"As anyone who's been using Linux for several years (or even a few) for everyday tasks knows, "ready for the desktop" is in the eye of the beholder."
For instance, if you're a zealot who isn't willing to admit that Linux has flaws, Linux *is* ready for the desktop.
On the other hand if you are a rational human being, you might think of Linux as a great server operating system, that some people, painfully, force themselves to use on the desktop for ideological reasons. Kind of like the computing equivalent of a hair shirt.
See? It's all a matter of perspective.
Another matter of perspective. If you tend to agree with me on this issue, you might mod this article "funny," but if you disagree, you might mod it "flaimbait."
and do handle merging much better than svn (which hardly manages merging at all) however, a lot of companies use perforce, because perforce does handle branching and merging well, whereas it also allows for a centrol source repository.
Mercurial and get are mostly built around the model of sending your patches in via email. They make some noise about how you *could* build a central repository, but it's not like that out of the box, and things like permissions often aren't there yet.
Basically, the development model that the open source guys use to do their work isn't the model that guys at big corporations need to use, and distributed VC tools are so heavily geared towards the open source model (in the case of git, not even supporting windows that well) that they can't really be drop in replacements for tools like perforce.
This looks pretty aweful...
on
Iron Sky Trailer
·
· Score: 0, Redundant
This might make a good cheezy video game, but I can't imagine sitting through 2 hours of youtube quality acting. Check out the preview for the other film these guys created "star wreck." It looks like a tediously bad fan comedy.
The trailer might look kind of impressive, but from the website: "At the moment, the team is finalizing the script" In other words, the trailer has no relation to the final product in quality... and when they actually have to find live action "actors" to do the script... well...
whether some of these fiascos are Balmer's fault. Particularly, what happened with vista. It's very plausible that Vista died of feature bloat because Balmer didn't pay enough attention to it in the beginning; however, it's hard to tell from the outside who was really responsible.
Either way though, It's clear that some kind of shakeup needs to occur for Microsoft to continue to compete. I just don't see Microsoft being able to expand into new markets using the clumsy "throw money at the problem" approach that post-gates Microsoft has used. Money's an important tool for a company the size of Microsoft, but it can't cover up underlying problems, like a project that suffering feature creep, or a corporate culture that suppresses bottom up innovation.
Bill Gates seemed to run a much tighter ship overall, with a supposedly fairly "hands on" management style. However, it is true that Microsoft was a much smaller company under Gate's tenure, and I'm not sure he would be the man to put back in charge of the new Microsoft.
I know there are 32 bit PPC macs, but I thought that all X86 macs were 64 bit, since they started with the core 2 duo systems.
What models of macs came with a 32 bit OS? Or are they simply referring to mac os 10.4, which I believe had some 32 bit support, but didn't have all API's (like cocoa) available in 64 bit versions.
in a startup might be relatively young, just out of college and willing to take risks. However, as soon as that startup gets enough funding to hire some more employees and pay them a salary, they aren't going to go after other college grads with limited experience.
Funded startups typically go after top tier developers with lots of experience to get their product off the ground. They need these guys because the challenges a startup faces are huge compared to the number of people and infrastructure available. Also, they can actually get top tier people by 1. offering them top dollar. 2. offering them lots of pre-ipo stock 3. last (but not least) giving them something really challenging to work on.
Only the first few employees can be expected to work for free, so these more experience guys that come later actually aren't taking that much of a risk, as they still get their higher than usual salary, plus stock, which is just gravy if it pays off.
>I don't think New York has the authority to >do this. But I sure would like to see the >supreme Court act.
That seems iffy to me. Most of the goods I purchase at a store aren't from in state, but I still pay sales tax on them.
Just because Amazon's store is online, doesn't mean that they shouldn't have to pay the same sales tax everyone else does.
I live in Washington state, in Seattle, the same city where Amazon is based, and I already pay state sales taxes on Amazon goods.
Maybe there is some legal loophole that prevents states from raising taxes in these situation, but it seems like a loophole that should be closed. This isn't a interstate tariff we're talking about here.
He's basically putting up all the money for the operation on the vague hope that it will pay off someday. They really don't have a business model, just a really generous investor/CEO.
So... it's basically a charity based operating system.
