What they didn't mention were a few interesting security fixes from bugs in 2.4.12, probably due to the self-imposed DMCA "gag order."
Actually, it is more likely that Linus just didn't bother to write everything down. His changelogs are usually very brief (but then, he didn't used to write them at all a while ago!)
He does mention updates to I2O and USB subsystems (just not what they were), and the "Alan Cox: more merging" entry can contain almost anything;-)
Will the linux kernel allow a user process to be killed that is blocked in a kernel call? In my experience, Solaris and Tru64 do not: a user program that is blocked in a kernel call will stay blocked until the kernel call returns, regardless of any action (short of rebooting the machine) that a user can take. I assume that there is some well-thought-out reasoning behind this, but sometimes (e.g. during device driver development) I wish it were somehow a configurable behavior.
In most cases, Linux will block like every(?) other *nix kernel. The reason for this is that inside kernel the process may have reserved some locks or resources, and the process itself is the only one which knows how to release them. Also, the process may be in the middle of programming some low level device, and stopping it might leave the device in undefined state. So in the general case, it is impossible to stop a process during system call.
Some system calls which can take longer (for example connecting to a remote TCP/IP socket) are designed so that they can be interrupted at certain points.
You mean, "Intel should have made some half-assed extension to a painfully obsolete ISA rather than have the balls to invent a radically new and innovative ISA that might actually be able to exploit the rediculously large number of transistors that will be on future chips for something other than cache."
Actually, when I first heard about IA-64 (the new ISA for Itanium) I thought it was about time.
...It was probably around four years ago, and Itanium is still not finished...
Also, nobody has any experience of using VLIW (or EPIC, whatever) style ISA on a general purpose processor before, so that will probably cause problems in the beginning. (Ok, Transmeta CPUs are VLIW, but you won't run them natively, and so far their performance haven't been stellar either.)
So just fixing up some of the worst problems of IA-32 ISA might actually be a good idea (another way would have been to design another 64bit ISA and just make better work than Intel has been doing). AA-64 ISA gives us twice the number of general purpose registers (which are twice as big as the old ones) and gives us register based FPU insted that horrible stack based monster we have suffered two decades. (Well, it was probably a good idean when FPU was it's own CPU but Intel should have at least given us a register based alternative to access it when it designed Pentium...)
The bottom line: IA-64 looks good in paper, but after the difficulties in implementation it doesn't look so good anymore. The AA-64 doesn't look that good in paper but doesn't seem to suffer same kinds of problems as IA-64 chips. However, neither is yet officially released (and AA-64 won't for a long time) so it will take some time untill we see which is actually better.
AMD can't get OEM agreements with the big PC manufacturers - did it ever occur to them that it might be because Joe Consumer doesn't ask Gateway or Compaq about the latest Athlon machines?
I got the feeling that AMD is currently selling as much processors as it can manufacture (and they are selling well - Intel reported slowdowns lately but AMD have been increasing both sales and profits). So the very reasonable prices and word-of-mouth are probably enough for now.
(AMD is building more factories but those things take time to do.)
OTOH if you want to sell a CPU which is much more expensive than the competitions CPU (or even your own older models) and which performs worse expect in few special cases - then you need an expensive advertising campain!
Kiitos! Heck, 370.000 FIM, that would kill our family budget thrice! It's not like the guy walked away unpunished.... jeez!
It was a large scale business, worth of several million FIMs so compared to that he got off lightly. (And he definitely deserved it because he cheated his customers to pay full price for a software which didn't have manufacturer support nor upgrade possibility.)
I'm not aware of any online translator from Finnish to English so here is my own very short summary:
A man was selling reselling upgrade, replacement, OEM and other cheaper or free
versions of programs as a full versions (he had made fake packages so
they seemed like full versions).
The Finnish Court of Appeals judged him guilty of fraud, but not to
copyright violations, since he hadn't made illegal copies.
(That fraud come because his customers weren't eligble to normal support
and upgrades to these programs.)
In short, EULAs are not valid in Finland.
