Bunk. If you can do raid over USB, you can do it over 10/100 ethernet. As long as it's just used for data storage, the loss in speed should be no big deal. Windows, at least, would not notice.
I really don't understand how this was modded "3, Insightful". It may be funny, or a troll, depending on intent. But insightful? Only if you didn't bother to read the linked article.
Because the linked article talkes about RAID'ed floppy drives over USB. Yes, for slow, obsolete floppy drives, USB is fast enough.
The claim that one wouldn't notice it using 10/100 ethernet just proves that the previous poster never compared a network file system with a local one for daily work.
Even with 100MBit, my current project compiles in just over 3 minutes (on a old 500Mhz CPU). I have linked the directories for the generated binary files to the local disk (max transfer about 13MB/s), because that speeds up compiling by a factor of 3 [1]. Now imagine using 10MBit and implied slow-down by a factor of about 10 and then reclaim that one wouldn't notice. With a current computer the difference would be even more staggering.
Ah, and if compiling does not fall into the "data storage" category: Well, simply copy that 50MB log file around, and some seconds become minutes (regarding the nobody would notice a "10 MBit" link).
[1] Considering the max transfer of the network file system (about 7MB/s) and the disk (13MB/s), I'd expected a smaller difference, especially because the OS may cache the files. Apparently latency plays a bigger role than I anticipated.
Try this with your "quick" Linux way of doing it. Grab a URL into your clip board. Now, in your web browser, replace the URL in the address bar with the one you just copied. Not so quick now, eh?
Let's see. To my best knowledge (let me know, if there is an easier way) it goes like this on Microsoft Windows:
Mark the text to copy,
Ctrl-C,
Move mouse to address bar of your browser,
Double click in the address bar (or mark the text by manually),
Ctrl-V.
On Linux it's:
Mark the text to copy,
Move mouse to address bar of your browser,
(Single-)Click in the address bar,
Ctrl-U
Press the middle mouse button.
Now, where is that not as quick? Aside from that, with my browser (Mozilla 1.4), in most cases I don't want to take the trip to the address bar (I only need this, if I want to modify to URL afterwards). Loading a page from a copied URL is simple as:
Mark the text to copy,
Move mouse anywhere into the browser window,
Press the middle mouse button.
Now, let's talk about quickness again.:-)
That said, I agree with your point, that selecting implying copying can be problematic. A simple solution is that you can either use Ctrl-C/V additionally (as a separate buffer). And guess what? That's already how most of the applications I use work meanwhile.
Say I create my own programming language (lets call it G) and write a compiler for it and keep it both the language specification and compiler secret. Then I write a program linking to GPL code and release the source code (in G).
I think most people would agree that I have not released the source code in the preferred form for making modifications to it (as nobody has a G compiler except me). If I release my G compiler (with source code) everyone would be happy.
Now the question is whether code for a C compiler, unavailable to the general public, which may have incompatabilities with other widely available compilers, can ever be the preferred form for distribution of GPLed code. There is no way for another developer to recreate the distributed executable with the distributed source code.
ROTFL
Well, that is exactly my argument/interpretation to begin with.:-)
And you had me already (well, half) convinced that the wording of the GPL doesn't support that.
But I stand half convinced. I think in your example above, your are right. The compiler, only written for that project, becomes part of the source code.
But in the case where I write a compiler. And then a lot of projects with that compiler (which is not too unusual within for a big company), then the compiler is an independend project, just not with a big user group. A real-world example of something like this would be the in-house interpreters/template engines written by Yahoo for the web sites (there are 2 or 3 of them; sorry, but don't have the link to the slide show anymore).
Well, let's say that a in-house compiler is a grey area that will only be decided in a court.
To be more precise, it was "getting as many people on Linux as possible, and AWAY from insecure Windows stuff"
suddenly leap to "I want a Linux monopoly as powerful as Microsoft's was!"?
Well, doing as you suggest, i.e. excercising "getting as many people on Linux as possible" consequently enough, could result in a Linux monopoly (as you called it - I never used that expression) sooner than you think.
Thank you for taking my comments and extrapolating them to the furthest possible extreme. I appreciate it, really.
You are welcome.
Well, I am sorry, I missed the "stop before becoming a monopoly" part in "as many as possible". Maybe the misunderstanding here is a question of backgrounds. I, for one, have no doubt that Linux has the potential to get the major player in the OS market, which is all required for getting a monoculture.
I will now propose a "half and half" conversion plan where Linux Zealots, instead of issuing blanket calls for Linux conversion, will simply go down the row going Linux, Microsoft, Linux, Microsoft.
That should do it.
How about "Use whatever fits your needs best?" -- I like this one a lot.
My reading of this section of the GPL is slightly different.
The source code for a work means the preferred form of the work for making modifications to it. 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.
This sentence does not say that anything about distributing executables required to build the work, only source code.
I am not sure if I understand what you meant to say. I did not say anything about executables. I am fine with the source of your compiler.:-) (As long as there is a compiler available for the language it is written in, e.g. the C source for a fortran compiler.)
Regarding executables: scripts used to control compilation and installation can mean a non-executable Makefile, but also something which is directly executable (e.g. "configure"). I even tend to say the autoconf scripts themselves are also covered.
But now I think I understand what you mean: That it talks about definition files and control scripts and so on, but not explictly about the compiler (executable) itself. You are right on this, but that doesn't change my point, because I took the need to also distribute the compiler itself, from the exception clause below (by the argument that the exeption clause wouldn't be necessary, if the above wouldn't be meant to include the compiler). But as you will see below, I am not sure of that interpretation anymore.
