the right to free speech by technical means, there is no way to distinguish legitimate users from assholes. The software cannot do this, by definition.
Well, you're half right, but you're coming to a bogus conclusion.
Yes, I agree that it's probably infeasible for a software package alone to currently do a good job of determining whether someone is an "asshole". However, we have trust networks and other mechanisms, where software acts as a powerful tool to assist human rating. Note that Freenet is inherently something of a trust network, as its native interface is a Web browser with Web pages. You presumably link to people that are not "assholes", who link to people that *they* feel are not "assholes".
I've even seen posts saying (to paraphrase) 'everyone should have free speech except kiddie pornographers and nazis'.
Sounds pretty European to me. A lot of European countries *really* still have a lot of social crap left over from after World War II. Garmany and France, in particular, are incredibly uptight (at least from an American standpoint) about Nazi-related stuff. The UK has some kind of pedophiliaphobia. I mean, sure, nobody likes the worst-case sort of sexual content related to kids -- kids getting abducted and raped or similar. However, the British are absolutely rabid about avoiding any kind of surfacing of pedophilia. Really unusual.
Arguing that a lot of Europeans should loosen up about Nazism is probably a fun debate, but it doesn't have as much impact in the US, so I'll argue the child pornography standpoint.
Frankly, I never really saw how banning child pornography has significant social benefits (especially since posession and distribution of blood sport content *is* legal). Sure, there can be all kinds of bad things associated with child pornography -- people worry about their kids getting abducted and raped or something -- but I don't see quite how eliminating distribution of such material on Freenet does anything to avoid real life sexual abuse.
The first argument I see in favor of censorship of child pornography might be that if there is a profitable industry for producing child pornography, then children will be involved in production of that pornography (as opposed to actors/actresses appearing to be children, or CG doctoring, or pornographic animation containing depictions of underage sex). This may lead to one of two potentially bad things: first, children may be nude in videos. In the Victorian tradition, being seen nude (particularly females) somehow "degrades" or damages future social standing. I've nver seen this as an immutable -- our society happens to have a nudity taboo, but it is arguable as to whether that is at all beneficial to society. The second bad point is that children may be physically injured in the production of such content. There are laws already dealing with injuring children -- I'm not sure why a special point needs to be made for pornography-related content.
The second issue might be that distribution of child ponography tends to exacerbate pedophiliac behavior. This is certainly a decent thought, but I'm not sure how grounded in reason it is. We in the United States allow distributing movies showing images of people being shot (actually, I just posted regarding this earlier today). It doesn't seem that the violence present alone (see my complaints about cartoon violence, which are different) significantly have had an impact on the increase of violence. Why would we think that sexual content alone would drive someone to engage in a sexual act?
Anti-child pornography laws are one of the few near-global laws, and I'm a bit curious why, as they seem to be the product of fear and emotion rather than a particularly reasoned decision -- at the very least, they are inconsistent with decisions about censorship that we have made in other content areas.
I'll let others wonder why the US can get away without censoring Nazi content and yet doesn't have massive Nazi surges, yet France feels the need to prevent people from having Nazi content.
But hey, way to go with your sly anti-businessman attack. Because as everyone knows, MBAs are all simpletons and schoolyard bullies.
Phipps is actually not an MBA -- he's an electrical engineer.
Frankly, I wish that both Phipps and ESR would stop acting like reactinary idiots. It does absolutely nothing but hurt open source. ESR blasted off a ridiculously inaccurate and amaturish letter to Sun, and decided to make it an "open" letter to piss off the maximal number of people. He managed to trade in some Open Source credibility to advance his whim-of-the-day. Phipps, instead of doing the right thing and either ignoring ESR or sending him a form letter, actually responded. Had he just ignored ESR, the whole thing would have died. Now it's alive *again*. Christ. This whole argument is so incredibily stupid.
Read ESR's essay again...and if you still don't get it, I'm sorry for ya.
How is that an argument?
