Ironically, despite all your insistence that we should question authority, you yourself failed to question whether 1+1=2 is truly an absolute. It isn't.
In the finite field of two elements, the only elements are 0 and 1. There is no element named 2, and for this reason it is simply not possible for 1+1 to equal 2 (in fact, 1+1 equals 0 in this context). More generally, 1+1 equals 0 in any ring of characteristic 2.
This example illustrates the reason that I think you are wrong and that the Wikipedia founder who wrote the articles being referenced in the story is right. You claim that
If you have critical thinking skills as someone else mentioned, you would be able to filter what is good information from what is bad. Wikipedia is just a starting point.
which is literally true, but misleading for what it leaves out. What it leaves out is that you need unrealistically strong critical thinking skills in order for you alone to match up with the entirety of human knowledge that has been accumulated in the generations before you. I believe that no one person can ever be truly an expert in even a single field of human knowledge. How then can you expect anyone to accurately filter good information from bad, across the entire spectrum of human knowledge?
It is easy to say that in principle one could distinguish good information from bad by thought alone. However, as you yourself just demonstrated, doing it in practice is much harder than it seems in theory. The truth is that 1+1 does not always equal 2, but very very very few people would ever even initiate the process of questioning such a fact. The number of people who can recognize inaccuracies in a statement such as "you can [only] get one answer. 1+1=2." is very very small. It is simply not true that most people can filter good information from bad in specialized subjects.
Although I said I would stop this thread, I will try to answer your questions as helpfully as I can.
How exactly is it rational to allow distribution of a GPLed program except by the author of a proprietary OS with that OS?
Your question's premise is false. A correct rephrasing would be:
How exactly is it rational to allow distribution of a GPLed program except with a proprietary OS?
The GPL actually prohibits any and all distribution of a GPLd program together with a linked proprietary OS. The author of the proprietary OS is not the only person enjoined from such actions. Your question continues to perpetuate the falsity that the author of the proprietary OS is the only "particular entity" prevented from bundling. This presumption is false. Anybody and everybody is prohibited from bundling a GPL executable together with a linked proprietary library.
A Windows GPL executable can be legally distributed, but it must be distributed separately from Windows. This holds for all entities in the world, not just one particular entity.
How can you argue that the GPL isn't viral when in this case it is DESIGNED to be viral
Straw man. I have never at any point in this entire thread argued that the GPL isn't viral.
The FSF absolutely does not agree. All of the examples that you give (compiler and kernel, editor and shell) do not involve linking to libraries. In fact, none of the four objects you mention (compiler, kernel, editor, shell) is a library at all. The FSF position is that a "shared address space almost surely means combining them into one program" and all direct linking to system libraries always involves sharing address space.
I'm getting tired of repeating the same points over and over again so I will summarize my points for the last time. Feel free to get in the last word if you must.
As an exception to the previous exception, you are not allowed to take advantage of the previous exception if you are distributing the executable accompanied by the library itself. (Source:...unless that component itself accompanies the executable.)
The FSF's web site makes the exact same error that you make -- namely, they fail to mention that there exists one, and exactly one, exception. This error should probably be corrected. However, in their defense, the FSF FAQ is written in a different context than the debate that we are having. It is not reasonable to expect a short FAQ answer to cover all the contingencies in detail.
I don't know how many times I have to repeat this point. The GPL itself contains the word "exception". The word "exception" is pretty clear in meaning. If the FSF FAQ neglects to mention this exception, then the FSF FAQ is wrong.
Now, under my (and, by the way, FSF's) interpretation, most programs do NOT link so closely to system libraries as to become a single program with that library.
The FSF does not agree with your interpretation. To quote the very same FSF FAQ that you quoted:
If the modules are included in the same executable file, they are definitely combined in one program. If modules are designed to run linked together in a shared address space, that almost surely means combining them into one program.
Either your interpretation is wrong, or the GPL is fucking ridiculous. Sorry, but that's just the way it is.
A lot of people think the GPL is ridiculous, but very few people think that the GPL completely prohibits distribution of any Windows executables in all circumstances.
If you're going to argue that legally, linking a GPLed program into Windows makes them more than mere aggregation, then the program's distributor CANNOT legally distribute it under the GPL.
Your argument would be true, except for the fact that the text of the GPL itself contains a "special exception" which specifically allows the exact thing that you claim is not allowed. This special exception only applies when the library is an OS library that does not accompany the executable.
I again encourage you to search for the phrase "special exception" within the text of the GPL. The phrase only appears once and it leads you to the exact same text that I have been quoting ever since the start of this thread.
At *most*, MS might be required to release the source to those libraries; as much of that source is already released as part of the MS compilers, I can't imagine they would have an issue doing that.
The issue for Microsoft is not that they would have to release the source to those libraries, but that they would have to release the source of those libraries under the GPL.
Under your interpretation, exactly how would *any* GPL Windows executable be legal to run? If it can be installed by the end user, why can't MS distribute it?
The GPL (unlike most EULAs) does not require that you accept the license in order to run the program. The license terms apply only to distributors.