However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
This sentence says that you don't need to distribute source for any major component shipped with your OS unless the component is distributed with your work. It says nothing about components not shipped with your OS.
Yes, it does. Indirectly. Since it is declared as an exception, one can deduce that (at least) the authors think there is a rule which implies the contrary, i.e. since it is an exception to the first paragraph above, that this first paragraph is to be understood in a way that the implied rule stands.
But you are right, by a different argument. Presuming that "source code" in the first paragraph does not include the compiler itself, but only the "helper stuff" (the possibility of which I agree to), then this second paragraph is only an exception to this helper stuff, and therefore the compiler itself still wouldn't be part of the requirement.
This reading is seconded by the wording "anything that is normally distributed [...] with the major components [...] of the OS" (emphasis mine). If it was meant to mean these components itself, it would probably be worded without "with". IMHO, the "with" suggests that only parts of the components are discussed, i.e. the "helper stuff".
On the other hand, the GPL FAQ states that this exception also affects libraries coming with the compilier, which are not directly mentioned in "source code" definition above, either. I think they are covered under "modules it contains", although dynamic libaries are not contained in an executable. So we have some inconsistency here.
It provides exceptions to the requirement to distribute source code, not additional requirements.
Nevertheless, describing an exception can clarify the meaning of a rule. For example, the existance of the LGPL makes absolutely clear that GPL is meant to "infect" via dynamic linking (again, at least in the author's point of view). If it wouldn't be the case, there would be (almost) no reason for the LGPL to e
I'm not sure I understand why everyone's so sure they're violating the GPL.
RTFA. It covers all your points.
If the missing kernel modules are entirely original, why are they required to release them under the GPL?
They are not. But the LKML posting clearly states that they are not talking about missing modules, but stuff that has been linked statically.
AFAIK, you are only required to release source that contains GPL-ed code, or that is based on GPL-ed code.
Correct.
If you write a kernel module from scratch, it is not bound by the GPL - Linux API compatibility notwithstanding.
Agreed we don't know if their code is original without seeing it. But if it's original they're not obliged to show it. That's the catch-22 of the GPL I've always found interesting.
Correct, but irrelevant to the case at hand.
Short summary: They show that compiling in network support (and the Linksys kernel binary comes with network support), errors out due to missing files. That violates section 3 of GPLv2 (that "source" means everything needed to compile the binary, except for standard software coming with the OS).
Additionally they show that the static Linksys kernel binary contains symbol names not contained in the whole source Linksys offers (and that those symbols reside in the main kernel address space, not in the module address space, i.e. they are part of the static kernel).
But don't take my word for it, simply read the article itself.
It's a good time, the GPL is bad. As some slashdotter used in his/her sign: I hate MS, but I find the GPL as a greater threat. I do prefer the wxWindows license, and in second place the BSD.
Well, you made a great effort to make your argument clear.;-)
You say the GPL is bad? Why? And for whom?
When I write software, the GPL is the best choice for me. The BSD license allows other to rip off my efforts. With the GPL, I get the deal, that if they build on it and distribute it, I get a part back. From my point of view, that's good, because my target group is me. The software I write in my free time, I write mostly for myself. Publishing it for others, is just a nice side-effect. Giving something back or so.
Sure, the BSD license is nicer to others. E.g. Subversion (a CVS-replacement) uses a BSD-like license and they deny any GPLed stuff to go in. I understand and second their motivation, because their target group are CVS users. They want Subversion to be broadly adopted and for that they need a weak license.
So before saying GPL is bad, it would be good to clarify the relations you have in mind.
These guys use "we can't compile it" argument just to prove that the source code is incomplete
That summary is incomplete:
1. They also show with great detail that the downloadable binary contains symbol names not contained in the published source.
2. They don't say "we can't compile it", but "compiling the network gives errors about missing files."
3. Since the distributed binary comes with network support, yes, "2." alone would already be enough to show that the GPLv2 has been violated (see section 3, the part about what "source code" is defined as; see quote below), even if just some script is missing.
If someone released source code under GPL, do they also need to make sure that others can compile it?
Depends on what you mean by that. Of course, they don't have to handle the case if your system has a broken installation. If you mean, whether they have to assure that you have everything needed to compile it, then the answer is: yes, with one exception (from section 3 of the GPLv2):
The source code for a work means the preferred form of the work for making modifications to it. 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 (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
So, yes, they have to make available everything needed to create the binary from source, except what is easily available elsewhere.
suppose that linksys ported linux to INTERCAL++ and released full source code but did not provide the compiler they made in-house. Would that still violate GPL?
That rule above is also the reason why you cannot simply create a special compiler in order to circumvent the GPL. The compiler wouldn't be usually distributed with your OS, so it is not covered by the exception, and you have to distribute it with the source.
They only have to release their code if it is part of a derivative work based on the GPLed code. A port of the Linux kernel to INTERCAL++ would probably be a derivative work, but the compiler might well be a separate work, so they would not need to release its source.
That's not entirely correct. Of course, nothing can force them to release their compiler, but the GPL forbids them to release the derivative work (here: the kernel) without the in-house compiler.
See section 3 of the GPL v2:
The source code for a work means the preferred form of the work for making modifications to it. 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 (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
So, they only don't need to include the compiler, if it is normally distributed with the OS, which isn't the case, of course, for an in-house compiler.
This is an often overlooked rule, because having the compiler easily available is the rule and not the exception with most OSes.