Several US based companies including Microsoft tried to get big compensations
both on copyright violations (2400000 FIM) and for legal costs (980000 FIM).
From the fraud, the man had to pay 300000 FIM (I can't say from the article
if that sum went to his customers or the US companies) and for legal costs
the court agreed that 70000 FIM would be the right amount.
I'm surprised you are having so much trouble with Mozilla on a 650MHz Athlon. I run it on a 466MHz Celeron, 64MB RAM, and although it takes a bit more time than Netscape to load, its not unusuable.
Let's just say that currently my top two favourite browsers are w3m (text based) and Opera (which I use at work on W2k). Both of these two are fast and stable, and I like their basic user interface principles. Mozilla, however isn't currently very stable (since I tried to use it last) nor it is particulary fast. Well, it can display pages fast enough, but seem to be rather sluggish when creating new windows (which I use a lot). And to top it all, I'm not quite sure I like the direction they are taking the UI.
Actually, I'm currently using Mozilla 0.6 and it seems somewhat better that the last nightly build I tried. Now if I could only get used to some of the user interface perver^H^H^H^H^H^Hfeatures I might finally start using it insted of NS4.7X as my graphical Linux browser...
If not having to pay for an operating system is important, why would you pay for software that is already available for operating systems you can pay for?
Because the reason you are using a free operating system is because it is best for the purpose, not because it is free?
I use Linux at home and so far free (as in beer) software has been enough for my purposes. I will buy Opera if they get it stable before Mozilla is usable on 650MHz Athlon;-). I would also buy games if they would release something I want to play (Alpha Centauri could be the first if they would release it).
However, at work we are just moving from MySQL to DB2 on our Linux server. MySQL is not up to the task, and it is much cheaper (not to mention faster) to purchase a license of DB2 than use the engineering hours either to tweak MySQL or our application to work around the limitations. (And then possibly have it fail when our customers are using it thus earning us bad reputation.)
I still wouldn't install DB2 for my own computer, even when the personal edition is downloadable for free.
If Linus has a problem with ReiserFS, it is that he probably fears some of the exciting new ``disruptive'' concepts that Hans Reiser has planned. ReiserFS is truly innovative.
IIRC from the time the ReiserFS inclusion was discussed in the kernel list (last spring, I think) Linus said that he would like to see ReiserFS included in 2.4 (he seriously doubted it would go to 2.4.0, but some 2.4.X anyway), however there were few problems, at least
In some cases, software RAID and ReiserFS conflicted and caused hangs.
NFS v2 and ReiserFS had trouble together in few strange cases (this is mostly because certain NFS v2 perversions rather than ReiserFS, but NFS v2 clients are still widely used)
ReiserFS was still in state of rapid development, and that kind of modules aren't usually accepted in the kernel (for example RAID, ISDN, PCMCIA and devfs lived quite a long time in separate patches, too).
ReiserFS port for 2.3 kernel was relatively recent, and it hadn't the testing nor the user base of 2.2 version.
When all this is added to the fact that (yet again, IIRC) Linus had already announced feature freese, I don't think it's the new concepts Linus has problems with. (Unless you mean that "No, we are not going to muck up the whole VFS during the feature freeze.")
Since then most (all?) problems have been solved, so it is possibile that ReiserFS will go in some 2.4.X kernel. (I certainly hope so - even when I managed to encounter one of the file eating bugs found in it, I quite liked the file system.)
Metadata-only journalling is useful in situations where good write performance is required, since journaled filesystems by definition write everything twice, once to the journal and then finally to the location on disk where the data belongs.
Actually, you don't need to write journaled data twice - you can just write data to emptly blocks and after that change the metadata so that it points to the new blocks. This way you write as many block as with metadata only -journaling.
Of course this will lead to fragmentation problems, so filesystem using this kind of journaling would probably need to do some kind of defragmenting in the background.
Isn't there a way to have some combination of encryption and compression so that a file has two keys, one returns benign data and
the other returns the real contents?
Using two different steganographic algorythms (or at least keys) in a same file? You would only need to choose the bits to hide the information so that they won't overlap.
