I fully agree that LWN's response to the management issue is weak, but on the other hand: so is yours to LWN.
Major slippage occurs in professionally managed close source projects as well. In fact, many (Brook's The Mythical Man-Month comes to mind) will claim that it often is the norm. In any case, I've surely experienced it first hand. But to give a much more visible example: just consider the number of times MicroSoft has missed its intended OS deadlines. Just ask youself: how often they announced the one and only unified Windows version that would finally put DOS to rest for the home user? XP is years behind schedule!
It caught on because it is a logical thing to do if/usr/local is defined as "you can play with this in whatever way you want to your heart's delight". In that kind of context,/usr/local cannot be relied upon for anything, so...
I know a place where/usr/local is NFS mounted (and has been so for 12 years or more). It's used for all the shared stuff that is site-local (as opposed to machine-local). Then if you have to install an optional compiler on one machine, where do you go?/opt is the answer Sun and HP etc. came up with, because it's the only thing outside (/usr)/bin that they could safely claim some control over.
Re:Subscriptions should add value
on
Slashdot Updates
·
· Score: 1
Continued existence only is an "added value" of an subscription service if it actually happens and would not happen otherwise.
If sufficient people leave slashdot because they do not want to swallow the larger ads, these ads do not save the site, even if one of those pays more than one traditional one. This, they might in fact might cause the site's downfall. (Hey, I'm not predicting anything, just analysing a mechanism.)
Exactly the same reasoning holds for subscriptions.
The real 2.4 mistake was to push it out even when it wasn't ready.
Producing 2.2 took so long, that Linus sort of vowed to make 2.4 take only +-25% as much time to hit the shelves. When he easily passed that deadline without a 2.4 being in sight, people started to complain and pressurize him. In the end, he sort of gave in (even though I suspect that he himself doesn't agree with that point of view), and pushed 2.4 out too soon.
Of course, the big question is how many of the bugs in 2.4 would not yet have been found and fixed by now if there hadn't been a "stable" version for everyone to jump on and scream about.
This is easy: download patch-2.4.11-dontuse.bz2 AND patch-2.4.12.bz2, rename the former to patch-2.4.11.bz2, and run the patch-kernel script. It will see that it needs to apply both patches and will then sing all the magic songs for you.
He didn't just take it, and not from a fighter squadron.
It was the symbol of a famous Italian WW1 fighter pilot (who got killed while on duty, by the way). His mother later donated the logo to Ferrari for use on his cars. Here's the full story..
Hardware implementations can also be reverse engineered. I once talked to a chip designer who was an citizen of Eastern Germany before the wall came down. One of the things he had worked on, was to duplicate certain chips (from DEC, IIRC, so it could have been VAX cpus) based only on samples. They succeeded.
OK, Joe Random Haxor can't do this, but foreign governments certainly can.
At last someone who does what I just asked people to do in over here: move the discussion forward to why the government doesn't listen to the "criminals don't care about laws" argument and to what it is that can be done to address that.
Everytime something like mandatory key escrow or backdoors is mentioned here, tens of posters reiterate the same old (and by now rather boring) song, namely that criminals do not care about such a law. And most of them even get moderated +5 insightful for doing so.
While I fully agree with the point they try to make, I really cannot imagine that it hasn't been made in the legislative bodies as well. Your aaverage politician really is not that stupid, even if it is trendy to claim otherwise.
So I'd like to request that instead someone who has talked to these people or who has read the proceedings of their meetings tells us exactly why this argument isn't being accepted, or why it is being overruled. No speculation and no "because their morons" statements, please. Just the facts.
Reiterating the same thing over and over in front of the same crowd of devoted followers is not going to change anything other than your/. karma. What really needs to be done, is to find (and then propagate) the proper reply to the reasons why the people who see things differently hold that opinion. Only then do we stand a chance of getting anywhere.
I would assume that this has been a real issue of engineering such buildings for a long time.
Imagine having to tear down the WTC (something that was going to happen anyway at some point in time). Your job becomes a lot easier of the original designers gave it some thought as well.
As far as volume goes - maybe there should just be a cap on patents. Something like "x patents a year can be awarded".
You've got to be kidding. Such a system could mean that a company cannot patent anything between, say, July and December, only because its competitors filled up the cap with "bogus" patents during the first half of the year. How long before the latter becomes a generally applied strategy? Smells like a preprogrammed denial of service attack to me.
Non-acid-free paper isn't the only problem hurting traditional books. There's also something called ink corrosion, which will destroy even good old quality paper in the long run.
But yes, I too prefer the genuine article over any of the modern extremely volatile e-replacements.
Basically, the stuff that I use to keep track of what is really important to me (some of those things are just memories "attached" to a book or document that anybody else will regard as "just a pile of old paper") should at least be able to outlast me. With the life expectancy of an average European being somewhere near 75 years, that means any technology that hopes to replace paper must reliably last 60 years before I will consider using it exclusively. And of course it needs to be as convenient as real paper as well.