How about "Read the Communist Manifesto again...and if you still don't get it, I'm sorry for ya."? Did *that* do anything to convince you?
ESR is a voice in the Open Source movement BECAUSE HE IS ONE. A lot of people interview him for things. A lot of people seek his opinions. He gets invited to speak and consult. People listen to him. He is the embodiment of the movement's very principles! If you don't like how he does "spokesperson", I suggest you go try and replace him. Put out a better implementation - yourself, or whoever.
He just did. He threw out some verbal criticism, which is what's at issue, not software.
Sun gave us OpenOffice.
ESR gave us fetchmail.
Which do you think helped Open Source more?
But yeah, the thing that ticks me off the most is that they say they understand Open Source.
Did you read ESR's original letter? It was awful.
Sun understands Open Source a lot better than ESR understands business issues, that's for sure.
Here's *my* open letter to Sun management:
Don't respond to stpuid letters when you dnn't need to do so. The Open Source community will not hold you responsible.
ESR has some name recognition, but you are under absolutely no obligation to respond.
If you wanted to do so, a nice "thanks, we'll take your thoughts into consideration" note would work nicely at tamping the thing down. ESR clearly didn't understand the issues involved, and made quite an ass out of himself with his letter. It would have been quite easy to simply ignore the letter or send back a form letter and let the whole thing die down and be forgotten. Now you've managed to insult him and start a potential flamewar. For Chrissake, use some common sense.
The idea of (even parental) censorship is not without value, but I think that it's overvalued.
The rationales for parental censorship that I can see go something like this (w/ my responses):
Issue: If I expose my child to this scary content, he/she is not old enough to have mental constructs or required knowledge in place to prevent him/her from being overwhelmed with irrational fear.
Response: I don't think I can agree. The mental constructs issue is, I think, not a convincing argument. The way people seem to develop contructs to deal with frightening-but-not-dangerous content is by being exposed to exactly that content. You can wait until they're twenty or start when they're six. I think that, ultimately, kids are going to have to have a few frightening-but-not-dangerous-experiences to learn how to deal with fright. My own parents put strict limits on what frightening movies I could watch as a child. The Thing was the first R-rated movie I saw. When I finally saw it, I was quite frightened. My friends, who had been watching frightening movies earlier than I was, were decidedly unintimidated. Today, I shrug off The Thing and similar movies. It took exposure to a good amount of frightening content, though, to build up that state of mond. I don't think that age is a factor so much as experience.
Knowledge base is a bit more convincing. I really think that a large factor is knowledge directly related to frightening content. In this case, parents can pretty easily talk about it. For example, we really don't have any reason to think that Jason from Friday the 13th could exist. It's just as reasonable to think that there's a horde of guardian angels running around overhead, a decidedly less frightening prospect. Furthermore, Jason is some guy in a (not all that great, frequently) costume. If a child is frightened by a movie, a parent can do a bit of explaining to avoid most of it.
I think that there *is* something to the argument that there is non-movie-specific knowledge that must be had to help deal with fear, however. Generally, if a movie represents something that isn't a possible danger, there's no reason to be frightened of it. A lot of things can be ruled out as possible dangers with a knowledge base -- i.e. it's pretty unlikely that an A.I. is going to fall in love with one of its creators and kill off the creator's spouse, because such behavior is not exactly easy to impart, is probably not easily accidently evolved in the kind of environment an A.I. exists in, and is pretty complex. There's not a lot of way to ensure that a child has a wide enough knowledge base to produce explanations and assign those explanations a high enough "convincing factor" (for lack of a better term). Furthermore, risk aversion is a common mental irrationality that humans are prone to that children seem to be even more vulnerable to. If shown something that is a real, potential danger, such as dying from a rattlesnake bite in California, people do not deal well with grappling with the very small degrees of probability involved, and will assign too much danger to that bite, even though car crashes pose a far greater risk to them. That's something that I think it takes a significant amount of experience to overcome. It is *possible* that a small child simply has not taken enough life risks and reasoned about them to be able to assign slight dangers a proper amount of risk.