For that matter, under your interpretation, how can *any* GPLed Windows app legally exist, excepting of course the case of the non-dependent app?
According to the part of the GPL that I quoted, you do not have to provide operating system source code unless you distribute the operating system itself together with the GPL program.
The vast majority of regular people ("party A" in your example) do not distribute Microsoft Windows as part of a bundle together with their GPL apps. Indeed, unless you are Microsoft or have a license from Microsoft to do so, distributing Windows is completely illegal.
If the author of a GPLed program links to a non-GPL system library, then distributes it, aren't they violating the GPL themselves?
First of all, the owner of a program is always the one entity in the world that can legally violate their own licensing terms. But in this case it's not even a violation unless the author actually distributes both the GPL program and the system library as a combined bundle.
This effectively requires all third party distributors to link the executable dynamically, since static linking forces you to bundle the EXE and the library. You might ask, why is dynamic linking given special treatment here and not in my previous post? It's because the "special exception" in the GPL (search for it, it only appears once) applies for unbundled operating system components, but not for bundled operating system components.
If GPLed Windows apps can legally exist, then I can't see how MS distributing them becomes any different than the author distributing them - that's one of the nice things about the GPL.
Microsoft is allowed to distribute GPL applications. However, nobody (not Microsoft, not you, not me, not party A, and not party B) is allowed to bundle a GPL application together with Windows and distribute the combined bundle, or else they violate the GPL.
There are in fact ways for Microsoft to distribute GPL software without relicensing Windows. I already mentioned two of the ways: they can distribute the GPL program in source form, or they can distribute it on a separate CD as a separate product. The GPL does not discriminate against Microsoft; what it discriminates against is bundling the app with a proprietary OS, no matter who is doing it.
Re:Microsoft can't legally bundle GPL software
on
Linux Users Are Spoiled
·
· Score: 2, Informative
if some of the source code replicates a component normally distributed within the operating system, you're exempted from being required to provide the source code to that component
No. The GPL does not mention replication. In fact, I just grepped the text of the license to make sure. The word replicate is your own invention and appears nowhere within the GPL.
The GPL says if some of your required source code IS an operating system component, you are exempted from having to distribute that source code unless you include that component.
distributing a GPL Windows executable with the system is generally going to count as mere aggregation...
This is generally not true, at least according to the author of the GPL.
The most clear cut case is that of a statically linked executable which includes calls to system libraries. In this case, the static executable itself includes an entire copy of the library in question -- and any reasonable person (more importantly, any reasonable courtroom) would agree that including a copy of the library within the executable itself together with calls to that library goes beyond mere aggregation onto common media.
The grey area is the case of dynamic linking with system libraries. Richard Stallman argues that this action is equivalent to static linking and therefore makes the combination of WindowsExecutable + SystemLibrary a derived work of the WindowsExecutable, for which distribution requires accompanying source code of all components and in particular the SystemLibrary component. I'm not quite sure I entirely agree, but I do recognize that it is debatable. In any case, interpretations of this part of the GPL have never been tested in court and I think it would be quite irresponsible of Microsoft to volunteer themselves as a test case.
The third case, of an executable that includes no dependence on any system library, would be perfectly fine for Microsoft to distribute, but few applications can build without system libraries.
Finally, I would like to note that the major proprietary UNIX vendors seem to agree with my interpretation of the GPL, since none of them include any GNU utilities with their operating systems even though the GNU utilities are vastly superior to their own. For example, back when their business was based on actual products, SCO's Skunkware CD which consisted mainly of GPL software for SCO Unix was sold as a separate disc.
Microsoft can't legally bundle GPL software
on
Linux Users Are Spoiled
·
· Score: 2, Informative
I don't see why Microsoft is so constrained about the software they can bundle. They would be perfectly within their rights to install Mozilla, Open Office, AbiWord, gcc and emacs...
This statement is actually extremely false. Now, there's a lot of FUD making the rounds about the so-called "viral" nature of the GPL, but what I'm about to say is fact not FUD. Microsoft would have to GPL all of Windows in order to bundle a GPL program together with Windows.
The relevant section of the GPL is Section 3, which states in part (edited for space but with no significant change in meaning):
You may copy and distribute the Program... in object code or executable form... provided that you also [provide the complete source code]...
For an executable work, complete source
code means all the source code for all modules it contains, plus any
associated interface definition files, plus the scripts used to
control compilation and installation of the executable. However, as a
special exception, the source code distributed need not include
anything that is normally distributed... with the...
operating system on which the executable runs, unless that component
itself accompanies the executable.
In other words, when distributing a GPL windows executable, you are not required to distribute operating system source code unless you are also bundling the executable with the operating system components. Unfortunately for Microsoft (and other proprietary OS vendors such as Sun, HP, and IBM), bundling is exactly what we are talking about here.
Microsoft can bundle GPL source code with Windows, but they aren't allowed to bundle GPL executables.
I generally prefer to use Linux over Windows when I have a choice in the matter (if you don't believe me, refer to my home page, or to the fact that I wrote this), but toolset and shell is not any big advantage for Linux anymore.