Which raise the point, why is this douchebag http://www.interopnews.com/news/is-ubuntu-selling-out-or-growing-up.html writing an article about how the company is "selling out" by making some very small moves to make money off of an operating system they spend large amounts of money on, and give away for free?
It kind of pisses me off that random internet idiots who don't make software for a living call anyone who tries to a "sellout."
The article mentions that they are trying to recoup a small amount of the money they are dumping into Canonical by selling some proprietary software.
So what? I'm sick and tired of internet morons tearing apart people that actually have to work for a living. It's not enough that they give away most of their software for free and under an open source license, but if they charge for *anything*, if you develop one line of proprietary code and sell it to make a buck, some random jerkoff will mouth off at you about how "software wants to be free," and you're "oppressing" them with your price tag and your non-gpl license.
Free software isn't a business model. None of the distros that don't make you pay money *per install* make any money. Canonical loses money, Suse loses money. The only people who make money making operating systems do so by selling some proprietary code, or (as with red hat) devising schemes to make people pay money for shrink wrapped copies of open source code. Ubuntu has by far taken the least obnoxious approach, i.e. giving away most of their software, and letting you use their repository for free updates (which others don't do), but developing some proprietary stuff they let you buy separately.
you're facing a difficult ethical dilemma, so clearly you made the right decision by asking for help from a bunch of anonymous internet forum morons.
In addition to conspiracy theories about how Microsoft and the RIAA conspired to kill your friend's brother, you can expect to recieve such insightful gems on your crisis as, "in soviet Russia, myspace suicides you."
why is this news? Professionals might use dreamweaver or equivalent for a little quick prototyping, but nothing spat out by a wysiwyg is going on the web.
The days of everyone's websites being filled with "authored by frontpage" comments are long over.
How can he be debunked? His primary contribution is the observation that unconstrained population grows exponentially. He then observed, that for his time the primary constraint on population growth was food supply.
There's really not much to debunk there.
There are other points that he makes that you can attack, but those are not points I relied on.
that "35 year high" for fertility is only 2.1 for each woman. That is extremely low by global standards. It is *only* a high for industrialized countries.
It's like saying that Anchorage is warm *compared to the rest of alaska* and concluding from that, that anchorage must be a tropical tourist resort.
One problem, is that even if there are "open source" drivers available, for what platform? Mac OSX? Solaris? BSD? Linux? Windows?
Or maybe they are drivers for an older kernel version? New kernels don't link against old drivers, and new versions of GCC don't necessarily *compile* old drivers.
And then of course, there's installation of drivers. Common stuff like video drivers tend to be included in distros... but what about things like fingerprint readers, USB printers, etc? Installing a driver, doesn't just mean running an installer. There's also the extra step of compiling for your exact kernel version. Which means you must have GCC installed, on a desktop machine that you may not want to do any development on at all.
Does this seem like a really bad system to anyone else? Why do we *ever* need to write the same driver more than once? Why not just have a standard interface for talking to X86 drivers for things like audio, video, printers, harddrives etc?
Then, one driver for Linux, OSX, Windows, etc can be passed around and we don't have to continually reinvent the wheel. Then a linux installer can get extra drives by pulling them out of a linux install. Then we can write whole new kernels without having to write tons of custom drivers for it.
Seriously, someone needs to the Linux kernel team about something called "software reuse" and "decoupling."
>Still, ReiserFS could continue without Hans, right? There's no such thing as software without a developer.
Open source isn't some magical formula that makes software grow itself. There have to be people knowledgeable enough, and interested enough to work on it. Also, if the work is involved enough, there generally has to be someone paying them to work on it. Even in open source, core developers on big projects are usually paid for their work either directly or indirectly. It looks like the means of paying for ReiserFS development is gone.
Also, the high degree of quality in ReiserFS was due to... Reiser and his employees. A new crop of developers is a new roll of the dice, and there's no guarantee you will get software that is nearly as good.
Every time I read about this case, it sounded like the sort of thing that couldn't possibly end in a conviction.
They found no weapon, no body, no witnesses. In fact, last time I read about this case, there was no real solid evidence whatsoever. How could they possibly reach a conviction? Was new evidence introduced that I haven't heard of? Did his lawyers just suck?
I hope he appeals this, from all the news articles I read, it sounds like his case was the *definition* of "reasonable doubt."