All of that is easily taken care of by anybody willing to spend more than two minutes to read a manual.
Translation: You're incompetent.
I am not sure why this is moderated as flamebait. OK, calling someone incompetent isn't nice, but I was about to say the same thing (in different words), until I saw that somebody else already made that (IMHO correct conclusion). Maybe it helps the moderators, if the statement is more specific:
You're incompetent as a Linux admin.
Or as I would have said it: All you have proven is that you are a more competent Microsoft Windows admin than Linux admin.
(A proof of that is that you use a third party FTP server on Microsoft Windows, but seemingly complain about the default one for your Linux distro. Same rights for both: either use default software for both, or use VsFtpd or ProFtpd on Linux).
It seems self-evident that getting as many people on Linux as possible, and AWAY from insecure Windows stuff, would be a boon to the Internet as a whole.
Nope. Monocultures are bad. Although Linux based distributions offer more diversity than Microsoft Windows, security-wise a good mixture of OSes would probably be best.
You don't need to take my word for it. Just read the recent article - featured on Slashdot - on why Microsoft monopoly is bad. They explicitly state that Solaris having Microsoft Windows' marketshare would be almost equally bad. And that comes from some of the best security experts.
As I said, Linux distros are not as bad (by an order of magnitude, IMNSHO), because they often come with several packages for the same goal (e.g. postfix/sendmail/exim/...). But still, an attack targeting a commonly used feature (TCP stack?) would still have catastrophic effects. And don't forget that the mass of installations would be end-users, thus default installs. That means less diversity.
Well, I'm sorry, but Linux can't be everything to everybody at all times.
In that case, people shouldn't bitch or complain about "skewed" statistics when Windows Server 2003 takes some installations from Linux.
I hope you do realize that (most of) the self-proclaimed Linux advocates bitching about "skewed" statistics aren't necessarily (well, not even probably) Linux developers to begin with (i.e. the kind of people who can, do and will influence the direction Linux takes).
Or in other words: Just because some kids on Slashdot get a break-down on seeing those numbers, doesn't mean anybody else sees it their way.
You do realize that sarcastic analogy makes utterly no sense, right?
Unless you're seriously suggesting that Linux is only for, and should only ever be aimed at, certain specialized markets. Is that, in fact, your intent?
Although his analogy looks misfitting at first glance, his point is completely correct. To be more explicit: The analogy was an example of the fact that usually marketing has a target group as part of the strategy. And that claiming a marketing failure without first agreeing on the target group "makes utterly no sense", as you correctly observed.
So (this is more directed to JayBlalock) what is the target group for Linux? If you take Linux, the kernel, I guess Linus can be taken as authority. So you get this answer (from a recent interview):
The thing is, at least to me personally, Microsoft just isn't relevant to what I do. That might sound strange, since they are clearly the dominant player in the market that Linux is in, but the thing is: I'm not in the ''market.'' I'm interested in Linux because of the technology, and Linux wasn't started as any kind of rebellion against the ''evil Microsoft empire.'' Quite the reverse, in fact: from a technology angle, Microsoft really has been one of the least interesting companies. So I've never seen it as a ''Linus versus Bill'' thing. I just can't see myself in the position of the nemesis, since I just don't care enough. To be a nemesis, you have to actively try to destroy something, don't you? Really, I'm not out to destroy Microsoft. That will just be a completely unintentional side effect.
So from Torvald's point of view, as he says, he is not in the market.
If you mean Linux, the OS (more precisely: Linux based distributions), I doubt you can get a single answer. Ask Debian, Mandrake, SuSE, Lindows, Slackware or all the others of over 150 distributors, and you will get as many answers, I bet.
So, before telling anything a marketing failure, which marketing strategy do you presume? And by whom?
And no, simple "world domination" aka "Linux on every PC" doesn't count. That might sound catchy, but isn't a marketing strategy. It is an delusion and will get you laughed out of any serious marketing agency (except if you mean present it as slogan, not as serious aim).
1. Why do you presume they had measured fuel use in the very first test they did? Since we cannot know, 0 points for your claim.
2. But what makes you think, that this cannot be engineered to be only be the skin for the time of passing the supersonic barrier? Remember the Concorde, which moved its nose for supersonic flight? The part with the friction sounds plausible enough. So 1 point for your claim, 0 points for it becoming a problem (which wasn't your direct claim, I know).
3. This only suggests that it seems more probable it won't use less fuel. Not that it will use more. 0 points to the claim that they will be even bigger gas guzzlers than normal.
Of course, the NYT had to do all these lot of hours of work in any case, because they had a huge, whopping security hole. What makes you think they can presume he was the first/only one to find it?
Yours:
What makes you believe that's ALL he did? As to a security hole, the asshole coulda just dropped them a note, not taken advantage of it to vandalize 'em.
Very interesting. Now, please enlighten me what your reply has to do with my statements? Are you just trolling?
Regardless. I never claimed "that's" ALL he did. (Btw, what "that"? I didn't mention any actions by him.) I claimed that he caused them no more work than had to do anyway (i.e. if they had found the hole themselves).
asshole coulda just dropped them a note
Well, you will laugh, that's what he did.
not taken advantage of it to vandalize 'em
Ah, so adding his contact info to their database is now called vandalism. Okay, you are trolling.
Furthermore that's what he claims he did, NYT then had to go and verify that is actually all he did, that it and other databases were still valid, etc. Tha'ts a lot of hours of work.