You can get a good toolset and shell on windows by installing Cygwin. Yes, it does have the drawback of being a big download, but at least it's only one download instead of several hunt-and-peck sessions of downloading. Cygwin provides all of the standard unix tools you mentioned: nice shells, grep, awk, sed, tar, and yes even perl and XFree86. The programs mostly behave the way you expect, because they're compiled from the same source code as the linux versions. You will find some bugs in Cygwin but none so bad as to cause data loss or hinder productivity.
Cygwin isn't as good as the command line environment in Linux (cause it's slow, has bugs, and requires third-party download), but it goes a long way towards filling the gap.
Unfortunately I can't recommend any Windows program to fill the role of apt-get....
I'd like to delete anything with a score > 15, simply store anything with a score > 5, and send an auto-reply for scores between 5 and 10 indicating that the message was marked as spam and I'll probably never look at it.
I can't speak for auto-replies, but you can do the sorting part client-side. The key is that spamassassin adds a line like "X-Spam-Level: *****" where the number of *'s is the score of the email. Almost any email client can filter mail to different folders based on headers. The unary representation of the spam score ensures that even a primitive filter can work.
For example, one popular client is Microsoft Outlook, and there are several web pages in google (such as this one) that explain how to reroute mail to specific folders depending on the spamassassin score.
You're right of course that both windows and macintosh support virtual desktops via add-on utilities. But the support provided by such utilities is inferior for two reasons. First of all, the very fact that you have to go out of your way to add them in the first place is a drawback. Second, the third party utilities are just not as functional as even the first generation implementations of virtual desktops from GNOME 1.x and KDE 1.x.
For example, Virtuawin does not provide you with a pager that shows window contents within the pager, something that GNOME and KDE have provided for years. Also, on every single Windows virtual desktop manager that I have tried, the desktop switching suffers periods of lagginess where it takes seconds or even minutes to switch desktops. This problem is especially pronounced when more than 20 or so windows are open at once.
I have nothing against the concept of virtual desktops on Windows, but all implementations I have seen so far are substandard.
I can't believe that no one has mentioned virtual desktops yet. An efficient virtual desktop implementation is the single most important multitasking enabling feature in the world today. Neither the Windows taskbar nor the Mac Expose desktop can compare to virtual desktops in utility.
Virtual desktops, for those who don't know, are multiple desktops which are all active on your computer at the same time. You switch between desktops by clicking a pager or pressing a key (e.g. on my linux desktop I use F1-F4 to activate desktops 1-4).
The advantage of virtual desktops is that they let you group programs and switch between them consistently and rapidly. For example, I always put ssh sessions on desktop 1, web pages on desktop 2, mail windows on desktop 3, and programming IDE on desktop 4. Each group of programs is always in the same place every day, and I can switch to whatever I want very quickly. Compare this to the Windows taskbar, where the taskbar icons are never in consistent locations and you have to hunt and peck for the right taskbar icon literally every single time you switch applications.
Even the Mac Expose desktop is less efficient than the simpler alternative of virtual desktops, since it is very difficult under Expose to group applications together and to perform consistent, single-keystroke navigation of applications.
Windows is like one folder on one desk, and Mac is like shifting lots of papers around on one desk. Virtual desktops is like having several desks at hand and switching between them at the touch of a button. The last one is the only paradigm that I would consider truly designed for multi-tasking.
i know someone who installed a red hat box and within a couple days, it was rooted...because they did a standard install of red hat. it might have changed since then...
It definitely has changed since then. Redhat 6 was the last version where you could root a default install.
Every Redhat version since Redhat 7 has a default installation with zero open ports and a firewall blocking all ports below 1024. The Fedora incarnations of Redhat are even better--on those distros, a default installation includes a firewall blocking all the ports. It now takes active effort on the part of the administrator to make Redhat/Fedora vulnerable to rooting.
The default security issue is something that Redhat has had problems with in the past and learned from their mistakes.
Speakeasy allows residential customers to share their broadband connection through a home network that utilizes technology such as Wi-Fi. However,
if a Speakeasy member is collecting access fees from any individual accessing their Wi-Fi network, the member must be subscribed to the NetShare service as a NetShare Admin, and the individual must be subscribed as a NetShare Customer. (emphasis added)
In other words, the Terms of Service state that I can share service without subscribing to the program. I have the option of collecting fees and I must give Speakeasy a cut if I do so, but it is not required.
You get to share your Internet connection with neighbors to eventually get your higher prices reduced down to what I'm paying to begin with.
I'm sorry pal, but to me, that sucks.
There's a saying that you get what you pay for. Find me another ISP that gives out four static IPs, allows you to run servers in their TOS, never performs port blocking, and allows connection sharing, for less money, and I would seriously consider switching to it, because I'm not at all deeply attached to the price of my current service.
Your complaint seems to be that you can't get the good features along with the cheap price. If the only alternative was a $750/mo T1 line then I could agree, but when it comes down to a simple choice between competing ADSL providers then that's just the free market at work.
If I share my ADSL 1.5/384 connection with my neighbors, I'm violating terms of service, and could lose my (very important to me) Internet connection.
Your ADSL provider sucks. My provider not only allows WiFi sharing, but even encourages it.
