NIS is simple and easy to maintain. LDAP is harder. From memory (10 years ago) Kerberos was geared towards as single user on a single machine, is that still the case?
Lots of big organizations still use NIS because its flaws, while real, are well understood.
Actually most Europeans armies are already or moving to all-volunteers now, it's only a recent development though, since the end of the cold war. Before that there was a very serious possibility that Europe could be invaded by Warsaw Pact countries, which all had huge conscript armies. The remote possibility of land invasion of the US made the systematic conscription unnecessary there.
In case of land invasion you really want the cannon fodder because the threat is widespread and a volunteer-only army, while better trained, is never big enough.
So it's not a question of `catching up to the fact' than a change of circumstances.
BTW I hadn't realized that draftees were less-than-useless, or even more so than volunteers in Vietnam. What about in Israel, are they useless there too? Are you just trying to be offensive or am I too sensitive?
Every democratic country I know has some sort of deal for they citizen that basically say they can be drafted at any moment. BTW this holds for the US too.
There is a big difference between going to the Army because you are forced to or going there because you are interested.
Most people I know who did their compulsery military service ended up loathing the army with a vengeance. The usual measure of discipline was how tight you made your bed in the morning (at 6:00 am) and camaraderie was all about how fast you could open bottles of beer with your teeth. Believe it or not a lot of people do not want to live in that way.
BTW I don't like your sentence `most people do a military service unless they are cowards'. There are other reasons not to want to do a military service, such as hating discipline and the repression of independent thought. In fact doing one's military service is just going with the pack. Not doing it requires thought and action.
I wouldn't call that re-invention, they still sell languages (C# being the latest), OSes (Windows in all its implementations), office suites and hardware.
They've been accreting businesses, not shedding any as they went. That's the problem. From a nice player now they cover all aspects of the dominant computing platform of today. Will they keep that position? if they do, what will they need to do?
Microsoft knows very well how to squash traditional competition. They have made a few unsuccessful moves against OSS, they are not giving up yet, that's what the whole DRM stuff is all about. The smart money is still on Microsoft.
1- Put cd in drive 2- open file on CD with any program 3- push the eject button on the cd drive.
What happens: nothing. The CD is stuck, you need to close whichever program opened the file on the CD first. Not obvious.
How do you find out about which programs have opened files on the CD? Read the lsof man page, good luck.
Automount doesn't help unmounting that much, you need to understand what is going on. Distros could make it easier with a GUI that would run lsof if unmout failed and tell you which program you need to close first, but none that I know do that at present.
Hi point may have been that standard video sequences used in motion research (e.g. the taxi sequence, the garden sequence, the rubik cube sequence, etc) are not available uncompressed. He's looked hard and the only way to get an uncompressed sequence is to make it yourself, precisely what he wrote.
Overall you are right, the US did try to contribute earlier in WWII though.
Actually F.D. Roosevelt wanted to get involved in WWII much earlier and much more, but he US opinion was against it (look up `America First' and the political involvement of Charles Lindberg). FDR did help the allies a great deal before entering the conflict via export laws that were neutral-sounding but in fact favoured the UK much more, such as `cash and carry', which ensured the UK was much less likely to get low on matériel.
Some people say FDR knew of the impending attack on Pearl Harbor but let it happen (shipping away the aircraft carriers and leaving obsolete ships at shore only), so that the US could declare war on Japan and therefore against Germany too.
Sorry about the lengths of my replies, this is my bane. I'll try and keep it short.
I think you are right in saying that Linux has already reached pretty much everybody willing to give it a try and that it's not ready for grandma.
The next level is hard. Apple has thrown huge resources at MacOS/X and so has Microsoft at win2k and XP, in terms of usability. I'm not sure the OSS community has the patience to develop great consistent GUIs. I'm impressed by KDE, but most developers write small things, and even medium things like kontour show signs of inconsistencies.
The GIMP is a hugely ambitious project, but it's interface is dreadful, very unintuitive. Yet I'm sure their developers think it's OK.
It's not a matter of choosing one GUI platform and one widget set, it's the way the developers use them.
First of all they tell you how to fetch, recompile and configure a whole lot of standard packages. Want to enhance your RH9 distribution with ALSA? head for these guys' documentation and you'll be home and dry in no time.
Second of all this documentation covers a lot of things in a very tight format. Want to configure an anonymous CVS server? there you go.
Finally this effort frees the whole community from the grips of the distribution vendors. If all goes to hell in a handbasket and your favourite distro turns its back on you, you can still survive, fix, patch and generally maintain what you've got, or start from scratch.