You don't really believe yourself, do you?
Of course, the NYT had to do all these lot of hours of work in any case, because they had a huge, whopping security hole. What makes you think they can presume he was the first/only one to find it?
not the crime he did commit (to walk through an unlocked door).
Excuse my ignorance, but is this really a crime in the USA? AFAIK local laws, in Germany anyone can walk into any open (as in "not closed", not "not locked") area as it pleases him/her, until and only until, you say him he is not welcome. Then you can call the police if he stays or reenters.
That's probably why most estates have garden fences. Most of them don't stop anyone, but they declare the garden a "closed" area (presumed that the fence gate is closed, of course).
So, yes, I can simply walk into a stranger's house, as long as he left the door open, and given that I don't do anything illegal additionally, there was no crime. (But that doesn't mean that the owner won't call the police and the police won't hold me and investigate what illigal activity I might have done in the house, if I don't have a reasonable explanation for being in that house).
Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway.
Yeah, but the latency sucks.;)
Re:If I try it, will I like it?
on
Mandrake 9.2 RC1
·
· Score: 4, Insightful
So, is this going to be a worthwhile investment of my time, or is some/most of the above going to stop working? While I'd like to get off my MS Habit, I have this need to use most of the above that keeps pulling me back:-)
If you "just" want to give it a try, please, PLEASE, don't install the RC, but wait until the final release is made. Even better, wait some weeks (2-3) after final release for the most important glitches to be known and read the errata before you are going to install.
The RCs aren't for "newcomers" to try out. They for are testing and bug fixing by people willing investing the time to do so.
But if you want to get the best chance that the final will support as much as possible on your notebook and you are willing to invest some time, then by all means, install it, test everything that matters to you and report any bugs you find (but be sure to look beforehand if the bug is already known).
So, whether the investment of your time is worthwhile depends on your view. If you are not willing, are not interested or simply don't have the time for testing/bug reporting, the RC isn't for you. Well, it wasn't ever meant for you.
Else, now is the time to get involved before it is too late (for getting any updates in before the freeze gets to freezy - they are already in version freeze).
While just saying "Hello" in my example is not really even a small-talk conversation, I think it's used pretty much the same way. It's just one person testing another to see how they resonate with them.
Hm. I usually get onto an elevator and forget about the existing occupant, following some thoughts until I subconsiously notice that this "ping" was for my floor, and hurry to get out before the door closes again.
Sometimes it happens, that I get unnerved by the other occupant who "radiates" his discomfort too strongly, until I take a heart and make at least eye-contact, which initiates the usual awkward silence. Well, at least, until I fall into deep thoughts again and forget about the environment again.:-)
And of course, there is the other variation, where I am in the elevator and hardly notice the poor fellow who just entered and now desparately tries to make eye-contact. (Btw, most people don't greet if they didn't made a successful eye-contact beforehand.)
I think it is a kind of "Friend or Foe" detector.
Despite of what I said above, I agree. It's just that its use seems to be restricted to certain personality types.
My biggest annoyance with Linux is this attitude that
* all Linux users can code,
* all Linux users want to code,
* all Linux users know every api to code,
* all Linux users want to join every devel mail list,
* all Linux users know every application's architecture,
* all Linux users have infinite time to solve obvious problems.
I am a software developer by trade. I know a fair amount of user interaction design principles. That doesn't mean I have the lifestyle that affords me ninety hours a week to add nothing but polish the nits out of the hundred different Linux applications I use every week.
You got this all the wrong way around:
If you are using a Linux based distribution, you are mainly using software for free (in the sense, that you didn't compensate the authors). Then you observe a glitch or bug and report it and expect it to get fixed. Again for free.
If the issue is real, can be reproduced and viewed as important enough (relative to what else has to be done), it will usually be fixed, implemented or put on the TODO. Even if the request is rejected, others have already invested their time for your interests.
But a lot of reports are really more of the kind "I don't like it being/behaving this way, couldn't you make it that way". And such issues usually are only fixed, if there is not too much controversy, someone agrees, and is willing to implement it for you. If your concern is not important enough to any author, you are probably asked to provide a patch. Also, in the whole process, there are often requests to provide more detailed info, the compiling of which may take some time.
This is not about attitude, it is a simple "if you want us to fix/program/integrate it, then provide a testcase/requirements/patches". You want someone do something for free for you[1], so provide some initiative.
Most "real"[2] projects I observed handle such request quite politely (as a rule, exceptions happen, of course). Naturally, there are also the other ones (mplayer comes to mind;), which may be not all that nice. Although those may lack in patience or manners, the underlying issue is the same: Don't complain, if you aren't willing to help fixing it.
Or in other words: Don't expect others to do, what you aren't willing to do yourself. If you cannot respect this, the attitude isn't with the "Linux people".
You might think "But I am helping them to improve their software!". Well, think about it. If the authors had observed/needed/wanted whatever you reported, and found it important, don't you think, they had already done it themselves? In real, most times you are effectively not helping the author, but other users like yourself.
So, to conclude, yes, I completely understand that it may be out of proportion for you to invest the time and effort needed to fix whatever annoyances you perceive. But if other people are not interested to fix "your" annoyances, and ask you to look into it yourself, that has seldom to do something with attitude.
Coming back to the start, "That doesn't mean I have the lifestyle that affords me ninety hours a week to add nothing but polish the nits out of the hundred different Linux applications I use every week." is an interesting statement on how you value your time with regard to the time of others.