An oversimplified example would be that if you hide information in an (uncompressed).wav file, you store information A to the lowest bit of the left channel and information B to the lowest bit of the right channel.
For this kind of operation you should propably choose a big file with plenty of 'insignificant' bits to choose (like MPEG video) so it would be easy to explain why you didn't use every possible bit to store data.
Consider our own chances of being detected. We've sent out our own deliberate markers,
but we've put out a much larger number of transmissions that were never intended for
communication with aliens, and a much greater variety of them, too.
The problem with unintentional singnals is that many of them will be impossible to detect (by any means we know of) in just a few lightyears (because of a backgrould noise in space), and it gets harder as technology advances more. There is two reasons for this: compression and directional transmissions.
First compression (and possible encryption) will make signal itself look more ol less like random noise and then the transmissions themselves are usually directional so they can't be detected expect from relatively small area (and use lower power anyway). Then there is already radio transmitters which use several different wawelengths to transmit the singnal (most current uses are military, I think, but civilian applications are already coming). To sum it all: the unintentional signals we are now putting out are both harder to detect and weaker in the first place. And continue getting more so.
(And to make things really challenging, quite a many frequencies get absorbed or reflected in the athmosphere and couldn't be detected even from geosyncronous orbit. I think that I read somewhere that we would be able to detect our own level of civilization only from something under 100 lightyears unless it delibitely put out a signal to our direction.)
This is the second time the EU has had something to say about the merging of two US companies.
The US been messing with affairs of other countries for centuries. And that includes it's supposed allies in Europe.
It would be easy to find few dozens examples in economical, political and military areas within last few years in case you are interested. (Good starting points would be searching for gene manipulated food, individual privacy and Bosnian crisis related issues, preferably from European sources.)
As a personal opinion, it seems that many politicians think that one of the most important reasons for strenghtening EU is guarding it againts outside bullies. Of which US is currently by far the worst.
That gives you the odds of an asteriod hitting you tommorow of 1 in 365,000,000. I'd buy a lottery ticket long before losing much sleep.
Actually, as someone else already pointed out, the change to be killed in an asteroid collision is pretty high compared to many other ways of dying. This is because even moderate hit can kill millions and the largest ones could kill most or all of the human race!
This is also a reason we should prepare for the possibility: There are quite a many rocks out there capable of destroying our civilization if not human as a species. Even while the probability of that happening next 100 years is low, if we miss one we might not get another change. Ever.
As for Tom's objectivity and bias, I stand by my original statement. I did not say he was incorrect, just that he has repeatedly been very forgiving of AMD and very quick to bash Intel.
I do remember Tom bashing AMD at least on Irongate problems. (The AGP problems and super bypass 'upgrade'.)
IMO Tom hasn't been bashing anyone without reason, but neither he has been overly nice to anyone. But looking back, most (all?) of his claims and predictions have turned out to be true.
And considering latest Intel moves, I can understand if he doesn't excactly like them (or does anyone here believe Intel RAMBUS probaganda? Not even Intel does anymore;-)
You're forgeting something... if you release unstable software to the public, you could always distribute a patch later to fix the problems. This is not true of CPU's..
Actually, Since Pentium Pro (I think) it has been possible to upgrade the microcode of Intel processors. These upgrades can fix some problems.
These upgrades are usually part of the BIOS so re-flashing your BIOS might also fix your CPU!;-)
I wrote a graphical editor in Java. On a 1024x768 display (modest for the times), the performance penalty of Swing made some parts of it barely usable on a dual P3-550 Xeon box with a TNT2.
Actually, in terms of swing speed, there are drastic differerences between different JVMs.
When I was working on a simple graph editor (bitmap background, boxes with text on foreground) I noticed that Sun JDK 1.2.2 run on NT was clearly slower than IBM JDK 1.1.8 run in Linux but used through X server on the same NT! (The network between NT and Linux was 10MB ethernet.)