Are you saying (in your sig) that all integers should really be handled as floats (or doubles)? There is a reason they are not, you know?
For the unitiated
float x = 1/2;
x will be 0.0, because 1 and 2 are both integers, therefore 1/2 is an integer division and its results is the integer 0, which is then converted into a float.
If you really wanted a float division you should write
This is very interesting, what is the nature of the difference that you hear? Can you hear distorsion, masking, missing frequencies, other form of artifacts? Can you describe them in words?
I can never hear any difference between an mp3 above 128kbit and the unencoded stream. I thought I had good ears, I used to work as a sound engineer. With audiophile equipment (not mine) I can hear a difference between different audio cables quality. But mp3 encoding artifacts, never for sure.
Can you detect an mp3 encoding (> 128kbits) without reference to the original? Have you tried to do a double blind experiment like so:
- find a recording that you think would be suitable but don't know already. - Encode it as an Ogg and mp3. - generate random names for them and play them in random order, several times. - write for yourself which one do you think are raw and which are encoded (same equipment, etc) - compare with truth.
A German magazine did the equivalent of that some years ago and concluded that a small but significant proportion of audiophiles *could* detect encoding at 128kbit on absolute top notch equipment, but that none of them could detect the encoding at 256kbit, compared with CD quality (i.e. their answer were indistinguishable from random).
To me this is conclusive proof that above 128kbit mp3 are perfectly fine.
I've never tried BeOS, so I might be completely wrong. Remember that there is the perception of speed and speed itself. Going back to my Neverwinter experience, every user that has tried it and reported say that the framerate is better under Linux, by about 10% (enough to be noticeable). However when I tried it, I found that the Windows experience was maybe a bit smoother. The framerate said otherwise, and in fact the Windows version crashed on me! (it never did under Linux). I'm at a loss as to why I thought the Windows version was smoother. I'm thinking maybe the *sound* was better (and it is, EAX is not supported under Linux). My point is it is very hard to judge video speed objectively.
With NVidia, everything that can possibly be is already in the kernel. The rest communicates with it via shared memory, there is no slowdown there; I'm not even sure there are missing optimization. On the main console the X protocol does not go through the network layers, it uses Unix domain sockets, which are zero copy with newer kernels.
Obviously if you wrote everything to talk directly to the frame buffer it would be faster, but that would be unwise for such mundane things as redrawing windows and painting text, because you would throw away all that flexibility for very little gain. When you really need the speed (video overlay) the direct access is available already, (DGA, XV, in true Unix fashion there is more than one way to do it).
Yes there are many layers, possibly one or two of which could be removed for some tangible gain, but not as much as you think. For example try the following experience: watch a DVD movie with mplayer, forcing it to work through the old antiquated X11 shared memory interface, and with the newer XV. You will find that the X11-SHM interface was already pretty much up to the job. It was already fast enough on a P-III 500MHz with an old Radeon 7000 (the entry level card), on a newer P-IV with a GF4 it's even better.
The point is all this direct hardware access is *not* necessary. It will make your life harder, not easier. All these layers add flexibility and reliability, because they can be debugged independently and they are individually simpler.
Under Unix, X11 is the standard, and it gives you choice.
> The bulk of users are still paying Microsoft > $100 every two years and sticking with Windows. > We did something wrong somewhere.
Linux is not ideal but I've been happy for it to be my main desktop ever since I've had a computer good enough to run it. People don't come to Linux because it's not good enough -- remember that according to Google Zeitgeist the most popular O/S is still win98, yet win2k and winXP are heaps better without requiring any retraining.
The main problem is inertia and perception. Trying Linux is hard for a non-computer person. Most user don't do a great deal with their computers already (read mail, browse web, write letter to grandma). Linux can do all of that capably enough and certainly better than win98, but Linux does not come pre-installed and requires a different way of working.
You and I are flexible enough because we are interested. Most people aren't. That's fine with me, I'm not sure Linux needs to reach all people and be the be all and end all for everybody. Windows certainly isn't.
So in conclusion, yes you are right Linux could be more optimized on the desktop, probably at some cost. This is not what will make or break Linux however.
If X is so bad and bloated, explain to my why I'm getting better framerate on Neverwinter Nights under X11/Linux than under the slim, fast and optimized Windows system, with my puny GF4-MX?
I honestly think that if you did all that you talk about you wouldn't gain a thing speed wise. I agree however that the windows manager layout introduces inefficiencies and inconsistencies, but you do get flexibility.