That is not to say I actually do share my 1.5/384 ADSL connection, but I could if I wanted to. Moreover, speakeasy's ADSL prices are way below T1/T3 (although still above el-cheapo baby bell DSL prices).
for a portable player with limited battery life... why in the world would anyone choose to get 75% performance with a negligable increase in sound quality (from headphones)?
The sentence in the article about ogg's battery life is very misleading. Yes, it is true that "you get about 25% less battery life" on ogg vs. mp3. However this comparison is done at the same bitrate -- that is to say, 128 kbps ogg will only have 75% the battery life of 128 kbps mp3.
But, what the quote doesn't take into account is that nobody uses oggs and mp3s at the same bitrate. I for one find that ogg can match mp3 in sound quality at about 60% of the bitrate. When you use a smaller bitrate, battery life goes up, because your hard drive activity is less. My firsthand experience is that you can get 15 hrs of continuous ogg playback on the karma, if you use a lower bitrate like 64 or 72 kbps. Also, you will note that even if we hypothetically penalized this real-world measurement of 15 hours by a theoretical 25%, it would still be better battery life than an iPod.
As to your dismissal of headphone sound quality, there are a greatmanyheadphones that are good enough to tell the difference. Even without good headphones, 72 kbps mp3 is so bad that anyone who is running out of disk space on their portable can easily justify the switch to vorbis.
You never see the ad./. gets the money. Why do they get the money? B/c no one knows you never see the ad
Actually, privoxy doesn't work that way.
Privoxy is a web proxy, not a browser plugin. That means it slipstreams itself in between your browser and the server. When using privoxy, your actual web browser never actually directly requests anything from the web site itself. All of its requests go through privoxy, and (crucially) privoxy does not actually pass all of the incoming requests through to the remote server.
The result is that when you go to slashdot's home page and there is an ads.osdn.com banner at the top of the page, privoxy doesn't work by first downloading the ad from the server and then preventing you from seeing it. Instead it works by recognizing ads.osdn.com as an advertising site, and not even sending the HTTP GET request at all.
Now, it is true that privoxy has a second, independently functional ad-blocking mechanism that does rely on post-processing the ad after it is downloaded, but ads.osdn.com is well known enough that privoxy can (and does) already decide to eschew even the initial GET request based purely on the URL input.
On the other hand, the Karma beats the Neuros in many ways. You already mentioned playlist management; to this I would add gapless playback (does firmware 2 fix this for the Neuros? please tell me it does) and internationalization (the Neuros LCD display is not even physically capable of rendering Chinese characters).
I do want Neuros to succeed very badly -- I bought one of them the first day, and I like their software philosophy and their support of the USB mass storage standard. But when it comes down to the practical matter of playing music, I much prefer the Karma to the Neuros.
ctrl-L will highlight the current url *without* overwriting your other highlighted text. Then hit delete or backspace to clear the url bar, and middle-click to paste in the new url.
The quickest way to paste a new URL into mozilla on Linux (and, I assume, FreeBSD as well) is to middle click directly into the main browser window. As long as you don't middle click a hyperlink, the browser will automatically load the pasted URL.
The title of this article is inaccurate. A "one-time password" is not the same thing as a "one-time pad".
A "one-time password" means a password that is used once and discarded. This password is typically used only for authentication purposes. By contrast, a "one-time pad" is used for encryption purposes.
One-time pads are almost never useful for typical internet situations because they are very easy to misuse and very insecure when misused. They also don't solve any problem worth solving -- conventional encryption is already strong enough that the added security of a one-time pad has no value in typical internet situations.
One-time passwords, on the other hand, do potentially have some value, because the currently available password authentication systems are quite weak compared to the strength of the corresponding encryption systems.
XV does work, but only sometimes. For instance, if you have a web browser open with a bunch of tabs, it seems to affect XV for some unknown reason. Is your XV working all of the time?
XV does indeed work all of the time. I usually have 20 tabs open in my web browser all the time, whether or not I'm watching movies.
The only movie playing problem I ever noticed is that in fullscreen mode mplayer will jerk the playback window to the side when switching between chapters on one or two of my DVDs, but I don't think this is an XVideo problem, because the playback window still works without interruption. None of the other DVD programs (xine, ogle, etc.) have this problem.
I am of course using the opensource radeon drivers since ATI hasn't released 64-bit proprietary drivers. The drivers do not support direct rendering, but this doesn't matter to me since I don't use any 3D programs.
WEP is completly insecure, and can be broken really easily, its really not worth it. I think making sure you are not broadcasting your ID, and setting up MAC address filtering, is the way to go.
You're right that WEP is insecure, but relying on your alternative recommendations is even worse. Kismet on Linux can sniff out cloaked SSIDs and MAC addresses of any devices on the network as long as the wireless network in question is actively transmitting packets at the same time that Kismet is being used. I have personally verified this functionality firsthand while running Kismet against my own wireless network.