I'd like to point out that there are already two implementations of java on javascript aside from the one in the article:
1. Google Web Toolkit compiles java to javascript. 2. XML11 project actually implements the JVM in javascript. That's right, their script *interprets* and *executes* java bytecode. They let you do things like run existing AWT applications unmodified, by emulating AWT in HTML. Of course, this was pretty slow the last time I looked.
I'm enthusiastic about the potential of this device. Here's some use cases:
1. Sending text messages to your girlfriend while dodging sniper fire. 2. Checking slashdot while reloading. 3. Twittering "Hey, I just shot some guy. PK PK."
>Hunger is by far a distribution problem, >not a food production problem.
This has only been true for a limited span of time, since the green revolution, which started in 1943, massively increased farming output, which had been only increasing at a steady linear pace in the past. Since then, populations have surged to meet the new level of food supply.
Historically (before 1943) population was kept in check by a limited supply of food. A large number of people would be born to each generation, more than could possibly be fed, and a certain percentage at the bottom of the socioeconomic ladder would just starve to death.
Read Thomas Malthus' "The Principle of Population" to see how economics, agriculture, and famine worked in the 18th century.
In recent times, the massive increase in food raised the cap on the population, and birth rates skyrocketed. Between 1900 and now, the population moved from 1.6 billion to 6 billion, multiplying by 3.75 in only 100 years. In comparison, over the prior 100 years, the population only multiplied by 1.6.
Unfortunately, instead of food production growth, in many places food output is actually shrinking as populations displace prime farm land. The geometic growth rates of food production are nearing and end.
The question now is, do we go back to being the way we used to be, with society relatively impoverished and the bottom parts of society constantly starving to death?
Some people are hopeful that birth control and education will put a check on population growth, independent of food supply. Indeed, the united states, japan, and europe are all below replacement rates while at the same time being awash in food. Unfortunately, the rest of the world has yet to follow suit, so who knows?
explaining why GC has to halt the world periodically, which causes stutter in applications with user faces UI's like video.
>You can write an OS in a garbage collected language. >(The Lisp machines did it)
The kernel would be a really bad place to put a garbage collector, because that would mean the entire system would halt on a compaction.
I don't know for a fact, but I doubt the LISP machines used garbage collection in the kernel. Even if the kernel source was some variation of LISP, it doesn't mean it was garbage collected LISP.
It would be incredibly stupid to try to write drivers with garbage collection. I'm not saying it is impossible, I'm just saying that the performance would be unusably bad even on modern hardware.
>You don't know jack shit about garbage collection. >You think malloc() and free() have zero cost?
Well, first of all, in C and C++ you often *don't have* to use malloc and free if you don't want to. Apache, for instance, uses memory pools. Hell, you can even plug a garbage collector into C and C++ if you want, like Boehm gc. The point of C++ is that it can do *anything* and doesn't force you into a particular model.
Also, before you swear at me, please look up the term "amortized complexity" and compare it to worst case complexity.
Many algorithms have low amortized, or average complexity, but extremely high worst case. Garbage collection has decent is an algorithm that has good amortized time an atrocious worst case complexity for memory allocation.
Whereas in comparison, malloc and free are not cheap, but their worst case complexity can be made reasonable.
What is the worst case for a java garbage collector? When you try to allocate memeory on the heap, but can't because the heap is full. Then, java has to stop the world (freeze *all* threads) and free whatever memory it can on the heap. This operations time is proportional to the *total memory* your program uses, so it can easily make your app hang for a few seconds. Obviously, if you are processing video this will cause unavoidable stuttering.
GC implementations use a number of techniques to reduce the number of compactions, but there's no way to avoid them completely, which is part of why java desktop applications always have a sluggish UI experience, even if they work fine for things like web sites or batch jobs where the subjective interactivity of the process doesn't really come into play.
C++ is going "out of fashion"? What industry is this again? Who cares what is "in fashion."
C and C++ let you do things that you *cannot* no matter how hard you try in other languages. Java and Python have their place, and that place has expanded over time, but there are areas that they cannot tread.
No matter how much you improve the JVM, you will *never* be able to write a video codec in java (when the jvm compacts, it has to stop the world, and thus the video too). Nor will you be able to write kernel modules. Nor will you be able to write a large commercial video game. The list goes on and on...