As for the nice, simple and fast MacOS/X display, I've tried it several times and I like X11 better. More choice, more flexibility, and yes, more stability (although the zoom feature is neat).
It's impossible to defend the Gulag and the Soviet policies, I'm not attempting to do that. The parent poster was saying that clearly the US was the good guys and the USSR the bad guys. It's a bit more subtle than that. David Bohm's career was pretty much ruined for no other reasons that he had opinions that did not agree with some committee's. I have no doubt that had McCarthy been given the power he would have put up some version of a Gulag too. So yes there is a very significant difference in execution, but not in intent -- both reek of striking injustice.
You can somehow expect a totalitarian government to behave very badly and shut people off in a very constraining manner. BTW some governments did a lot worse than USSR. Pol Pot anyone? You can also expect a functionning democracy to work better and not kill people off randomly.
That does not make magically all people in the US morally better than in USSR. The system is better, no serious person contests that. Evil lurks everywhere.
In the US too scientists were condemned for holding supposedly communist views. The one that springs to mind is quantum physicist David Bohm. During the McCarthy era he refused to testify, and was stripped of his US citizenship. His position at Princeton was revoked and he was forced into expatriation.
So it's not quite as bad as sending people to the Gulag but it's not altogether nice, wouldn't you say?
However, the key argument (I find) against this sort of argument is that the justice system is not impartial and make a lot of mistakes. One of the reasons is the jury system, where ordinary people don't understand scientific evidence given to them and tend to trust eyewitnesses above everything else [the weakest form of evidence in my view].
A significant number of innocent people have been put to death. It could be you one day, think about it.
Slashdot Mirror
It's not all rosy like `use LDAP'
NIS is simple and easy to maintain. LDAP is harder. From memory (10 years ago) Kerberos was geared towards as single user on a single machine, is that still the case?
Lots of big organizations still use NIS because its flaws, while real, are well understood.
Actually most Europeans armies are already or moving to all-volunteers now, it's only a recent development though, since the end of the cold war. Before that there was a very serious possibility that Europe could be invaded by Warsaw Pact countries, which all had huge conscript armies. The remote possibility of land invasion of the US made the systematic conscription unnecessary there.
In case of land invasion you really want the cannon fodder because the threat is widespread and a volunteer-only army, while better trained, is never big enough.
So it's not a question of `catching up to the fact' than a change of circumstances.
BTW I hadn't realized that draftees were less-than-useless, or even more so than volunteers in Vietnam. What about in Israel, are they useless there too? Are you just trying to be offensive or am I too sensitive?
Every democratic country I know has some sort of deal for they citizen that basically say they can be drafted at any moment. BTW this holds for the US too.
There is a big difference between going to the Army because you are forced to or going there because you are interested.
Most people I know who did their compulsery military service ended up loathing the army with a vengeance. The usual measure of discipline was how tight you made your bed in the morning (at 6:00 am) and camaraderie was all about how fast you could open bottles of beer with your teeth. Believe it or not a lot of people do not want to live in that way.
BTW I don't like your sentence `most people do a military service unless they are cowards'. There are other reasons not to want to do a military service, such as hating discipline and the repression of independent thought. In fact doing one's military service is just going with the pack. Not doing it requires thought and action.
I wouldn't call that re-invention, they still sell languages (C# being the latest), OSes (Windows in all its implementations), office suites and hardware.
They've been accreting businesses, not shedding any as they went. That's the problem. From a nice player now they cover all aspects of the dominant computing platform of today. Will they keep that position? if they do, what will they need to do?
Microsoft knows very well how to squash traditional competition. They have made a few unsuccessful moves against OSS, they are not giving up yet, that's what the whole DRM stuff is all about. The smart money is still on Microsoft.
Cheers.
Try the following:
1- Put cd in drive
2- open file on CD with any program
3- push the eject button on the cd drive.
What happens: nothing. The CD is stuck, you need to close whichever program opened the file on the CD first. Not obvious.
How do you find out about which programs have opened files on the CD? Read the lsof man page, good luck.
Automount doesn't help unmounting that much, you need to understand what is going on. Distros could make it easier with a GUI that would run lsof if unmout failed and tell you which program you need to close first, but none that I know do that at present.
Cheers.
Oh yesss, like this new shiny GeForce FX or that great Radeon 9700 (both supported in Linux BTW). Where is your bargain bin?
FYI Linux on Opteron already works 64-bit. Where is your Windows version that does that?
Driving a car is never trivial.
