Or this? MirrorCentral is different to FreeNet - FreeNet is a great idea, but it's software and will take time to develop and deploy. In the meantime MirrorCentral is the kind of solution that Runna^Muck proposed.
No, but if the shell (explorer) crashes, that's just as bad. Used to happen a lot with SP4, still sometimes happens with SP5, in my limited experience.
If you as a company can't prevent grevious injuries from mishaps which even intelligent people will have, then you shouldn't serve coffee through the drivethrough.
Oh come on
Cars kill a lot more people than spilt coffee. A lot more. Even intelligent people sometimes kill themselves and other people in car accidents.
So should we impose a ban on the sale of all cars?
What about the customer's responsibility to prevent such accidents? Like, don't put a cup of scalding hot coffe between your legs? It's kinda stupid, you gotta admit.
Right, but it's not quite that simple. If you modify the behaviour of the program but keep your changes sufficiently insulated (for an extreme example, write a whole new executable that runs on a different machine and communicates by TCP/IP) then the "insulated" part of your code does not have to be GPLed.
That's how non-GPL'd code such as Netscape can legally call the GPL'd kernel, because they are sufficiently separated.
My question is, what counts as "sufficiently insulated"? Any legal opinion on this?
With the LGPL or MPL, it's pretty clear - if you change any of the original source files, those changes must be covered by the original license - any new source files you create, you can do what you want with, right?
True, but the number of people killed by murderers (outside war situations) is insignificant compared to the number of people killed by, say, eating unhealthy food, malnutrition, drinking unhealthy water, smoking, being hit by vehicles, or radioactive dust (in Iraq, Japan, and Kosovo). The media has its priorities all wrong - unsurprisingly, since bizarre murders are just the sort of thing that can't reasonably be blamed on dominant institutions in our society, whilst the others can, at least partly.
Well, seeing as Netscape (now a subsidiary of AOL) provides the hosting for Mozilla, this is probably one of the few open source projects that won't be attracted to move to SourceForge!
If you knew anything about CPU design you would know that CPUs now often contain (onboard), routines written in microcode (which is a level below "machine code" as traditionally understood). These microcode routines could be open sourced, as they are just software.
Um, that wouldn't be very sensible. A JIT compiler for Java, for example, doesn't compile source code, it compiles bytecodes into native codes. Since bytecode can be considered an assembly/machine language for an abstract virtual machine, it's effectively an optimising cross-assembler - the word "compiler" in the phrase "just in time compiler" is a bit misleading.
And you know this how? You're the director of the NSA, right?
And you're sure there aren't any top-secret sub-agencies?
Only a few weeks ago the Guardian newspaper revealed that a secret nuclear facility in England had several times been close to a nuclear accident at least as severe as the one in Japan. Democratic governments do hide the truth from people. They do have secret facilities.
If you make a copy of a piece of software *and* you give it to a friend *and* he uses this software instead of buying a copy from the local shop, how is this different than shoplifting a copy? (Other than the fact you've taken a free copy of the documentation and gotten spiffy packaging with it.) That's one less copy sold, and that's money out of the developer's pocket either way, isn't it?
Okay, what if, immediately after you give it to your friend, you erase all your own copies, if any?
Friend uses the gift instead of buying a new copy
One less copy sold
Money out of the developer's pocket
Therefore, by analogy, it's still theft... Um, I don't think so. Not even Microsoft thinks so.
You need a better analogy to prove your case, because the analogy can also be applied to "prove" that giving away is theft, which is nonsense.
If I have a heart attack and something goes wrong, what's my doctor going to do...Post on/. or look it up in a newsgroup. No way.
That's an issue of technical support, not open vs closed. You can still buy technical support (and very high quality tech support at that, in some cases) for OSS products. There is no law somewhere that states "The most support you're going to get for an OSS product is newsgroups and developer mailing lists." It just isn't true.
...based on the accurate and educated decisions of a few highly qualified experts...not the collective decisions and changes from anybody, anywhere.
This is not how it works. E.g. Linux - not just anyone can patch the (official) Linux kernel. All modifications have to be approved first, by experts like Alan Cox. In fact, there is nothing stopping you from doing QA on an OSS product. By stark contrast, corporations like Microsoft won't let you analyse the source code to most of their products! Get a clue.
(unless the spammer host was a recipient, in which case the message is processed as normal - this defeats the spammer's check of the effectiveness of the spamming)
Unfortunately, the spammer could have any number of free email accounts on e.g. hotmail, bigfoot, his ISP etc. which you wouldn't know about. Still, it'd be something!
Unfortunately, you can't run Java 100% natively because there are no bytecodes for accessing hardware or anything like that. And I don't think having the garbage collection algorithm executed as bytecodes would work very well, either...
Servlets are not run through CGI. The two are different methods. With CGI you would probably have a separate Java Virtual Machine spawned for every single connection, which is obviously extremely inefficient (hogs memory and CPU, and entails a significant delay for each one to start up). The whole idea of servlets is that you have a single VM that stays continuously running, and all of the servlets run inside that VM (though not necessarily through the same classloader).