[1] Even integrating a seemingly perfect patch can easily hours! It includes reviewing and testing the patch, maybe writing missing test-cases and so on.
[2] This is meant to exclude wanna-be projects, i.e. count the projects that are included in major distros or have a reasonable chance of being included.
But it would be totally pointless if someone did. They can't do anything WITH the source on the system. It's a closed system. They can't get into it without violating the extremely expensive warranty they purchased. The only purpose in asking would be merely to cause us trouble.
There is a very valid reason: If you go out of business (or there is some other reason, why they cannot/want not do business with you any longer), they can (at least theoretically) do something theirselves, hire another company, whatever.
It only sounds like "merely to cause you trouble" to you, because you are unable to see a valid reason; and (a version of) bash (that runs your scripts without error) is easy to come by these days. Think of what is in 10 or 20 years. You get the idea.
If someone asked, and they were a licensed customer (not a third party), then we would do it.
You only have to give it to a customer, except, of course, if the customer selled the system. But if you didn't at least hand out the "you can get the source from us" notice, you have already broken the GPL.
ext2/ext3 filesystems trashed beyond repair : 4-6 per year
Wow, I never seen ext2 getting beyond repair (except on hardware failure, of course). Mind sharing on how much systems these numbers are based (i.e. 4-6 for 10 system is huge, for 10000, not that much;)?
Slashdot Mirror
Bunk. If you can do raid over USB, you can do it over 10/100 ethernet. As long as it's just used for data storage, the loss in speed should be no big deal. Windows, at least, would not notice.
I really don't understand how this was modded "3, Insightful". It may be funny, or a troll, depending on intent. But insightful? Only if you didn't bother to read the linked article.
Because the linked article talkes about RAID'ed floppy drives over USB. Yes, for slow, obsolete floppy drives, USB is fast enough.
The claim that one wouldn't notice it using 10/100 ethernet just proves that the previous poster never compared a network file system with a local one for daily work.
Even with 100MBit, my current project compiles in just over 3 minutes (on a old 500Mhz CPU). I have linked the directories for the generated binary files to the local disk (max transfer about 13MB/s), because that speeds up compiling by a factor of 3 [1]. Now imagine using 10MBit and implied slow-down by a factor of about 10 and then reclaim that one wouldn't notice. With a current computer the difference would be even more staggering.
Ah, and if compiling does not fall into the "data storage" category: Well, simply copy that 50MB log file around, and some seconds become minutes (regarding the nobody would notice a "10 MBit" link).
[1] Considering the max transfer of the network file system (about 7MB/s) and the disk (13MB/s), I'd expected a smaller difference, especially because the OS may cache the files. Apparently latency plays a bigger role than I anticipated.
Try this with your "quick" Linux way of doing it. Grab a URL into your clip board. Now, in your web browser, replace the URL in the address bar with the one you just copied. Not so quick now, eh?
:-)
Let's see. To my best knowledge (let me know, if there is an easier way) it goes like this on Microsoft Windows:
Mark the text to copy,
Ctrl-C,
Move mouse to address bar of your browser,
Double click in the address bar (or mark the text by manually),
Ctrl-V.
On Linux it's:
Mark the text to copy,
Move mouse to address bar of your browser,
(Single-)Click in the address bar,
Ctrl-U
Press the middle mouse button.
Now, where is that not as quick? Aside from that, with my browser (Mozilla 1.4), in most cases I don't want to take the trip to the address bar (I only need this, if I want to modify to URL afterwards). Loading a page from a copied URL is simple as:
Mark the text to copy,
Move mouse anywhere into the browser window,
Press the middle mouse button.
Now, let's talk about quickness again.
That said, I agree with your point, that selecting implying copying can be problematic. A simple solution is that you can either use Ctrl-C/V additionally (as a separate buffer). And guess what? That's already how most of the applications I use work meanwhile.
Say I create my own programming language (lets call it G) and write a compiler for it and keep it both the language specification and compiler secret. Then I write a program linking to GPL code and release the source code (in G).
:-)
I think most people would agree that I have not released the source code in the preferred form for making modifications to it (as nobody has a G compiler except me). If I release my G compiler (with source code) everyone would be happy.
Now the question is whether code for a C compiler, unavailable to the general public, which may have incompatabilities with other widely available compilers, can ever be the preferred form for distribution of GPLed code. There is no way for another developer to recreate the distributed executable with the distributed source code.
ROTFL
Well, that is exactly my argument/interpretation to begin with.
And you had me already (well, half) convinced that the wording of the GPL doesn't support that.
But I stand half convinced. I think in your example above, your are right. The compiler, only written for that project, becomes part of the source code.
But in the case where I write a compiler. And then a lot of projects with that compiler (which is not too unusual within for a big company), then the compiler is an independend project, just not with a big user group. A real-world example of something like this would be the in-house interpreters/template engines written by Yahoo for the web sites (there are 2 or 3 of them; sorry, but don't have the link to the slide show anymore).
Well, let's say that a in-house compiler is a grey area that will only be decided in a court.
Well, I am sorry, I missed the "stop before becoming a monopoly" part in "as many as possible". Maybe the misunderstanding here is a question of backgrounds. I, for one, have no doubt that Linux has the potential to get the major player in the OS market, which is all required for getting a monoculture.How about "Use whatever fits your needs best?" -- I like this one a lot.
I am not sure if I understand what you meant to say. I did not say anything about executables. I am fine with the source of your compiler. :-)
(As long as there is a compiler available for the language it is written in, e.g. the C source for a fortran compiler.)