After investigating a little it seems that Sun JVM rendered the image in core memory and just dumped everything to graphics card when needed. OTOH IBM JVM drew the picture 'right' so the graphics card accelerator was used. (Sun JVM on Linux/X was unusably slow, IBM JVM on NT was the fastes combination.)
Please note these observations probably won't reflect performance of current 1.3 JVMs.
It's called a VM. If you knew how Windows worked, you wouldn't say stupid stuff like that. Command.com (the DOS prompt) is (and always HAS) run under a virtual machine in Windows 95.
I believe that is true for NT. But at least Win 95 used to run more on top of DOS more than use it only as a loader. There are some evidence for that, like that DOS protected mode memory managers (EMM386 or third party like QEMM) are needed for loading Win 95, and I have seen few times Win 95 crash to a dos prompt! OTOH while Win 98 still needs EMM386 (I haven't tried QEMM on it) I have always got the comfy BSOD when it crashes so maybe there is less DOS in it.
This will never happen without a major philosophy change at MS.
Actually, it is entirely possible for MS to port Office to Linux. This is because MS currently have two cash cows: Windows (all variations) and Office.
If Linux would threaten to surpass Windows as a desktop OS and there would not be Office for it, MS would lose both. OTOH if MS released Office for Linux it could probably keep much of the market share it currently has in office applications (at least for a while).
Of course, MS wouldn't release Office for Linux unless it had to because it would significantly help Linux on Desktop market.
If people really want a Linux DVD player they could work with a licensed vendor to create Linux DVD software.
There have been several attempts to get a vendor (at least Creative Labs and Sigma Design) to make a player for Linux, as well as get (parts of) DVD documentation licensed. So far without success...
(Well, SD said it would do a Linux player for their next DVD card, and Creative has Dxr2 DVD player on their opensource page, but AFAIK Creative didn't help with it. And I don't know how functional it is at the moment.)
Slashdot Mirror
Actually, it is more likely that Linus just didn't bother to write everything down. His changelogs are usually very brief (but then, he didn't used to write them at all a while ago!)
He does mention updates to I2O and USB subsystems (just not what they were), and the "Alan Cox: more merging" entry can contain almost anything ;-)
In most cases, Linux will block like every(?) other *nix kernel. The reason for this is that inside kernel the process may have reserved some locks or resources, and the process itself is the only one which knows how to release them. Also, the process may be in the middle of programming some low level device, and stopping it might leave the device in undefined state. So in the general case, it is impossible to stop a process during system call.
Some system calls which can take longer (for example connecting to a remote TCP/IP socket) are designed so that they can be interrupted at certain points.
Actually, when I first heard about IA-64 (the new ISA for Itanium) I thought it was about time.
Also, nobody has any experience of using VLIW (or EPIC, whatever) style ISA on a general purpose processor before, so that will probably cause problems in the beginning. (Ok, Transmeta CPUs are VLIW, but you won't run them natively, and so far their performance haven't been stellar either.)
So just fixing up some of the worst problems of IA-32 ISA might actually be a good idea (another way would have been to design another 64bit ISA and just make better work than Intel has been doing). AA-64 ISA gives us twice the number of general purpose registers (which are twice as big as the old ones) and gives us register based FPU insted that horrible stack based monster we have suffered two decades. (Well, it was probably a good idean when FPU was it's own CPU but Intel should have at least given us a register based alternative to access it when it designed Pentium...)
The bottom line: IA-64 looks good in paper, but after the difficulties in implementation it doesn't look so good anymore. The AA-64 doesn't look that good in paper but doesn't seem to suffer same kinds of problems as IA-64 chips. However, neither is yet officially released (and AA-64 won't for a long time) so it will take some time untill we see which is actually better.
I got the feeling that AMD is currently selling as much processors as it can manufacture (and they are selling well - Intel reported slowdowns lately but AMD have been increasing both sales and profits). So the very reasonable prices and word-of-mouth are probably enough for now. (AMD is building more factories but those things take time to do.)
OTOH if you want to sell a CPU which is much more expensive than the competitions CPU (or even your own older models) and which performs worse expect in few special cases - then you need an expensive advertising campain!