Issue: Marketers design their campigns specifically to appeal to humans. I don't want them to be exposed to tobacco/alcohol, and let those marketers to get their hooks into my child.
Response: I really feel that a better way to deal with marketing of potentially harmful products to a child is to attempt to innoculate him to that product, rather than isolate him from it. You simply are not going to be able to shield a child forever from a product, and I think that it would be better to simply ensure that they can make an intelligent decision early on. Explaining to a child why cigarettes are b
Didn't Fred Anderson also become acting CEO of Apple for some time after Gil left?
The name may not be as familiar as past Apple CEOs, but he's certainly been a major player.
IBM is a huge company. The people making hard drives probably have very little to do with the people making Linux other than the same style of standardized HR forms. IBM in particular seems to operate divisions with a good deal of modularity, from what I've heard from people working there.
You can also always get lower-capacity drives in the form of SCSI devices rather than IDE...
For Chrissake, man, IBM knows relationships. IBM has been around for, what, 50 years now?
.com boom is probably a good example), but you chose IBM and Intel for examples of companies that are "bad...at relationships"? Come *on*!
Your complaint is true of some tech companies (the
iChat is a perfect example in os x, where aim popup windows are initially transparent and out of the way. You are probably aware of how windows AIM handles this. it's obnoxious, gets in your way, and overlaps whatever you working on. what if that window appeared transparently on top of your current window allowing you to first finish your task then answer the message? this is just one of many examples i could give where true transparency is not only "eye candy" but genuinely useful.
Sure. The problem is that in most scenerios where translucency would work, I generally see simply effective window-switching as being a better solution, because then the same pixels aren't having a bunch of stuff on top of it interfering with visual recognition. In your iChat example, I'd probably do what some IM clients have done before, and simply bring up a small (translucent, if you like) indicator that a new message has arrived. Tapping a button or a mouse acknowledges that message.
Transparency tends to be more usable if at least one of the two (or more, heaven forbid) layers involved is fairly simple and large, which is why I mentioned the track list information as being a valid use of transparency. Here, there is only one line of text, and even against a complext background, one does not generally run into difficulty or slowdowns dealing with two superimposed images. I'm not saying that transparency doesn't have a place, just that it's been oversold as a HCI element thus far, and barring new ideas in user interface (perhaps the concept of some sort of large general status indicators being placed on the screen in very transparent, muted colors), I think it's going to stay there.
Your other two suggestions -- the "genie effect" (and I believe that you're right, it can't be done efficiently as it relies on true transparency -- though it's not a vector graphic effect, rather mutation of a raster image) and a 3d rotate transition effect when switching desktops (similar issue, relies on dumping a screen as far as I know) are not bad in any way, but they are certainly not more significant than eye candy. The "genie effect" is functionally nothing more than a fancy form of the Finder ZoomRects that classic Mac OS had in at least System 6, and probably well before. The transition effect is pretty and currently novel, but ultimately adds nothing to the functionality of the system.
From what I can tell from the discussion here, XFree does not contain GPLed code. As a matter of fact, if the old XFree86 license and the GPL are actually incompatible, it had *darn* well better not.
There are a few people saying that code that was submitted to the Linux kernel was also submitted to XFree86 -- all that means is that the author did an implicit dual license.
The issue is (probably) not GPL incompatibility. While Red Hat pushes hard to be as Open Source as possible, (and harder than most of the major vendors, especially our favorite fallen angel, Caldera). They've shipped closed-source software before (Netscape Navigator) when there was no usable open source alternative, and they ship many GPLed packages.
The problem may be that they have to put this attribution snippet on any ads for their distro if they include XFree86 in a feature list.
Frankly, I really don't like the whole advertising clause. I like it as little as I like Stallman's stupid "GNU" prefixes. If I want to know who wrote software, there are many places to find out. The AUTHORS/CREDITS file, an about box in the program, docs...it's there *if you want to look it up*. On the other hand, XFree86 hijacks ad space on everyone else's ads, and the same goes for Stallman (well, in his ideal world.) While it certainly is someone's right to release their own work under whatever license they want, it would be nicer to let people get information if they want it, but not constantly throw it in their face.