A long call stack doesn't necessarily matter in terms of performance, because a good optimising JIT compiler can in principle do lots of inlining so that those calls and returns don't actually happen, they just get inferred from the original.class file if and when a stack trace needs to be printed (which is very rare in the big scheme of things).
Of course, that long call stack might mean it's doing things very inefficiently, or it might mean merely a very fine-grained level of modularity (you're bound to see a lot of methods being traversed that way), which to me speaks of a good, extensible design, not a bad one. Sometimes it's the former, obviously (I've found code which is orders of magnitude too slow, probably because of lack of cacheing of results), otherwise Swing wouldn't be so slow, but sometimes it's the latter.
Sun really haven't invested much effort in optimising Swing yet, they've concentrated on server-side which is hot right now, and they know of course that Moore's Law will do some of the work for them. One big problem is that Swing currently takes almost no advantage of any kind of graphics acceleration. Try switching Windows into vanilla 640x800 VGA mode and see how the graphics noticeably slows down now that it can't use acceleration. I think once they fix this, most of the complaints about speed (on Win32 anyway) will go away.
Or just a 21" monitor, and have the app on one side and the help files on the other? That's the kind of thing windowing systems were originally designed for, after all...
Slashdot Mirror
"Tension high after pepper spray used on protesters" (CNN, 31 Oct 97)
Includes disturbing video footage of the police actually rubbing pepper spray into protestors eyes. Ouch.
The Nazi bastards should have been tried and locked up for torture.
Oh come on
Cars kill a lot more people than spilt coffee. A lot more. Even intelligent people sometimes kill themselves and other people in car accidents.
So should we impose a ban on the sale of all cars?
What about the customer's responsibility to prevent such accidents? Like, don't put a cup of scalding hot coffe between your legs? It's kinda stupid, you gotta admit.
That's how non-GPL'd code such as Netscape can legally call the GPL'd kernel, because they are sufficiently separated.
My question is, what counts as "sufficiently insulated"? Any legal opinion on this?
With the LGPL or MPL, it's pretty clear - if you change any of the original source files, those changes must be covered by the original license - any new source files you create, you can do what you want with, right?
Um, that wouldn't be very sensible. A JIT compiler for Java, for example, doesn't compile source code, it compiles bytecodes into native codes. Since bytecode can be considered an assembly/machine language for an abstract virtual machine, it's effectively an optimising cross-assembler - the word "compiler" in the phrase "just in time compiler" is a bit misleading.
And you know this how? You're the director of the NSA, right?
And you're sure there aren't any top-secret sub-agencies?
Only a few weeks ago the Guardian newspaper revealed that a secret nuclear facility in England had several times been close to a nuclear accident at least as severe as the one in Japan. Democratic governments do hide the truth from people. They do have secret facilities.
Okay, what if, immediately after you give it to your friend, you erase all your own copies, if any?
- Friend uses the gift instead of buying a new copy
- One less copy sold
- Money out of the developer's pocket
Therefore, by analogy, it's still theft... Um, I don't think so. Not even Microsoft thinks so.You need a better analogy to prove your case, because the analogy can also be applied to "prove" that giving away is theft, which is nonsense.
That's an issue of technical support, not open vs closed. You can still buy technical support (and very high quality tech support at that, in some cases) for OSS products. There is no law somewhere that states "The most support you're going to get for an OSS product is newsgroups and developer mailing lists." It just isn't true.
This is not how it works. E.g. Linux - not just anyone can patch the (official) Linux kernel. All modifications have to be approved first, by experts like Alan Cox. In fact, there is nothing stopping you from doing QA on an OSS product. By stark contrast, corporations like Microsoft won't let you analyse the source code to most of their products! Get a clue.
Unfortunately, the spammer could have any number of free email accounts on e.g. hotmail, bigfoot, his ISP etc. which you wouldn't know about. Still, it'd be something!
It certainly is.
paint():
...
Reposition.
Oops, need to paint() again.
Reposition.
Oops, need to paint() again.
Reposition.
Of course, that long call stack might mean it's doing things very inefficiently, or it might mean merely a very fine-grained level of modularity (you're bound to see a lot of methods being traversed that way), which to me speaks of a good, extensible design, not a bad one. Sometimes it's the former, obviously (I've found code which is orders of magnitude too slow, probably because of lack of cacheing of results), otherwise Swing wouldn't be so slow, but sometimes it's the latter.
Sun really haven't invested much effort in optimising Swing yet, they've concentrated on server-side which is hot right now, and they know of course that Moore's Law will do some of the work for them. One big problem is that Swing currently takes almost no advantage of any kind of graphics acceleration. Try switching Windows into vanilla 640x800 VGA mode and see how the graphics noticeably slows down now that it can't use acceleration. I think once they fix this, most of the complaints about speed (on Win32 anyway) will go away.
http://www.darkink.com/~kitsune/
Score 2 Informative anyone?
http://www.lod.org/electrum/spkyburn.jpg