Regarding executables: scripts used to control compilation and installation can mean a non-executable Makefile, but also something which is directly executable (e.g. "configure"). I even tend to say the autoconf scripts themselves are also covered.
But now I think I understand what you mean: That it talks about definition files and control scripts and so on, but not explictly about the compiler (executable) itself. You are right on this, but that doesn't change my point, because I took the need to also distribute the compiler itself, from the exception clause below (by the argument that the exeption clause wouldn't be necessary, if the above wouldn't be meant to include the compiler). But as you will see below, I am not sure of that interpretation anymore.
Yes, it does. Indirectly. Since it is declared as an exception, one can deduce that (at least) the authors think there is a rule which implies the contrary, i.e. since it is an exception to the first paragraph above, that this first paragraph is to be understood in a way that the implied rule stands.
But you are right, by a different argument. Presuming that "source code" in the first paragraph does not include the compiler itself, but only the "helper stuff" (the possibility of which I agree to), then this second paragraph is only an exception to this helper stuff, and therefore the compiler itself still wouldn't be part of the requirement.
This reading is seconded by the wording "anything that is normally distributed [...] with the major components [...] of the OS" (emphasis mine). If it was meant to mean these components itself, it would probably be worded without "with". IMHO, the "with" suggests that only parts of the components are discussed, i.e. the "helper stuff".
On the other hand, the GPL FAQ states that this exception also affects libraries coming with the compilier, which are not directly mentioned in "source code" definition above, either. I think they are covered under "modules it contains", although dynamic libaries are not contained in an executable. So we have some inconsistency here.
Nevertheless, describing an exception can clarify the meaning of a rule. For example, the existance of the LGPL makes absolutely clear that GPL is meant to "infect" via dynamic linking (again, at least in the author's point of view). If it wouldn't be the case, there would be (almost) no reason for the LGPL to e
I'm not sure I understand why everyone's so sure they're violating the GPL.
RTFA. It covers all your points.
If the missing kernel modules are entirely original, why are they required to release them under the GPL?
They are not. But the LKML posting clearly states that they are not talking about missing modules, but stuff that has been linked statically.
AFAIK, you are only required to release source that contains GPL-ed code, or that is based on GPL-ed code.
Correct.
If you write a kernel module from scratch, it is not bound by the GPL - Linux API compatibility notwithstanding.
Agreed we don't know if their code is original without seeing it. But if it's original they're not obliged to show it. That's the catch-22 of the GPL I've always found interesting.
Correct, but irrelevant to the case at hand.
Short summary: They show that compiling in network support (and the Linksys kernel binary comes with network support), errors out due to missing files. That violates section 3 of GPLv2 (that "source" means everything needed to compile the binary, except for standard software coming with the OS).
Additionally they show that the static Linksys kernel binary contains symbol names not contained in the whole source Linksys offers (and that those symbols reside in the main kernel address space, not in the module address space, i.e. they are part of the static kernel).
But don't take my word for it, simply read the article itself.
It's a good time, the GPL is bad.
;-)
As some slashdotter used in his/her sign:
I hate MS, but I find the GPL as a greater threat.
I do prefer the wxWindows license, and in second place the BSD.
Well, you made a great effort to make your argument clear.
You say the GPL is bad? Why? And for whom?
When I write software, the GPL is the best choice for me. The BSD license allows other to rip off my efforts. With the GPL, I get the deal, that if they build on it and distribute it, I get a part back. From my point of view, that's good, because my target group is me. The software I write in my free time, I write mostly for myself. Publishing it for others, is just a nice side-effect. Giving something back or so.
Sure, the BSD license is nicer to others. E.g. Subversion (a CVS-replacement) uses a BSD-like license and they deny any GPLed stuff to go in. I understand and second their motivation, because their target group are CVS users. They want Subversion to be broadly adopted and for that they need a weak license.
So before saying GPL is bad, it would be good to clarify the relations you have in mind.
That summary is incomplete:
1. They also show with great detail that the downloadable binary contains symbol names not contained in the published source.
2. They don't say "we can't compile it", but "compiling the network gives errors about missing files."
3. Since the distributed binary comes with network support, yes, "2." alone would already be enough to show that the GPLv2 has been violated (see section 3, the part about what "source code" is defined as; see quote below), even if just some script is missing.
If someone released source code under GPL, do they also need to make sure that others can compile it?
Depends on what you mean by that. Of course, they don't have to handle the case if your system has a broken installation. If you mean, whether they have to assure that you have everything needed to compile it, then the answer is: yes, with one exception (from section 3 of the GPLv2):So, yes, they have to make available everything needed to create the binary from source, except what is easily available elsewhere.
suppose that linksys ported linux to INTERCAL++ and released full source code but did not provide the compiler they made in-house. Would that still violate GPL?
That rule above is also the reason why you cannot simply create a special compiler in order to circumvent the GPL. The compiler wouldn't be usually distributed with your OS, so it is not covered by the exception, and you have to distribute it with the source.
That's not entirely correct. Of course, nothing can force them to release their compiler, but the GPL forbids them to release the derivative work (here: the kernel) without the in-house compiler.
See section 3 of the GPL v2:So, they only don't need to include the compiler, if it is normally distributed with the OS, which isn't the case, of course, for an in-house compiler.
This is an often overlooked rule, because having the compiler easily available is the rule and not the exception with most OSes.
All of that is easily taken care of by anybody willing to spend more than two minutes to read a manual.
Translation: You're incompetent.