Franckly, I really don't expect something that satisfies these criteria to show up during my lifefime (and in my family we have a tendency to reach grow very old).
Actually, I'm not sure you read his response correctly. There have been some high profile cases in which internal e-mails turned into embarassing evidence. Microsoft vs. DOJ comes to mind.
Instead of organizing a mass deletion once a year, why not delete them right away and enjoy having a managable inbox in which you won't loose track of an important mail by accident? It's just a click of the mouse after all (which you have to spend anyway hitting "next message").
The issue is not whether C++ is standardised or not, but how compatible the various compilers are
with that standard (or at the very least with each other). As one who has to fight with the C++ compilers from Sun, HP, and GNU on a daily basis I understand very well why they preferred C over C++.
Sun's compiler is very bad indeed. The HP one (I'm talking about aCC here, not about their old CC
compiler mock up) is OK. It's definitely much better than what Sun tries to push down our throat. It also easily beats the old gcc 2.95 series.
Re:what a predicament ...
on
Linux Turns 10
·
· Score: 1
I hate to say it, but worse cases have been reported. Cases in which NOT all the swap had been used up. I've been reading LKML on a daily basis for long enough to be able to understand at least some of it. Now, I'm not saying that I actually verified these specific reports, but there have been some.
Concerning the advice to stop when one hits swap: that kills Linux for the use we currently make of our aging HP-UX servers. Of course we don't want them to swap. But they are big servers after all, shared by tens of potential users who 1) don't always know how to check these things; 2) don't sit right next to it (hence they get no audible warning from the disks going mad); and 3) run jobs that take days of weeks anyway (hence they do daily not look into why their results are slow to come in).
To make matters worse, we're a research organisation, so it is standard practice for some of us to start something of which the run time behaviour initially is unknown.
Re:what a predicament ...
on
Linux Turns 10
·
· Score: 1
This entire article is not about technical isues, so I tried to avoid starting an off-topic technical debate.
But if you really want to know why the early 2.4 VM is bad (it's finally improving a bit with 2.4.[89], fortunately), just go reading the Linux kernel mailing list archives. I scan the list in real time. And read a substantial part of it. Look for reports of machines that hang for minutes/hours while trying to sort out what to do with memory pages when the have hundreds of free megabytes of swap and other nasty things like that. Who cares about overcommit if the machine just locks up before even reaching the limit?
And now for another ad hominem attack: have you actually used 2.4.x like I do all day? Or have just just uttered an ad hominem attack against me for getting moderated up while mentioning a problem with Linux?
Re:Wrong Date?
on
Linux Turns 10
·
· Score: 2, Informative
I'm sorry to report you wrong. August 25 was the date that Linus first publicly announced that fact that he was working on something. The actual work had started some time before. Somewhere in April IIRC.
Re:what a predicament ...
on
Linux Turns 10
·
· Score: 2, Interesting
I'm not the kernel, but if I were, the top of my wishlist would be a VM that actually works.
No, this is not flamebait. It's an honest (and somewhat sadened) remark by a very long time Linux fan (cfr. my signature) who just finally managed to get the sysadmins at his office (very much against the desires of their NT-minded (blinded?) boss) to seriously consider Linux SMP servers with Gigabytes of RAM for some heavy E-CAD work as a replacement for our aging HP-UX boxes. In fact, one such Linux sweety will likely be ordered quite soon for evaluation purposes. And what happens? Precisely now the 2.4 kernels are taking over the various distributions while having major trouble with their VM in exactly the kind of conditions we want to use them for.
Just to short cut one kind of replies: Of course we can use an older distribution or build our own combination of things (heck, I don't even use a distribution at home and compile everything from scratch). But at work, I'm not a sysadmin, and we have to make do with a few UNIX and NT experts and lots of people who how how to fix NT problems, but whom one might suspect of fearing that a Linux box will explode if they push the wrong key. And no, the latter is not a reproach, its an simple observation. It's quite normal given their backgounds, but it's also a major problem for us Linux zealots.
I absolulety hate to say it, but Linux still has a few more years to go before really making it, even in certain non-desktop roles. It needs a several more improvements in the technical department. It definitely needs to loose its tendency to have stable versions that aren't stable at all until several months after initial release. But most of all, it needs a whole class of people who know a considerable amount about using computers (read as: Windows machines), but don't really understand them, to get used to something which, to them at least, is completely new and exotic.
If this is true, you should talk to Hotmail about them having a major security problem, because in that case all Hotmail users clearly are open to all sorts of very nasty denial-of-service attacks.
If your e-mail quota are filling up, they should simply refuse to accept more mail, not delete old stuff. This scheme too is prone to denial of service, but at least your correspondents will know that their message to you was lost and that they should try again later.