Hi point may have been that standard video sequences used in motion research (e.g. the taxi sequence, the garden sequence, the rubik cube sequence, etc) are not available uncompressed. He's looked hard and the only way to get an uncompressed sequence is to make it yourself, precisely what he wrote.
I was under the impression that Bink was not available under Linux. A quick google returns nothing of interest. Do you know any different?
Overall you are right, the US did try to contribute earlier in WWII though.
Actually F.D. Roosevelt wanted to get involved in WWII much earlier and much more, but he US opinion was against it (look up `America First' and the political involvement of Charles Lindberg). FDR did help the allies a great deal before entering the conflict via export laws that were neutral-sounding but in fact favoured the UK much more, such as `cash and carry', which ensured the UK was much less likely to get low on matériel.
Some people say FDR knew of the impending attack on Pearl Harbor but let it happen (shipping away the aircraft carriers and leaving obsolete ships at shore only), so that the US could declare war on Japan and therefore against Germany too.
Speaking as a parent, well said! Wish I could mod you up.
Cheers.
Sorry about the lengths of my replies, this is my bane. I'll try and keep it short.
I think you are right in saying that Linux has already reached pretty much everybody willing to give it a try and that it's not ready for grandma.
The next level is hard. Apple has thrown huge resources at MacOS/X and so has Microsoft at win2k and XP, in terms of usability. I'm not sure the OSS community has the patience to develop great consistent GUIs. I'm impressed by KDE, but most developers write small things, and even medium things like kontour show signs of inconsistencies.
The GIMP is a hugely ambitious project, but it's interface is dreadful, very unintuitive. Yet I'm sure their developers think it's OK.
It's not a matter of choosing one GUI platform and one widget set, it's the way the developers use them.
Thanks for that, I wish I could mod you up. Cheers.
OK, many thanks for that reply. I really would like to convince myself that mp3 compression artifacts can be heard.
By rights they should be but I've never experienced it myself. I love Bach so I'll try your reference recording, thanks again!
Of course there is, what are talking about? Whip out this Beofulf cluster of yours, it's ideal for the task.
There are huge benefits from this effort.
First of all they tell you how to fetch, recompile and configure a whole lot of standard packages. Want to enhance your RH9 distribution with ALSA? head for these guys' documentation and you'll be home and dry in no time.
Second of all this documentation covers a lot of things in a very tight format. Want to configure an anonymous CVS server? there you go.
Finally this effort frees the whole community from the grips of the distribution vendors. If all goes to hell in a handbasket and your favourite distro turns its back on you, you can still survive, fix, patch and generally maintain what you've got, or start from scratch.
Thanks B/LFS!
Are you saying (in your sig) that all integers should really be handled as floats (or doubles)? There is a reason they are not, you know?
For the unitiated
float x = 1/2;
x will be 0.0, because 1 and 2 are both integers, therefore 1/2 is an integer division and its results is the integer 0, which is then converted into a float.
If you really wanted a float division you should write
float x = 1.0/2;
This is very interesting, what is the nature of the difference that you hear? Can you hear distorsion, masking, missing frequencies, other form of artifacts? Can you describe them in words?
I can never hear any difference between an mp3 above 128kbit and the unencoded stream. I thought I had good ears, I used to work as a sound engineer. With audiophile equipment (not mine) I can hear a difference between different audio cables quality. But mp3 encoding artifacts, never for sure.
Can you detect an mp3 encoding (> 128kbits) without reference to the original? Have you tried to do a double blind experiment like so:
- find a recording that you think would be suitable but don't know already.
- Encode it as an Ogg and mp3.
- generate random names for them and play them in random order, several times.
- write for yourself which one do you think are raw and which are encoded (same equipment, etc)
- compare with truth.
A German magazine did the equivalent of that some years ago and concluded that a small but significant proportion of audiophiles *could* detect encoding at 128kbit on absolute top notch equipment, but that none of them could detect the encoding at 256kbit, compared with CD quality (i.e. their answer were indistinguishable from random).
To me this is conclusive proof that above 128kbit mp3 are perfectly fine.
Hi, thanks for the long response,
I've never tried BeOS, so I might be completely wrong. Remember that there is the perception of speed and speed itself. Going back to my Neverwinter experience, every user that has tried it and reported say that the framerate is better under Linux, by about 10% (enough to be noticeable). However when I tried it, I found that the Windows experience was maybe a bit smoother. The framerate said otherwise, and in fact the Windows version crashed on me! (it never did under Linux). I'm at a loss as to why I thought the Windows version was smoother. I'm thinking maybe the *sound* was better (and it is, EAX is not supported under Linux). My point is it is very hard to judge video speed objectively.