Garbage collection is *great* when you can use it. However, garbage collection has some fundamental performance and memory usage characteristics that aren't going to go away, and thus GC languages will always be precluded from certain domains.
Scheme, ruby, python, lua, and lots of other languages come and go, but C++ is a powerful tool that is here to stay.
note that even within the open source community there are factions opposed to the GPL.
The GPL is not the only open source license, and many people do not think it is a very good one. Licenses like BSD, MIT, BOOST, APACHE, and so on are seen as more free by some (including myself), and more compatible with other licenses.
The GPL is intentionally viral. Some people who live purely in the open source world see this as a good thing, obviously. However, myself and most other developers I know live in a mixed source world, and don't want the license from one library imposing it's license on your entire application.
GPL and LGPL are fine for some things; however, they have many surprising and not particularly well known clauses and ambiguities that make them poison when shipping commercial software.
>Hit one key on initial boot. >Insert driver floppy. Type A:\. Press enter. >Hardly a nightmare.
I have an a Lenovo X61, which has no floppy. Indeed, I can't recall seeing any computer on sale in the past few years that has a floppy drive.
>if you're not installing floppy drives >in computers it's your own fault.
Actually, I think it's Microsoft's fault for requiring a floppy for installing drivers, when computers don't even come with floppy drives anymore.
My understanding is that they *do not* support USB thumb drives for XP. Most people are forced to slipstream drivers if they can't install with the drivers on the install CD.
># Cheap enough that you can afford to >give away the media as part of >giving away the content;
That's what CD's are for. CD's are easier to come by, and probably cheaper than floppies. You can't rewrite (most) Cd's, but it doesn't matter since they are cheap enough to be disposable.
># Random access.
Floppy drives are *not* random access. Random access means there is no seek, you just give an address and get the value stored at that address in some fixed amount of time. Floppies have spin a magnetic disk to seek to a location just like harddrives and CD's. That's what that annoying *grinding* sound is.
Slashdot Mirror
"As anyone who's been using Linux for several years (or even a few) for everyday tasks knows, "ready for the desktop" is in the eye of the beholder."
For instance, if you're a zealot who isn't willing to admit that Linux has flaws, Linux *is* ready for the desktop.
On the other hand if you are a rational human being, you might think of Linux as a great server operating system, that some people, painfully, force themselves to use on the desktop for ideological reasons. Kind of like the computing equivalent of a hair shirt.
See? It's all a matter of perspective.
Another matter of perspective. If you tend to agree with me on this issue, you might mod this article "funny," but if you disagree, you might mod it "flaimbait."
"In other words, "shared source" is not open source, and shouldn't be confused with it."
which is why it's called... shared source and not open source.
OMG microsoft has set us up the bombs! Oh noes!
God... nerds.
and do handle merging much better than svn (which hardly manages merging at all) however, a lot of companies use perforce, because perforce does handle branching and merging well, whereas it also allows for a centrol source repository.
Mercurial and get are mostly built around the model of sending your patches in via email. They make some noise about how you *could* build a central repository, but it's not like that out of the box, and things like permissions often aren't there yet.
Basically, the development model that the open source guys use to do their work isn't the model that guys at big corporations need to use, and distributed VC tools are so heavily geared towards the open source model (in the case of git, not even supporting windows that well) that they can't really be drop in replacements for tools like perforce.
This might make a good cheezy video game, but I can't imagine sitting through 2 hours of youtube quality acting. Check out the preview for the other film these guys created "star wreck." It looks like a tediously bad fan comedy.
The trailer might look kind of impressive, but from the website:
"At the moment, the team is finalizing the script"
In other words, the trailer has no relation to the final product in quality... and when they actually have to find live action "actors" to do the script... well...
I'm not too optimistic.
whether some of these fiascos are Balmer's fault. Particularly, what happened with vista. It's very plausible that Vista died of feature bloat because Balmer didn't pay enough attention to it in the beginning; however, it's hard to tell from the outside who was really responsible.
Either way though, It's clear that some kind of shakeup needs to occur for Microsoft to continue to compete. I just don't see Microsoft being able to expand into new markets using the clumsy "throw money at the problem" approach that post-gates Microsoft has used. Money's an important tool for a company the size of Microsoft, but it can't cover up underlying problems, like a project that suffering feature creep, or a corporate culture that suppresses bottom up innovation.