I am not sure why this is moderated as flamebait. OK, calling someone incompetent isn't nice, but I was about to say the same thing (in different words), until I saw that somebody else already made that (IMHO correct conclusion). Maybe it helps the moderators, if the statement is more specific:
You're incompetent as a Linux admin.
Or as I would have said it: All you have proven is that you are a more competent Microsoft Windows admin than Linux admin.
(A proof of that is that you use a third party FTP server on Microsoft Windows, but seemingly complain about the default one for your Linux distro. Same rights for both: either use default software for both, or use VsFtpd or ProFtpd on Linux).
It seems self-evident that getting as many people on Linux as possible, and AWAY from insecure Windows stuff, would be a boon to the Internet as a whole.
Nope. Monocultures are bad. Although Linux based distributions offer more diversity than Microsoft Windows, security-wise a good mixture of OSes would probably be best.
You don't need to take my word for it. Just read the recent article - featured on Slashdot - on why Microsoft monopoly is bad. They explicitly state that Solaris having Microsoft Windows' marketshare would be almost equally bad. And that comes from some of the best security experts.
As I said, Linux distros are not as bad (by an order of magnitude, IMNSHO), because they often come with several packages for the same goal (e.g. postfix/sendmail/exim/...). But still, an attack targeting a commonly used feature (TCP stack?) would still have catastrophic effects. And don't forget that the mass of installations would be end-users, thus default installs. That means less diversity.
I hope you do realize that (most of) the self-proclaimed Linux advocates bitching about "skewed" statistics aren't necessarily (well, not even probably) Linux developers to begin with (i.e. the kind of people who can, do and will influence the direction Linux takes).
Or in other words: Just because some kids on Slashdot get a break-down on seeing those numbers, doesn't mean anybody else sees it their way.
Unless you're seriously suggesting that Linux is only for, and should only ever be aimed at, certain specialized markets. Is that, in fact, your intent?
Although his analogy looks misfitting at first glance, his point is completely correct. To be more explicit: The analogy was an example of the fact that usually marketing has a target group as part of the strategy. And that claiming a marketing failure without first agreeing on the target group "makes utterly no sense", as you correctly observed.
So (this is more directed to JayBlalock) what is the target group for Linux? If you take Linux, the kernel, I guess Linus can be taken as authority. So you get this answer (from a recent interview):
So from Torvald's point of view, as he says, he is not in the market.
If you mean Linux, the OS (more precisely: Linux based distributions), I doubt you can get a single answer. Ask Debian, Mandrake, SuSE, Lindows, Slackware or all the others of over 150 distributors, and you will get as many answers, I bet.
So, before telling anything a marketing failure, which marketing strategy do you presume? And by whom?
And no, simple "world domination" aka "Linux on every PC" doesn't count. That might sound catchy, but isn't a marketing strategy. It is an delusion and will get you laughed out of any serious marketing agency (except if you mean present it as slogan, not as serious aim).
2. But what makes you think, that this cannot be engineered to be only be the skin for the time of passing the supersonic barrier? Remember the Concorde, which moved its nose for supersonic flight? The part with the friction sounds plausible enough. So 1 point for your claim, 0 points for it becoming a problem (which wasn't your direct claim, I know).
3. This only suggests that it seems more probable it won't use less fuel. Not that it will use more. 0 points to the claim that they will be even bigger gas guzzlers than normal.
Very interesting. Now, please enlighten me what your reply has to do with my statements? Are you just trolling?
Regardless. I never claimed "that's" ALL he did. (Btw, what "that"? I didn't mention any actions by him.) I claimed that he caused them no more work than had to do anyway (i.e. if they had found the hole themselves).Well, you will laugh, that's what he did.Ah, so adding his contact info to their database is now called vandalism. Okay, you are trolling.
Furthermore that's what he claims he did, NYT then had to go and verify that is actually all he did, that it and other databases were still valid, etc. Tha'ts a lot of hours of work.
You don't really believe yourself, do you?
Of course, the NYT had to do all these lot of hours of work in any case, because they had a huge, whopping security hole. What makes you think they can presume he was the first/only one to find it?
not the crime he did commit (to walk through an unlocked door).
Excuse my ignorance, but is this really a crime in the USA? AFAIK local laws, in Germany anyone can walk into any open (as in "not closed", not "not locked") area as it pleases him/her, until and only until, you say him he is not welcome. Then you can call the police if he stays or reenters.
That's probably why most estates have garden fences. Most of them don't stop anyone, but they declare the garden a "closed" area (presumed that the fence gate is closed, of course).
So, yes, I can simply walk into a stranger's house, as long as he left the door open, and given that I don't do anything illegal additionally, there was no crime. (But that doesn't mean that the owner won't call the police and the police won't hold me and investigate what illigal activity I might have done in the house, if I don't have a reasonable explanation for being in that house).
Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway.
;)
Yeah, but the latency sucks.
So, is this going to be a worthwhile investment of my time, or is some/most of the above going to stop working? While I'd like to get off my MS Habit, I have this need to use most of the above that keeps pulling me back :-)
If you "just" want to give it a try, please, PLEASE, don't install the RC, but wait until the final release is made. Even better, wait some weeks (2-3) after final release for the most important glitches to be known and read the errata before you are going to install.
The RCs aren't for "newcomers" to try out. They for are testing and bug fixing by people willing investing the time to do so.
But if you want to get the best chance that the final will support as much as possible on your notebook and you are willing to invest some time, then by all means, install it, test everything that matters to you and report any bugs you find (but be sure to look beforehand if the bug is already known).