I suspect that the politics and feelings involved in XFree86 are a lot more complicated and have been going on for longer than Slashdot's random and recent reporting on the subject.
For all I know, Dawes could be an awful person, but (from a wholely ignorant outside viewer) it seems that the Slashdot story submitters have a habit of slagging on him a bit much.
I think that, no matter what, some kind of fork is going to happen. This really sucks, because if Dawes and the other core folks are willing to take all the crap they get (XFree86 and X11 may be two of the most unjustly maligned projects I know of) and can handle such a large package (there are *verY* few projects the size of X11 that constantly run as root, directly access hardware, and must remain stable and responsive)
I do hope that there isn't any bad blood over it. The Samba team managed to work things out without pain in much the same way the gcc team did. Basically, the folks that felt that the package needed to be stable just sat on their version and maintained it, only slowly improving it. The other folks took a version and ran with it, throwing tons of features in and going absolutely crazy. And in both cases, the projects were eventually re-merged (once the new fork was stable). They became simply long-term unstable branches under different leadership.
The simple truth is XFree86 is not capable of the features one should expect from a modern display system. Take one look at Mac OS X's Quartz Extreme in a CompUSA to get a good example as to why XFree86 (NOT X11, that isn't the problem) needs to shape up it's act.
Really? Specifically what problems do you have with XFree86, or are you just talking about the most common themes in XFree86 desktop environments?
Let the ludites running 486's keep their XFree86, and let us get on with our lives using a modern X11 implementation with real features like true transparency
You know, one of the things that people like about Linux is that it doesn't have crazy hardware requirements. Unlike OS X, or, to a lesser extent, Windows.
Transparency can be nice, but honestly, it adds very little functionality to a desktop environment. Antialiased text was a different story -- it allows a user to be given more data, by using gray levels. Plain old window transparency isn't good for a lot other than eye candy. And that eye candy is largely novelty ("look, I have transparent windows!"), and not necessary a long-term draw. I've tried working with transparent windows, and never been too impressed. Generally, interfaces are fairly modal at the window level -- I'm working with a single widget, and don't need to see what's behind it, and I'd rather devote the pixels composing that widget to making the widget easily recognizable, instead of giving some information about what's behind it. It just makes it harder to see what's being worked on. The reason windows are draggable is so that you can drag them into a configuration where you can see both windows that you're working with for the rare occasions when you need to have multiple windows visible at once.
There are a few cases for transparency. It's nice for onscreen display type elements -- if someone wants to display song titles from their player, for example, they might be into displaying it transluencly. Frankly, though, the desktop metaphor is not a transparency-oriented one, and I've yet to see good improvements suggested to it that require translucency.
vector scaling
XFree86 can do vector graphics via OpenGL.
and GPU acceleration.
XFree86 has extensive support for both 2d and 3d acceleration.
The rationale behind avoiding monoculture is that not all members have the same weaknesses, so an attack will not destroy the entire population. While this is a valid point for biological populations, there are some issues with it as apply to computer security. We are not dealing with "members" getting "killed" -- we are dealing with "computers" being "compromised".
The first issue is that many elements of the whole in some computer systems have the same degree of access. Perhaps half of the workstations at a company run Linux and half Windows. If all of them have roughly the same tasks (as opposed to devoting Windows to web browsing and Linux to email reading), then a compromise of *any* of them allows a compromise of all the important data. Many security systems are weakest-link -- if one element can be compromised, the whole system falls. In this case, all having a polyculture does is expose more weaknesses, reducing the security of the system as a whole.
The second element is somewhat similar -- most computer networks have some degree of trust relationship between members. It may be something explicit, like having IP-based rsh auth (though that's a bit of an old problem) or allowing access to various intranet Web pages to any internal computers. It may be just allowing a compromised computer to sniff a network that other computers pass traffic over. In this case, a compromise of one member of the network provides an attack vector against the other members of the network. Again, a polyculture exposes more weaknesses, weakening the security of the system as a whole.