With NVidia, everything that can possibly be is already in the kernel. The rest communicates with it via shared memory, there is no slowdown there; I'm not even sure there are missing optimization. On the main console the X protocol does not go through the network layers, it uses Unix domain sockets, which are zero copy with newer kernels.
Obviously if you wrote everything to talk directly to the frame buffer it would be faster, but that would be unwise for such mundane things as redrawing windows and painting text, because you would throw away all that flexibility for very little gain. When you really need the speed (video overlay) the direct access is available already, (DGA, XV, in true Unix fashion there is more than one way to do it).
Yes there are many layers, possibly one or two of which could be removed for some tangible gain, but not as much as you think. For example try the following experience: watch a DVD movie with mplayer, forcing it to work through the old antiquated X11 shared memory interface, and with the newer XV. You will find that the X11-SHM interface was already pretty much up to the job. It was already fast enough on a P-III 500MHz with an old Radeon 7000 (the entry level card), on a newer P-IV with a GF4 it's even better.
The point is all this direct hardware access is *not* necessary. It will make your life harder, not easier. All these layers add flexibility and reliability, because they can be debugged independently and they are individually simpler.
Under Unix, X11 is the standard, and it gives you choice.
> The bulk of users are still paying Microsoft
> $100 every two years and sticking with Windows.
> We did something wrong somewhere.
Linux is not ideal but I've been happy for it to be my main desktop ever since I've had a computer good enough to run it. People don't come to Linux because it's not good enough -- remember that according to Google Zeitgeist the most popular O/S is still win98, yet win2k and winXP are heaps better without requiring any retraining.
The main problem is inertia and perception. Trying Linux is hard for a non-computer person. Most user don't do a great deal with their computers already (read mail, browse web, write letter to grandma). Linux can do all of that capably enough and certainly better than win98, but Linux does not come pre-installed and requires a different way of working.
You and I are flexible enough because we are interested. Most people aren't. That's fine with me, I'm not sure Linux needs to reach all people and be the be all and end all for everybody. Windows certainly isn't.
So in conclusion, yes you are right Linux could be more optimized on the desktop, probably at some cost. This is not what will make or break Linux however.
Cheers
If X is so bad and bloated, explain to my why I'm getting better framerate on Neverwinter Nights under X11/Linux than under the slim, fast and optimized Windows system, with my puny GF4-MX?
I honestly think that if you did all that you talk about you wouldn't gain a thing speed wise. I agree however that the windows manager layout introduces inefficiencies and inconsistencies, but you do get flexibility.
As for the nice, simple and fast MacOS/X display, I've tried it several times and I like X11 better. More choice, more flexibility, and yes, more stability (although the zoom feature is neat).
It's impossible to defend the Gulag and the Soviet policies, I'm not attempting to do that. The parent poster was saying that clearly the US was the good guys and the USSR the bad guys. It's a bit more subtle than that. David Bohm's career was pretty much ruined for no other reasons that he had opinions that did not agree with some committee's. I have no doubt that had McCarthy been given the power he would have put up some version of a Gulag too. So yes there is a very significant difference in execution, but not in intent -- both reek of striking injustice.
You can somehow expect a totalitarian government to behave very badly and shut people off in a very constraining manner. BTW some governments did a lot worse than USSR. Pol Pot anyone? You can also expect a functionning democracy to work better and not kill people off randomly.
That does not make magically all people in the US morally better than in USSR. The system is better, no serious person contests that. Evil lurks everywhere.
In the US too scientists were condemned for holding supposedly communist views. The one that springs to mind is quantum physicist David Bohm. During the McCarthy era he refused to testify, and was stripped of his US citizenship. His position at Princeton was revoked and he was forced into expatriation.
So it's not quite as bad as sending people to the Gulag but it's not altogether nice, wouldn't you say?
However, the key argument (I find) against this sort of argument is that the justice system is not impartial and make a lot of mistakes. One of the reasons is the jury system, where ordinary people don't understand scientific evidence given to them and tend to trust eyewitnesses above everything else [the weakest form of evidence in my view].
A significant number of innocent people have been put to death. It could be you one day, think about it.
They are too, sorry, See this site for a short and concise explanation. They are different representation of the same number.
It's on the last page of Tom's review:
Opteron 240 1.4GHz $283
Opteron 242 1.6Ghz $690
Opteron 244 1.8Ghz $794
There you go.