Why does this work? Because the SSID and the MAC address of the destination device is included in plaintext in every wireless data packet (with or without WEP). SSID cloaking doesn't hide SSIDs within the data packets, it only prevents the router from broadcasting the stream of non-data packets announcing its SSID that it would otherwise broadcast. Therefore SSID cloaking is really only effective if you are not transmitting any data, or if your attacker is using an inferior sniffing platform that doesn't have raw access to the SSID in the data packets.
MAC address filtering is ineffective for the same reason -- every data packet is required by the specification to include the destination MAC address, so as long as you are transmitting data, your MAC addresses (all of them, not just the AP's MAC address) are exposed. Once an attacker gets your MAC addresses it is a simple matter to spoof them.
Your recommendation to use higher level layers of security is a very good one, because there is simply no way at the present time to secure wireless networks even to a level that is comparable to what we normally expect out of a wired network.
Just as an alternative (not Slackware related) data point, I'm using X.org on Fedora 2 x86-64 with an ATI Mobility Radeon 9600 and XVideo overlay DOES work, although it's much slower than it was on Fedora 1 using ATI's proprietary drivers. Unfortunately ATI has not yet seen fit to release 64-bit linux drivers. That said, I'm pretty happy that XV works at all... on my last laptop, it took several months before XV support for that card made it into XFree86.
He meant The first 64-bit distro for PPC, that is not a toy environment
That's certainly not what he said:
We are the first linux distribution to offer a 64-bit top-to-bottom solution which is not a toy environment.
Notice that "linux distribution" is not restricted in scope by any qualifiers such as "for PPC."
Even if he meant something different than what he said, it is certainly fair game to point out inaccuracies in what was said.
the further you get away from the "standard" i386 the worse support gets. Look at Fedora Core 2 for AMD64 - mysql is 32bit...
Um, this statement is false. The mysql server and client are fully 64-bit... here's proof.
While we have dozens of distributions there is not a single 64bit Linux out there that is even close to being as full-featured as debian, fedora, redhat, mandrake,... on i386 are...
If you want something as full-featured as i386, then (aside from simply running i386) x86-64 is the best game in town, because it actually runs i386 binaries. For example, my copy of Mathematica for i386 Linux runs perfectly in Fedora 2 x86-64.
As an aside, x86-64 clearly contradicts the story summary's claim of being the "first linux distribution to offer a 64-bit top-to-bottom solution which is not a toy environment." Red Hat Enterprise 3 for AMD64 was released six months ago with a full 64-bit userspace environment, and I don't think anybody can seriously argue that RHEL3 is a "toy environment" compared to a beta gentoo-ppc64 release.
Slashdot Mirror
In the finite field of two elements, the only elements are 0 and 1. There is no element named 2, and for this reason it is simply not possible for 1+1 to equal 2 (in fact, 1+1 equals 0 in this context). More generally, 1+1 equals 0 in any ring of characteristic 2.
This example illustrates the reason that I think you are wrong and that the Wikipedia founder who wrote the articles being referenced in the story is right. You claim that
which is literally true, but misleading for what it leaves out. What it leaves out is that you need unrealistically strong critical thinking skills in order for you alone to match up with the entirety of human knowledge that has been accumulated in the generations before you. I believe that no one person can ever be truly an expert in even a single field of human knowledge. How then can you expect anyone to accurately filter good information from bad, across the entire spectrum of human knowledge?It is easy to say that in principle one could distinguish good information from bad by thought alone. However, as you yourself just demonstrated, doing it in practice is much harder than it seems in theory. The truth is that 1+1 does not always equal 2, but very very very few people would ever even initiate the process of questioning such a fact. The number of people who can recognize inaccuracies in a statement such as "you can [only] get one answer. 1+1=2." is very very small. It is simply not true that most people can filter good information from bad in specialized subjects.
That, my friend, is why we need experts.
How exactly is it rational to allow distribution of a GPLed program except by the author of a proprietary OS with that OS?
Your question's premise is false. A correct rephrasing would be:
The GPL actually prohibits any and all distribution of a GPLd program together with a linked proprietary OS. The author of the proprietary OS is not the only person enjoined from such actions. Your question continues to perpetuate the falsity that the author of the proprietary OS is the only "particular entity" prevented from bundling. This presumption is false. Anybody and everybody is prohibited from bundling a GPL executable together with a linked proprietary library.A Windows GPL executable can be legally distributed, but it must be distributed separately from Windows. This holds for all entities in the world, not just one particular entity.
How can you argue that the GPL isn't viral when in this case it is DESIGNED to be viral
Straw man. I have never at any point in this entire thread argued that the GPL isn't viral.
I'm getting tired of repeating the same points over and over again so I will summarize my points for the last time. Feel free to get in the last word if you must.
I don't know how many times I have to repeat this point. The GPL itself contains the word "exception". The word "exception" is pretty clear in meaning. If the FSF FAQ neglects to mention this exception, then the FSF FAQ is wrong.
Now, under my (and, by the way, FSF's) interpretation, most programs do NOT link so closely to system libraries as to become a single program with that library.
The FSF does not agree with your interpretation. To quote the very same FSF FAQ that you quoted:
Either your interpretation is wrong, or the GPL is fucking ridiculous. Sorry, but that's just the way it is.