Bill Gates seemed to run a much tighter ship overall, with a supposedly fairly "hands on" management style. However, it is true that Microsoft was a much smaller company under Gate's tenure, and I'm not sure he would be the man to put back in charge of the new Microsoft.
I know there are 32 bit PPC macs, but I thought that all X86 macs were 64 bit, since they started with the core 2 duo systems.
What models of macs came with a 32 bit OS? Or are they simply referring to mac os 10.4, which I believe had some 32 bit support, but didn't have all API's (like cocoa) available in 64 bit versions.
The last starbucks I was at had ATAT wifi for free... or at least it let me log on.
in a startup might be relatively young, just out of college and willing to take risks. However, as soon as that startup gets enough funding to hire some more employees and pay them a salary, they aren't going to go after other college grads with limited experience.
Funded startups typically go after top tier developers with lots of experience to get their product off the ground. They need these guys because the challenges a startup faces are huge compared to the number of people and infrastructure available. Also, they can actually get top tier people by
1. offering them top dollar.
2. offering them lots of pre-ipo stock
3. last (but not least) giving them something really challenging to work on.
Only the first few employees can be expected to work for free, so these more experience guys that come later actually aren't taking that much of a risk, as they still get their higher than usual salary, plus stock, which is just gravy if it pays off.
>I don't think New York has the authority to
>do this. But I sure would like to see the
>supreme Court act.
That seems iffy to me. Most of the goods I purchase at a store aren't from in state, but I still pay sales tax on them.
Just because Amazon's store is online, doesn't mean that they shouldn't have to pay the same sales tax everyone else does.
I live in Washington state, in Seattle, the same city where Amazon is based, and I already pay state sales taxes on Amazon goods.
Maybe there is some legal loophole that prevents states from raising taxes in these situation, but it seems like a loophole that should be closed. This isn't a interstate tariff we're talking about here.
Canonical was founded by the billionaire Mark Shuttleworth.
http://en.wikipedia.org/wiki/Mark_Shuttleworth
He's basically putting up all the money for the operation on the vague hope that it will pay off someday. They really don't have a business model, just a really generous investor/CEO.
So... it's basically a charity based operating system.
Which raise the point, why is this douchebag
http://www.interopnews.com/news/is-ubuntu-selling-out-or-growing-up.html
writing an article about how the company is "selling out" by making some very small moves to make money off of an operating system they spend large amounts of money on, and give away for free?
It kind of pisses me off that random internet idiots who don't make software for a living call anyone who tries to a "sellout."
The article mentions that they are trying to recoup a small amount of the money they are dumping into Canonical by selling some proprietary software.
So what? I'm sick and tired of internet morons tearing apart people that actually have to work for a living. It's not enough that they give away most of their software for free and under an open source license, but if they charge for *anything*, if you develop one line of proprietary code and sell it to make a buck, some random jerkoff will mouth off at you about how "software wants to be free," and you're "oppressing" them with your price tag and your non-gpl license.
Free software isn't a business model. None of the distros that don't make you pay money *per install* make any money. Canonical loses money, Suse loses money. The only people who make money making operating systems do so by selling some proprietary code, or (as with red hat) devising schemes to make people pay money for shrink wrapped copies of open source code. Ubuntu has by far taken the least obnoxious approach, i.e. giving away most of their software, and letting you use their repository for free updates (which others don't do), but developing some proprietary stuff they let you buy separately.
you're facing a difficult ethical dilemma, so clearly you made the right decision by asking for help from a bunch of anonymous internet forum morons.
In addition to conspiracy theories about how Microsoft and the RIAA conspired to kill your friend's brother, you can expect to recieve such insightful gems on your crisis as, "in soviet Russia, myspace suicides you."
why is this news? Professionals might use dreamweaver or equivalent for a little quick prototyping, but nothing spat out by a wysiwyg is going on the web.
The days of everyone's websites being filled with "authored by frontpage" comments are long over.
How can he be debunked? His primary contribution is the observation that unconstrained population grows exponentially. He then observed, that for his time the primary constraint on population growth was food supply.
There's really not much to debunk there.
There are other points that he makes that you can attack, but those are not points I relied on.
that "35 year high" for fertility is only 2.1 for each woman. That is extremely low by global standards. It is *only* a high for industrialized countries.