It was a large scale business, worth of several million FIMs so compared to that he got off lightly. (And he definitely deserved it because he cheated his customers to pay full price for a software which didn't have manufacturer support nor upgrade possibility.)
I'm not aware of any online translator from Finnish to English so here is my own very short summary:
A man was selling reselling upgrade, replacement, OEM and other cheaper or free versions of programs as a full versions (he had made fake packages so they seemed like full versions).
The Finnish Court of Appeals judged him guilty of fraud, but not to copyright violations, since he hadn't made illegal copies. (That fraud come because his customers weren't eligble to normal support and upgrades to these programs.)
In short, EULAs are not valid in Finland.
Several US based companies including Microsoft tried to get big compensations both on copyright violations (2400000 FIM) and for legal costs (980000 FIM). From the fraud, the man had to pay 300000 FIM (I can't say from the article if that sum went to his customers or the US companies) and for legal costs the court agreed that 70000 FIM would be the right amount.
(I hope I didn't make too many mistakes.)
Actually, I'm currently using Mozilla 0.6 and it seems somewhat better that the last nightly build I tried. Now if I could only get used to some of the user interface perver^H^H^H^H^H^Hfeatures I might finally start using it insted of NS4.7X as my graphical Linux browser...
I use Linux at home and so far free (as in beer) software has been enough for my purposes. I will buy Opera if they get it stable before Mozilla is usable on 650MHz Athlon ;-). I would also buy games if they would release something I want to play (Alpha Centauri could be the first if they would release it).
However, at work we are just moving from MySQL to DB2 on our Linux server. MySQL is not up to the task, and it is much cheaper (not to mention faster) to purchase a license of DB2 than use the engineering hours either to tweak MySQL or our application to work around the limitations. (And then possibly have it fail when our customers are using it thus earning us bad reputation.)
I still wouldn't install DB2 for my own computer, even when the personal edition is downloadable for free.
(But everyone of us is using ad filters anyway?)
IIRC from the time the ReiserFS inclusion was discussed in the kernel list (last spring, I think) Linus said that he would like to see ReiserFS included in 2.4 (he seriously doubted it would go to 2.4.0, but some 2.4.X anyway), however there were few problems, at least
- In some cases, software RAID and ReiserFS conflicted and caused hangs.
- NFS v2 and ReiserFS had trouble together in few strange cases (this is mostly because certain NFS v2 perversions rather than ReiserFS, but NFS v2 clients are still widely used)
- ReiserFS was still in state of rapid development, and that kind of modules aren't usually accepted in the kernel (for example RAID, ISDN, PCMCIA and devfs lived quite a long time in separate patches, too).
- ReiserFS port for 2.3 kernel was relatively recent, and it hadn't the testing nor the user base of 2.2 version.
When all this is added to the fact that (yet again, IIRC) Linus had already announced feature freese, I don't think it's the new concepts Linus has problems with. (Unless you mean that "No, we are not going to muck up the whole VFS during the feature freeze.")Since then most (all?) problems have been solved, so it is possibile that ReiserFS will go in some 2.4.X kernel. (I certainly hope so - even when I managed to encounter one of the file eating bugs found in it, I quite liked the file system.)
Actually, you don't need to write journaled data twice - you can just write data to emptly blocks and after that change the metadata so that it points to the new blocks. This way you write as many block as with metadata only -journaling.
Of course this will lead to fragmentation problems, so filesystem using this kind of journaling would probably need to do some kind of defragmenting in the background.
An oversimplified example would be that if you hide information in an (uncompressed) .wav file, you store information A to the lowest bit of the left channel and information B to the lowest bit of the right channel.
For this kind of operation you should propably choose a big file with plenty of 'insignificant' bits to choose (like MPEG video) so it would be easy to explain why you didn't use every possible bit to store data.
The problem with unintentional singnals is that many of them will be impossible to detect (by any means we know of) in just a few lightyears (because of a backgrould noise in space), and it gets harder as technology advances more. There is two reasons for this: compression and directional transmissions.