A lot of people think the GPL is ridiculous, but very few people think that the GPL completely prohibits distribution of any Windows executables in all circumstances.
Your argument would be true, except for the fact that the text of the GPL itself contains a "special exception" which specifically allows the exact thing that you claim is not allowed. This special exception only applies when the library is an OS library that does not accompany the executable.
I again encourage you to search for the phrase "special exception" within the text of the GPL. The phrase only appears once and it leads you to the exact same text that I have been quoting ever since the start of this thread.
At *most*, MS might be required to release the source to those libraries; as much of that source is already released as part of the MS compilers, I can't imagine they would have an issue doing that. The issue for Microsoft is not that they would have to release the source to those libraries, but that they would have to release the source of those libraries under the GPL.
The GPL (unlike most EULAs) does not require that you accept the license in order to run the program. The license terms apply only to distributors.
For that matter, under your interpretation, how can *any* GPLed Windows app legally exist, excepting of course the case of the non-dependent app?
According to the part of the GPL that I quoted, you do not have to provide operating system source code unless you distribute the operating system itself together with the GPL program.
The vast majority of regular people ("party A" in your example) do not distribute Microsoft Windows as part of a bundle together with their GPL apps. Indeed, unless you are Microsoft or have a license from Microsoft to do so, distributing Windows is completely illegal.
If the author of a GPLed program links to a non-GPL system library, then distributes it, aren't they violating the GPL themselves?
First of all, the owner of a program is always the one entity in the world that can legally violate their own licensing terms. But in this case it's not even a violation unless the author actually distributes both the GPL program and the system library as a combined bundle.
This effectively requires all third party distributors to link the executable dynamically, since static linking forces you to bundle the EXE and the library. You might ask, why is dynamic linking given special treatment here and not in my previous post? It's because the "special exception" in the GPL (search for it, it only appears once) applies for unbundled operating system components, but not for bundled operating system components.
If GPLed Windows apps can legally exist, then I can't see how MS distributing them becomes any different than the author distributing them - that's one of the nice things about the GPL.
Microsoft is allowed to distribute GPL applications. However, nobody (not Microsoft, not you, not me, not party A, and not party B) is allowed to bundle a GPL application together with Windows and distribute the combined bundle, or else they violate the GPL.
There are in fact ways for Microsoft to distribute GPL software without relicensing Windows. I already mentioned two of the ways: they can distribute the GPL program in source form, or they can distribute it on a separate CD as a separate product. The GPL does not discriminate against Microsoft; what it discriminates against is bundling the app with a proprietary OS, no matter who is doing it.
No. The GPL does not mention replication. In fact, I just grepped the text of the license to make sure. The word replicate is your own invention and appears nowhere within the GPL.
The GPL says if some of your required source code IS an operating system component, you are exempted from having to distribute that source code unless you include that component.
distributing a GPL Windows executable with the system is generally going to count as mere aggregation...
This is generally not true, at least according to the author of the GPL.
The most clear cut case is that of a statically linked executable which includes calls to system libraries. In this case, the static executable itself includes an entire copy of the library in question -- and any reasonable person (more importantly, any reasonable courtroom) would agree that including a copy of the library within the executable itself together with calls to that library goes beyond mere aggregation onto common media.
The grey area is the case of dynamic linking with system libraries. Richard Stallman argues that this action is equivalent to static linking and therefore makes the combination of WindowsExecutable + SystemLibrary a derived work of the WindowsExecutable, for which distribution requires accompanying source code of all components and in particular the SystemLibrary component. I'm not quite sure I entirely agree, but I do recognize that it is debatable. In any case, interpretations of this part of the GPL have never been tested in court and I think it would be quite irresponsible of Microsoft to volunteer themselves as a test case.
The third case, of an executable that includes no dependence on any system library, would be perfectly fine for Microsoft to distribute, but few applications can build without system libraries.
Finally, I would like to note that the major proprietary UNIX vendors seem to agree with my interpretation of the GPL, since none of them include any GNU utilities with their operating systems even though the GNU utilities are vastly superior to their own. For example, back when their business was based on actual products, SCO's Skunkware CD which consisted mainly of GPL software for SCO Unix was sold as a separate disc.
This statement is actually extremely false. Now, there's a lot of FUD making the rounds about the so-called "viral" nature of the GPL, but what I'm about to say is fact not FUD. Microsoft would have to GPL all of Windows in order to bundle a GPL program together with Windows.
The relevant section of the GPL is Section 3, which states in part (edited for space but with no significant change in meaning):
In other words, when distributing a GPL windows executable, you are not required to distribute operating system source code unless you are also bundling the executable with the operating system components. Unfortunately for Microsoft (and other proprietary OS vendors such as Sun, HP, and IBM), bundling is exactly what we are talking about here.
Microsoft can bundle GPL source code with Windows, but they aren't allowed to bundle GPL executables.
You can get a good toolset and shell on windows by installing Cygwin. Yes, it does have the drawback of being a big download, but at least it's only one download instead of several hunt-and-peck sessions of downloading. Cygwin provides all of the standard unix tools you mentioned: nice shells, grep, awk, sed, tar, and yes even perl and XFree86. The programs mostly behave the way you expect, because they're compiled from the same source code as the linux versions. You will find some bugs in Cygwin but none so bad as to cause data loss or hinder productivity.