It's like saying that Anchorage is warm *compared to the rest of alaska* and concluding from that, that anchorage must be a tropical tourist resort.
One problem, is that even if there are "open source" drivers available, for what platform? Mac OSX? Solaris? BSD? Linux? Windows?
Or maybe they are drivers for an older kernel version? New kernels don't link against old drivers, and new versions of GCC don't necessarily *compile* old drivers.
And then of course, there's installation of drivers. Common stuff like video drivers tend to be included in distros... but what about things like fingerprint readers, USB printers, etc? Installing a driver, doesn't just mean running an installer. There's also the extra step of compiling for your exact kernel version. Which means you must have GCC installed, on a desktop machine that you may not want to do any development on at all.
Does this seem like a really bad system to anyone else? Why do we *ever* need to write the same driver more than once? Why not just have a standard interface for talking to X86 drivers for things like audio, video, printers, harddrives etc?
Then, one driver for Linux, OSX, Windows, etc can be passed around and we don't have to continually reinvent the wheel. Then a linux installer can get extra drives by pulling them out of a linux install. Then we can write whole new kernels without having to write tons of custom drivers for it.
Seriously, someone needs to the Linux kernel team about something called "software reuse" and "decoupling."
>Still, ReiserFS could continue without Hans, right?
There's no such thing as software without a developer.
Open source isn't some magical formula that makes software grow itself. There have to be people knowledgeable enough, and interested enough to work on it. Also, if the work is involved enough, there generally has to be someone paying them to work on it. Even in open source, core developers on big projects are usually paid for their work either directly or indirectly. It looks like the means of paying for ReiserFS development is gone.
Also, the high degree of quality in ReiserFS was due to... Reiser and his employees. A new crop of developers is a new roll of the dice, and there's no guarantee you will get software that is nearly as good.
Every time I read about this case, it sounded like the sort of thing that couldn't possibly end in a conviction.
They found no weapon, no body, no witnesses. In fact, last time I read about this case, there was no real solid evidence whatsoever. How could they possibly reach a conviction? Was new evidence introduced that I haven't heard of? Did his lawyers just suck?
I hope he appeals this, from all the news articles I read, it sounds like his case was the *definition* of "reasonable doubt."
I'd like to point out that there are already two implementations of java on javascript aside from the one in the article:
1. Google Web Toolkit compiles java to javascript.
2. XML11 project actually implements the JVM in javascript. That's right, their script *interprets* and *executes* java bytecode. They let you do things like run existing AWT applications unmodified, by emulating AWT in HTML. Of course, this was pretty slow the last time I looked.
I'm enthusiastic about the potential of this device. Here's some use cases:
1. Sending text messages to your girlfriend while dodging sniper fire.
2. Checking slashdot while reloading.
3. Twittering "Hey, I just shot some guy. PK PK."
>Hunger is by far a distribution problem,
>not a food production problem.
This has only been true for a limited span of time, since the green revolution, which started in 1943, massively increased farming output, which had been only increasing at a steady linear pace in the past. Since then, populations have surged to meet the new level of food supply.
Historically (before 1943) population was kept in check by a limited supply of food. A large number of people would be born to each generation, more than could possibly be fed, and a certain percentage at the bottom of the socioeconomic ladder would just starve to death.
Read Thomas Malthus' "The Principle of Population" to see how economics, agriculture, and famine worked in the 18th century.
In recent times, the massive increase in food raised the cap on the population, and birth rates skyrocketed. Between 1900 and now, the population moved from 1.6 billion to 6 billion, multiplying by 3.75 in only 100 years. In comparison, over the prior 100 years, the population only multiplied by 1.6.
Unfortunately, instead of food production growth, in many places food output is actually shrinking as populations displace prime farm land. The geometic growth rates of food production are nearing and end.
The question now is, do we go back to being the way we used to be, with society relatively impoverished and the bottom parts of society constantly starving to death?
Some people are hopeful that birth control and education will put a check on population growth, independent of food supply. Indeed, the united states, japan, and europe are all below replacement rates while at the same time being awash in food. Unfortunately, the rest of the world has yet to follow suit, so who knows?
explaining why GC has to halt the world periodically, which causes stutter in applications with user faces UI's like video.