First compression (and possible encryption) will make signal itself look more ol less like random noise and then the transmissions themselves are usually directional so they can't be detected expect from relatively small area (and use lower power anyway). Then there is already radio transmitters which use several different wawelengths to transmit the singnal (most current uses are military, I think, but civilian applications are already coming). To sum it all: the unintentional signals we are now putting out are both harder to detect and weaker in the first place. And continue getting more so.
(And to make things really challenging, quite a many frequencies get absorbed or reflected in the athmosphere and couldn't be detected even from geosyncronous orbit. I think that I read somewhere that we would be able to detect our own level of civilization only from something under 100 lightyears unless it delibitely put out a signal to our direction.)
The US been messing with affairs of other countries for centuries. And that includes it's supposed allies in Europe. It would be easy to find few dozens examples in economical, political and military areas within last few years in case you are interested. (Good starting points would be searching for gene manipulated food, individual privacy and Bosnian crisis related issues, preferably from European sources.)
As a personal opinion, it seems that many politicians think that one of the most important reasons for strenghtening EU is guarding it againts outside bullies. Of which US is currently by far the worst.
Actually, as someone else already pointed out, the change to be killed in an asteroid collision is pretty high compared to many other ways of dying. This is because even moderate hit can kill millions and the largest ones could kill most or all of the human race!
This is also a reason we should prepare for the possibility: There are quite a many rocks out there capable of destroying our civilization if not human as a species. Even while the probability of that happening next 100 years is low, if we miss one we might not get another change. Ever.
I do remember Tom bashing AMD at least on Irongate problems. (The AGP problems and super bypass 'upgrade'.)
IMO Tom hasn't been bashing anyone without reason, but neither he has been overly nice to anyone. But looking back, most (all?) of his claims and predictions have turned out to be true.
And considering latest Intel moves, I can understand if he doesn't excactly like them (or does anyone here believe Intel RAMBUS probaganda? Not even Intel does anymore ;-)
Actually, Since Pentium Pro (I think) it has been possible to upgrade the microcode of Intel processors. These upgrades can fix some problems.
These upgrades are usually part of the BIOS so re-flashing your BIOS might also fix your CPU! ;-)
Actually, in terms of swing speed, there are drastic differerences between different JVMs.
When I was working on a simple graph editor (bitmap background, boxes with text on foreground) I noticed that Sun JDK 1.2.2 run on NT was clearly slower than IBM JDK 1.1.8 run in Linux but used through X server on the same NT! (The network between NT and Linux was 10MB ethernet.)
After investigating a little it seems that Sun JVM rendered the image in core memory and just dumped everything to graphics card when needed. OTOH IBM JVM drew the picture 'right' so the graphics card accelerator was used. (Sun JVM on Linux/X was unusably slow, IBM JVM on NT was the fastes combination.)
Please note these observations probably won't reflect performance of current 1.3 JVMs.
I believe that is true for NT. But at least Win 95 used to run more on top of DOS more than use it only as a loader. There are some evidence for that, like that DOS protected mode memory managers (EMM386 or third party like QEMM) are needed for loading Win 95, and I have seen few times Win 95 crash to a dos prompt! OTOH while Win 98 still needs EMM386 (I haven't tried QEMM on it) I have always got the comfy BSOD when it crashes so maybe there is less DOS in it.
If Linux would threaten to surpass Windows as a desktop OS and there would not be Office for it, MS would lose both. OTOH if MS released Office for Linux it could probably keep much of the market share it currently has in office applications (at least for a while).
Of course, MS wouldn't release Office for Linux unless it had to because it would significantly help Linux on Desktop market.
There have been several attempts to get a vendor (at least Creative Labs and Sigma Design) to make a player for Linux, as well as get (parts of) DVD documentation licensed. So far without success...
(Well, SD said it would do a Linux player for their next DVD card, and Creative has Dxr2 DVD player on their opensource page, but AFAIK Creative didn't help with it. And I don't know how functional it is at the moment.)