Cygwin isn't as good as the command line environment in Linux (cause it's slow, has bugs, and requires third-party download), but it goes a long way towards filling the gap.
Unfortunately I can't recommend any Windows program to fill the role of apt-get....
I can't speak for auto-replies, but you can do the sorting part client-side. The key is that spamassassin adds a line like "X-Spam-Level: *****" where the number of *'s is the score of the email. Almost any email client can filter mail to different folders based on headers. The unary representation of the spam score ensures that even a primitive filter can work.
For example, one popular client is Microsoft Outlook, and there are several web pages in google (such as this one) that explain how to reroute mail to specific folders depending on the spamassassin score.
For example, Virtuawin does not provide you with a pager that shows window contents within the pager, something that GNOME and KDE have provided for years. Also, on every single Windows virtual desktop manager that I have tried, the desktop switching suffers periods of lagginess where it takes seconds or even minutes to switch desktops. This problem is especially pronounced when more than 20 or so windows are open at once.
I have nothing against the concept of virtual desktops on Windows, but all implementations I have seen so far are substandard.
Virtual desktops, for those who don't know, are multiple desktops which are all active on your computer at the same time. You switch between desktops by clicking a pager or pressing a key (e.g. on my linux desktop I use F1-F4 to activate desktops 1-4).
The advantage of virtual desktops is that they let you group programs and switch between them consistently and rapidly. For example, I always put ssh sessions on desktop 1, web pages on desktop 2, mail windows on desktop 3, and programming IDE on desktop 4. Each group of programs is always in the same place every day, and I can switch to whatever I want very quickly. Compare this to the Windows taskbar, where the taskbar icons are never in consistent locations and you have to hunt and peck for the right taskbar icon literally every single time you switch applications.
Even the Mac Expose desktop is less efficient than the simpler alternative of virtual desktops, since it is very difficult under Expose to group applications together and to perform consistent, single-keystroke navigation of applications.
Windows is like one folder on one desk, and Mac is like shifting lots of papers around on one desk. Virtual desktops is like having several desks at hand and switching between them at the touch of a button. The last one is the only paradigm that I would consider truly designed for multi-tasking.
It definitely has changed since then. Redhat 6 was the last version where you could root a default install.
Every Redhat version since Redhat 7 has a default installation with zero open ports and a firewall blocking all ports below 1024. The Fedora incarnations of Redhat are even better--on those distros, a default installation includes a firewall blocking all the ports. It now takes active effort on the part of the administrator to make Redhat/Fedora vulnerable to rooting.
The default security issue is something that Redhat has had problems with in the past and learned from their mistakes.
False, although you do have to read deeper than the front page to obtain the true terms and conditions under which sharing is allowed.
The Terms of Service say (and I quote):
In other words, the Terms of Service state that I can share service without subscribing to the program. I have the option of collecting fees and I must give Speakeasy a cut if I do so, but it is not required.
You get to share your Internet connection with neighbors to eventually get your higher prices reduced down to what I'm paying to begin with. I'm sorry pal, but to me, that sucks.
There's a saying that you get what you pay for. Find me another ISP that gives out four static IPs, allows you to run servers in their TOS, never performs port blocking, and allows connection sharing, for less money, and I would seriously consider switching to it, because I'm not at all deeply attached to the price of my current service.
Your complaint seems to be that you can't get the good features along with the cheap price. If the only alternative was a $750/mo T1 line then I could agree, but when it comes down to a simple choice between competing ADSL providers then that's just the free market at work.
Your ADSL provider sucks. My provider not only allows WiFi sharing, but even encourages it.
That is not to say I actually do share my 1.5/384 ADSL connection, but I could if I wanted to. Moreover, speakeasy's ADSL prices are way below T1/T3 (although still above el-cheapo baby bell DSL prices).
The sentence in the article about ogg's battery life is very misleading. Yes, it is true that "you get about 25% less battery life" on ogg vs. mp3. However this comparison is done at the same bitrate -- that is to say, 128 kbps ogg will only have 75% the battery life of 128 kbps mp3.
But, what the quote doesn't take into account is that nobody uses oggs and mp3s at the same bitrate. I for one find that ogg can match mp3 in sound quality at about 60% of the bitrate. When you use a smaller bitrate, battery life goes up, because your hard drive activity is less. My firsthand experience is that you can get 15 hrs of continuous ogg playback on the karma, if you use a lower bitrate like 64 or 72 kbps. Also, you will note that even if we hypothetically penalized this real-world measurement of 15 hours by a theoretical 25%, it would still be better battery life than an iPod.
As to your dismissal of headphone sound quality, there are a great many headphones that are good enough to tell the difference. Even without good headphones, 72 kbps mp3 is so bad that anyone who is running out of disk space on their portable can easily justify the switch to vorbis.
Actually, privoxy doesn't work that way.