So, whether the investment of your time is worthwhile depends on your view. If you are not willing, are not interested or simply don't have the time for testing/bug reporting, the RC isn't for you. Well, it wasn't ever meant for you.
Else, now is the time to get involved before it is too late (for getting any updates in before the freeze gets to freezy - they are already in version freeze).
Likewise: the really experienced people would probably chose option 2.
;-)
I agree. People in the know what they don't know.
Well, that's the trick. Option three would identify you als probably wanna-be and get you to the supporter for "problematic customers".
While just saying "Hello" in my example is not really even a small-talk conversation, I think it's used pretty much the same way. It's just one person testing another to see how they resonate with them.
:-)
Hm. I usually get onto an elevator and forget about the existing occupant, following some thoughts until I subconsiously notice that this "ping" was for my floor, and hurry to get out before the door closes again.
Sometimes it happens, that I get unnerved by the other occupant who "radiates" his discomfort too strongly, until I take a heart and make at least eye-contact, which initiates the usual awkward silence. Well, at least, until I fall into deep thoughts again and forget about the environment again.
And of course, there is the other variation, where I am in the elevator and hardly notice the poor fellow who just entered and now desparately tries to make eye-contact. (Btw, most people don't greet if they didn't made a successful eye-contact beforehand.)
I think it is a kind of "Friend or Foe" detector.
Despite of what I said above, I agree. It's just that its use seems to be restricted to certain personality types.
My biggest annoyance with Linux is this attitude that
;), which may be not all that nice. Although those may lack in patience or manners, the underlying issue is the same: Don't complain, if you aren't willing to help fixing it.
* all Linux users can code,
* all Linux users want to code,
* all Linux users know every api to code,
* all Linux users want to join every devel mail list,
* all Linux users know every application's architecture,
* all Linux users have infinite time to solve obvious problems.
I am a software developer by trade. I know a fair amount of user interaction design principles. That doesn't mean I have the lifestyle that affords me ninety hours a week to add nothing but polish the nits out of the hundred different Linux applications I use every week.
You got this all the wrong way around:
If you are using a Linux based distribution, you are mainly using software for free (in the sense, that you didn't compensate the authors). Then you observe a glitch or bug and report it and expect it to get fixed. Again for free.
If the issue is real, can be reproduced and viewed as important enough (relative to what else has to be done), it will usually be fixed, implemented or put on the TODO. Even if the request is rejected, others have already invested their time for your interests.
But a lot of reports are really more of the kind "I don't like it being/behaving this way, couldn't you make it that way". And such issues usually are only fixed, if there is not too much controversy, someone agrees, and is willing to implement it for you. If your concern is not important enough to any author, you are probably asked to provide a patch. Also, in the whole process, there are often requests to provide more detailed info, the compiling of which may take some time.
This is not about attitude, it is a simple "if you want us to fix/program/integrate it, then provide a testcase/requirements/patches". You want someone do something for free for you[1], so provide some initiative.
Most "real"[2] projects I observed handle such request quite politely (as a rule, exceptions happen, of course). Naturally, there are also the other ones (mplayer comes to mind
Or in other words: Don't expect others to do, what you aren't willing to do yourself. If you cannot respect this, the attitude isn't with the "Linux people".
You might think "But I am helping them to improve their software!". Well, think about it. If the authors had observed/needed/wanted whatever you reported, and found it important, don't you think, they had already done it themselves? In real, most times you are effectively not helping the author, but other users like yourself.
So, to conclude, yes, I completely understand that it may be out of proportion for you to invest the time and effort needed to fix whatever annoyances you perceive. But if other people are not interested to fix "your" annoyances, and ask you to look into it yourself, that has seldom to do something with attitude.
Coming back to the start, "That doesn't mean I have the lifestyle that affords me ninety hours a week to add nothing but polish the nits out of the hundred different Linux applications I use every week." is an interesting statement on how you value your time with regard to the time of others.
[1] Even integrating a seemingly perfect patch can easily hours! It includes reviewing and testing the patch, maybe writing missing test-cases and so on.
[2] This is meant to exclude wanna-be projects, i.e. count the projects that are included in major distros or have a reasonable chance of being included.
But it would be totally pointless if someone did. They can't do anything WITH the source on the system. It's a closed system. They can't get into it without violating the extremely expensive warranty they purchased. The only purpose in asking would be merely to cause us trouble.
There is a very valid reason: If you go out of business (or there is some other reason, why they cannot/want not do business with you any longer), they can (at least theoretically) do something theirselves, hire another company, whatever.
It only sounds like "merely to cause you trouble" to you, because you are unable to see a valid reason; and (a version of) bash (that runs your scripts without error) is easy to come by these days.
Think of what is in 10 or 20 years. You get the idea.
If someone asked, and they were a licensed customer (not a third party), then we would do it.
You only have to give it to a customer, except, of course, if the customer selled the system. But if you didn't at least hand out the "you can get the source from us" notice, you have already broken the GPL.
ext2/ext3 filesystems trashed beyond repair : 4-6 per year
;)?
Wow, I never seen ext2 getting beyond repair (except on hardware failure, of course). Mind sharing on how much systems these numbers are based (i.e. 4-6 for 10 system is huge, for 10000, not that much
That's incorrect. "data" journalling mode of ext3 does journal data. As does reiser4, which I take from this comment of the referenced benchmark: So yes, the systems being compared all supported journalling of data. ext3 with only meta-data journalling was included,