Third, there are security management issues. Most medium or large computer networks have someone or some group with some degree of responsibliity for computer security. That group usually has finite resources and budget. Much of their effort can generally be replicated across similar members -- for example, securing a plaintext authentication in Windows means a fix that just has to be replicated across all members in the network. If their time and money must be spread across multiple types of members, they are less able to spend resources on any one group, and each type of member may be less well managed.
Fourth, most networks do not follow a "Russian doll" approach, where a potential cracker must compromise first one computer, then another computer, then another computer to get in to the network proper from the outside. In such a scenerio, making each of the dolls different does improve security, since a cracker must compromise all, rather than just one, system. It's pretty common to just have a NATted network with all hosts inside at roughly the same level of internal access, however.
Overall, I *do* think that it's a good idea to move away from "Microsoft only" on computer networks. Competition tends to improve products, and Microsoft has a poor security track record (and doesn't focus on security very well). However, if an CIO has the sole goal of improving security, and has the choice of rolling out Linux or rolling out Kerberos on existing Windows boxes, I'd have to say that rolling out Kerberos is probably going to do more for security.
There are a number of reasons that CS majors generally don't need particularly powerful machines. For most work in CS, you simply aren't doing masses of computation. If you're writing something that does, you can generally run it on a powerful machine somewhere. Contrast this with, say, a mechanical engineering student, that may want to do stress analysis on their home computer to avoid having to go to a lab to do so.
Compiling code isn't a real-time, interactive task, and generally doesn't take all that long. We have pretty good systems to cache built components of software.
Furthermore, I've found that a lot of CS students know how to poke their system to run pretty efficiently, so they don't need a really jacked system.
If I had to recommend a college system...hmm.
First, laptops are popular right now, as in they're selling really strongly. Laptops can be nice, but neither are they necessary -- they tend to be more fragile, more easily lost or stolen, not very upgradable, and less comfortable to use for extended periods of time. I don't know a lot of people that do serious note-taking with laptops, though I have seen people playing games in classes with laptops. Oh, and they don't tend to be as nice for game-playing, and college dorms are a fun place to play multiplayer games. They don't seem to be a particularly necessary item. On the other hand, they do easily let you move to a lounge or library to work with someone else, they let you take advantage of more and more common wireless compus networks, and they're much easier to take home with you during the holidays.
I think that most people are going to want to upgrade their computer. They're probably going to want to upgrade at least hard drive and memory, some time in college. Since you're likely to do some upgrading, buying the fanciest system you can afford right at the start seems like a bit of a waste.
I'd recommend a quiet keyboard. Some roommates can be irritated by constant clicking. The same goes for a good pair of headphones with a long cord. College dorms are a place where you want to play speakers, but you're surrounded on all sides by people and likely have a roommate. You don't have to worry about rules about playing music too late, and it's much easier to afford really good quality headphones than really good quiality speakers.
I'm dubious about the story submitter's intention to get a tablet PC. If they're sure that that's what they want, well and good, but if they just really like the idea, I dunno if they'll be that happy with them. Tablets are expensive, less powerful than similarly-priced computers, aren't upgradeable, and really haven't caught on because folks don't seem to like them that much. The only real reason I can think of for getting a tablet is if you really want to use drawing input as a primary form of input. Frankly, for almost everything, drawing is a lousy form of input. It's slower than just about anything else. If you're taking art classes, a stand-alone drawing tablet (Wacom or similar) is inexpensive and mature.
In this very thread I covered the fact that 'massive' does not mean 'omnipresent'. There will be packages that are not covered, as you indeed go on to talk about.