Privoxy is a web proxy, not a browser plugin. That means it slipstreams itself in between your browser and the server. When using privoxy, your actual web browser never actually directly requests anything from the web site itself. All of its requests go through privoxy, and (crucially) privoxy does not actually pass all of the incoming requests through to the remote server.
The result is that when you go to slashdot's home page and there is an ads.osdn.com banner at the top of the page, privoxy doesn't work by first downloading the ad from the server and then preventing you from seeing it. Instead it works by recognizing ads.osdn.com as an advertising site, and not even sending the HTTP GET request at all.
Now, it is true that privoxy has a second, independently functional ad-blocking mechanism that does rely on post-processing the ad after it is downloaded, but ads.osdn.com is well known enough that privoxy can (and does) already decide to eschew even the initial GET request based purely on the URL input.
On the other hand, the Karma beats the Neuros in many ways. You already mentioned playlist management; to this I would add gapless playback (does firmware 2 fix this for the Neuros? please tell me it does) and internationalization (the Neuros LCD display is not even physically capable of rendering Chinese characters).
I do want Neuros to succeed very badly -- I bought one of them the first day, and I like their software philosophy and their support of the USB mass storage standard. But when it comes down to the practical matter of playing music, I much prefer the Karma to the Neuros.
The quickest way to paste a new URL into mozilla on Linux (and, I assume, FreeBSD as well) is to middle click directly into the main browser window. As long as you don't middle click a hyperlink, the browser will automatically load the pasted URL.
A "one-time password" means a password that is used once and discarded. This password is typically used only for authentication purposes. By contrast, a "one-time pad" is used for encryption purposes.
One-time pads are almost never useful for typical internet situations because they are very easy to misuse and very insecure when misused. They also don't solve any problem worth solving -- conventional encryption is already strong enough that the added security of a one-time pad has no value in typical internet situations.
One-time passwords, on the other hand, do potentially have some value, because the currently available password authentication systems are quite weak compared to the strength of the corresponding encryption systems.
XV does indeed work all of the time. I usually have 20 tabs open in my web browser all the time, whether or not I'm watching movies.
The only movie playing problem I ever noticed is that in fullscreen mode mplayer will jerk the playback window to the side when switching between chapters on one or two of my DVDs, but I don't think this is an XVideo problem, because the playback window still works without interruption. None of the other DVD programs (xine, ogle, etc.) have this problem.
I am of course using the opensource radeon drivers since ATI hasn't released 64-bit proprietary drivers. The drivers do not support direct rendering, but this doesn't matter to me since I don't use any 3D programs.
You're right that WEP is insecure, but relying on your alternative recommendations is even worse. Kismet on Linux can sniff out cloaked SSIDs and MAC addresses of any devices on the network as long as the wireless network in question is actively transmitting packets at the same time that Kismet is being used. I have personally verified this functionality firsthand while running Kismet against my own wireless network.
Why does this work? Because the SSID and the MAC address of the destination device is included in plaintext in every wireless data packet (with or without WEP). SSID cloaking doesn't hide SSIDs within the data packets, it only prevents the router from broadcasting the stream of non-data packets announcing its SSID that it would otherwise broadcast. Therefore SSID cloaking is really only effective if you are not transmitting any data, or if your attacker is using an inferior sniffing platform that doesn't have raw access to the SSID in the data packets.
MAC address filtering is ineffective for the same reason -- every data packet is required by the specification to include the destination MAC address, so as long as you are transmitting data, your MAC addresses (all of them, not just the AP's MAC address) are exposed. Once an attacker gets your MAC addresses it is a simple matter to spoof them.
Your recommendation to use higher level layers of security is a very good one, because there is simply no way at the present time to secure wireless networks even to a level that is comparable to what we normally expect out of a wired network.
Just as an alternative (not Slackware related) data point, I'm using X.org on Fedora 2 x86-64 with an ATI Mobility Radeon 9600 and XVideo overlay DOES work, although it's much slower than it was on Fedora 1 using ATI's proprietary drivers. Unfortunately ATI has not yet seen fit to release 64-bit linux drivers. That said, I'm pretty happy that XV works at all... on my last laptop, it took several months before XV support for that card made it into XFree86.
That's certainly not what he said:
Notice that "linux distribution" is not restricted in scope by any qualifiers such as "for PPC." Even if he meant something different than what he said, it is certainly fair game to point out inaccuracies in what was said.
Um, this statement is false. The mysql server and client are fully 64-bit... here's proof.
While we have dozens of distributions there is not a single 64bit Linux out there that is even close to being as full-featured as debian, fedora, redhat, mandrake,... on i386 are...
If you want something as full-featured as i386, then (aside from simply running i386) x86-64 is the best game in town, because it actually runs i386 binaries. For example, my copy of Mathematica for i386 Linux runs perfectly in Fedora 2 x86-64.
As an aside, x86-64 clearly contradicts the story summary's claim of being the "first linux distribution to offer a 64-bit top-to-bottom solution which is not a toy environment." Red Hat Enterprise 3 for AMD64 was released six months ago with a full 64-bit userspace environment, and I don't think anybody can seriously argue that RHEL3 is a "toy environment" compared to a beta gentoo-ppc64 release.