>You can write an OS in a garbage collected language.
>(The Lisp machines did it)
The kernel would be a really bad place to put a garbage collector, because that would mean the entire system would halt on a compaction.
I don't know for a fact, but I doubt the LISP machines used garbage collection in the kernel. Even if the kernel source was some variation of LISP, it doesn't mean it was garbage collected LISP.
It would be incredibly stupid to try to write drivers with garbage collection. I'm not saying it is impossible, I'm just saying that the performance would be unusably bad even on modern hardware.
>You don't know jack shit about garbage collection.
>You think malloc() and free() have zero cost?
Well, first of all, in C and C++ you often *don't have* to use malloc and free if you don't want to. Apache, for instance, uses memory pools. Hell, you can even plug a garbage collector into C and C++ if you want, like Boehm gc. The point of C++ is that it can do *anything* and doesn't force you into a particular model.
Also, before you swear at me, please look up the term "amortized complexity" and compare it to worst case complexity.
Many algorithms have low amortized, or average complexity, but extremely high worst case. Garbage collection has decent is an algorithm that has good amortized time an atrocious worst case complexity for memory allocation.
Whereas in comparison, malloc and free are not cheap, but their worst case complexity can be made reasonable.
What is the worst case for a java garbage collector? When you try to allocate memeory on the heap, but can't because the heap is full. Then, java has to stop the world (freeze *all* threads) and free whatever memory it can on the heap. This operations time is proportional to the *total memory* your program uses, so it can easily make your app hang for a few seconds. Obviously, if you are processing video this will cause unavoidable stuttering.
GC implementations use a number of techniques to reduce the number of compactions, but there's no way to avoid them completely, which is part of why java desktop applications always have a sluggish UI experience, even if they work fine for things like web sites or batch jobs where the subjective interactivity of the process doesn't really come into play.
C++ is going "out of fashion"? What industry is this again? Who cares what is "in fashion."
C and C++ let you do things that you *cannot* no matter how hard you try in other languages. Java and Python have their place, and that place has expanded over time, but there are areas that they cannot tread.
No matter how much you improve the JVM, you will *never* be able to write a video codec in java (when the jvm compacts, it has to stop the world, and thus the video too). Nor will you be able to write kernel modules. Nor will you be able to write a large commercial video game. The list goes on and on...
Garbage collection is *great* when you can use it. However, garbage collection has some fundamental performance and memory usage characteristics that aren't going to go away, and thus GC languages will always be precluded from certain domains.
Scheme, ruby, python, lua, and lots of other languages come and go, but C++ is a powerful tool that is here to stay.
note that even within the open source community there are factions opposed to the GPL.
The GPL is not the only open source license, and many people do not think it is a very good one. Licenses like BSD, MIT, BOOST, APACHE, and so on are seen as more free by some (including myself), and more compatible with other licenses.
The GPL is intentionally viral. Some people who live purely in the open source world see this as a good thing, obviously. However, myself and most other developers I know live in a mixed source world, and don't want the license from one library imposing it's license on your entire application.
GPL and LGPL are fine for some things; however, they have many surprising and not particularly well known clauses and ambiguities that make them poison when shipping commercial software.
>Hit one key on initial boot.
>Insert driver floppy. Type A:\. Press enter.
>Hardly a nightmare.
I have an a Lenovo X61, which has no floppy. Indeed, I can't recall seeing any computer on sale in the past few years that has a floppy drive.
>if you're not installing floppy drives
>in computers it's your own fault.
Actually, I think it's Microsoft's fault for requiring a floppy for installing drivers, when computers don't even come with floppy drives anymore.
My understanding is that they *do not* support USB thumb drives for XP. Most people are forced to slipstream drivers if they can't install with the drivers on the install CD.
># Cheap enough that you can afford to
>give away the media as part of
>giving away the content;
That's what CD's are for. CD's are easier to come by, and probably cheaper than floppies. You can't rewrite (most) Cd's, but it doesn't matter since they are cheap enough to be disposable.
># Random access.
Floppy drives are *not* random access. Random access means there is no seek, you just give an address and get the value stored at that address in some fixed amount of time. Floppies have spin a magnetic disk to seek to a location just like harddrives and CD's. That's what that annoying *grinding* sound is.