[shrug] I know people that use only packaged software. But suppose we do install, say, ten unpackaged software packages -- maybe some special emacs mode and some driver for some oddball hardware. I do know that it's a hell of a lot easier to manage a RH system with packages than with not -- 990 packages are being managed for me, and I only have to worry about ten. Handling unmanaged systems without automatic updating is a pain in the ass that I don't really want to ever go back to, having escaped it.
You can do that on Mac systems too, using mkbom and the developer tools for package management.
The entire point was that I could create a package and have that package updated and handled automatically once the software package is packaged by my vendor. Apple doesn't provide any such mechanism.
And me. Most of my systems management experience is in Solaris, and Linux. I hate SysV init. You've never fully appreciated overkill until you observe that their are eleven states you can put your server into.
Well, I don't have enough time on BSD systems to reasonably complain one way or another. I think that the fact that an overwhelming number of Linux distributions, given the choice between BSD init and SysV init, went with SysV init is a pretty strong argument, though.
Well, in SysV, you've got an interesting race condition on your hands. The BSD signal handler introduce ways in which signals could be vetoed or suspended over blocks of code.
Wow...I had to look at the GNU signal man page to even figure out what you were talking about. This issue hasn't been seen on Linux since glibc 2 was introduced -- what is that, seven years ago?
Yes it was, although Classic MacOS didn't have fat binaries. Todays topic: NeXT Computer, Inc. (later NeXT Software, Inc.)
Classic Mac OS certainly *did* have fat 680x0/PPC binaries. Fat binaries were a common way of shipping software for years after the 6100/7100/8100 debuted. Look at old entries on the Info-Mac or Merit archives, and you'll see a ton of fat binaries. For a while it was very common to ship fat binaries. The PPC code went in the data fork, and the 680x0 code in CODE resource entries in the resource fork.
To be honest, I'm not sure why you'd want fat binaries on Mac OS X, given that it only runs on PPC, and only will do so for the forseeable future.
No it's not. But that's only because you're using a system that doesn't support it. Think of this: I can develop an application on my black slab, then bundle it up and distribute that one bundle to people using SPARC, M68k, Intel and HP-PA. This saves me a lot of work. Have a look too at the Darwin installation CD available from Apple. That one CD will install on both PowerPC and IA32 architectures. Let's say I decided that I didn't like any of the systems available today, and went out to create uber-UNIX for IA32, IA64, PowerPC, Itanium, SPARC, M68k, ARM, HP-PA, PDP-7 and Alpha. Wouldn't it be great if I only had one distribution set to maintain? Wouldn't it be great if all of my developers could compile for all supported platforms, without the tedium involved with cross-compiling on less aware systems?
No, it wouldn't be great, because it would make every software release ten times as large (assuming all the architectures you listed above) as it would normally be. That's *wasteful* -- why would I want to download forty CDs worth of software when I could download four CDs worth of software (a typical Linux distribution)? It's not really hard to just download a PowerPC package, you know?
I'm sorry, but the article you linked to doesn't give enough information for those benchmarks to be supported. For instance, when comparing file system latencies, why does the benchmarker not tell us which filesystem he is us
Actually, with Executor, you can run 680x0 Mac stuff on Linux or Windows.
Emulating the PPC on the x86 efficiently is much harder than the other way around because of the number of registers on the PPC.
Virginia Tech
That's funny. I know an engineering student at VT, and they have to use a Windows box.
now Apple has taken the OS to where nearly every Computer Science PHD speculated in the late 80s when they imagined: "What do we *really* want from an OS."
If by that you mean "slow, making poor use of RAM, and full of useless pixmaps".
Windows does not add $500 to the price tag of an OEM system. From what I've been able to figure out, big OEM costs are about $30 a machine. If anything, the Mac OS probably adds at *least* $30/machine in OS development costs, given the worse economies of scale.
OS X (and OS 9, as well, if that's your cup of tea) wakes from sleep in less than a second (to displaying the desktop),
Just to clear you up on this point, laptops today do this by dumping the video image to disk before sleeping, since usually people don't hit a key for a couple secs or so. The OS design has very little to do with percieved wakeup speed.