I'm talking about using an encryption code when writing a new password on it. One that is simple enough for me to remember and decrypt without a computer (a pen and another "temporary" piece of paper will do).
But don't count on me explaining it here. The enemy might be reading... The day that I die, the algorithm will die with me. Especially since those who find the paper are unlikely to know what it is.
Then I'd suggest simply adding a "sleep 3600" to "make world" for X. You will see a whole new world of opportunities open up in front of your very eyes.
I'm currently tracking 25 passwords (two of which are not computer related, actually). Some of them change too often to be sure that I'll remember them when in high need.
So I've had to write them down "somewhere" bloody safe. As it happens, I ended up encrypting the piece of paper, such that the only thing that I definitely have to remember is the non-trivial decryption scheme. Of course, I also remember the passwords that I need most often, but for the others my encrypted paper has occasionally worked miracles.
Slashdot Mirror
I fully agree that LWN's response to the management issue is weak, but on the other hand: so is yours to LWN.
Major slippage occurs in professionally managed close source projects as well. In fact, many (Brook's The Mythical Man-Month comes to mind) will claim that it often is the norm. In any case, I've surely experienced it first hand. But to give a much more visible example: just consider the number of times MicroSoft has missed its intended OS deadlines. Just ask youself: how often they announced the one and only unified Windows version that would finally put DOS to rest for the home user? XP is years behind schedule!
It caught on because it is a logical thing to do if /usr/local is defined as "you can play with this in whatever way you want to your heart's delight". In that kind of context, /usr/local cannot be relied upon for anything, so...
/usr/local is NFS mounted (and has been so for 12 years or more). It's used for all the shared stuff that is site-local (as opposed to machine-local). Then if you have to install an optional compiler on one machine, where do you go? /opt is the answer Sun and HP etc. came up with, because it's the only thing outside (/usr)/bin that they could safely claim some control over.
I know a place where
Continued existence only is an "added value" of an subscription service if it actually happens and would not happen otherwise.
If sufficient people leave slashdot because they do not want to swallow the larger ads, these ads do not save the site, even if one of those pays more than one traditional one. This, they might in fact might cause the site's downfall. (Hey, I'm not predicting anything, just analysing a mechanism.)
Exactly the same reasoning holds for subscriptions.
Producing 2.2 took so long, that Linus sort of vowed to make 2.4 take only +-25% as much time to hit the shelves. When he easily passed that deadline without a 2.4 being in sight, people started to complain and pressurize him. In the end, he sort of gave in (even though I suspect that he himself doesn't agree with that point of view), and pushed 2.4 out too soon.
Of course, the big question is how many of the bugs in 2.4 would not yet have been found and fixed by now if there hadn't been a "stable" version for everyone to jump on and scream about.
This is easy: download patch-2.4.11-dontuse.bz2 AND patch-2.4.12.bz2, rename the former to patch-2.4.11.bz2, and run the patch-kernel script. It will see that it needs to apply both patches and will then sing all the magic songs for you.
He didn't just take it, and not from a fighter squadron.
It was the symbol of a famous Italian WW1 fighter pilot (who got killed while on duty, by the way). His mother later donated the logo to Ferrari for use on his cars. Here's the full story..
Hardware implementations can also be reverse engineered. I once talked to a chip designer who was an citizen of Eastern Germany before the wall came down. One of the things he had worked on, was to duplicate certain chips (from DEC, IIRC, so it could have been VAX cpus) based only on samples. They succeeded.
OK, Joe Random Haxor can't do this, but foreign governments certainly can.
At last someone who does what I just asked people to do in over here: move the discussion forward to why the government doesn't listen to the "criminals don't care about laws" argument and to what it is that can be done to address that.
While I fully agree with the point they try to make, I really cannot imagine that it hasn't been made in the legislative bodies as well. Your aaverage politician really is not that stupid, even if it is trendy to claim otherwise.
So I'd like to request that instead someone who has talked to these people or who has read the proceedings of their meetings tells us exactly why this argument isn't being accepted, or why it is being overruled. No speculation and no "because their morons" statements, please. Just the facts.
Reiterating the same thing over and over in front of the same crowd of devoted followers is not going to change anything other than your /. karma. What really needs to be done, is to find (and then propagate) the proper reply to the reasons why the people who see things differently hold that opinion. Only then do we stand a chance of getting anywhere.
I would assume that this has been a real issue of engineering such buildings for a long time.
Imagine having to tear down the WTC (something that was going to happen anyway at some point in time). Your job becomes a lot easier of the original designers gave it some thought as well.
You've got to be kidding. Such a system could mean that a company cannot patent anything between, say, July and December, only because its competitors filled up the cap with "bogus" patents during the first half of the year. How long before the latter becomes a generally applied strategy? Smells like a preprogrammed denial of service attack to me.
Non-acid-free paper isn't the only problem hurting traditional books. There's also something called ink corrosion, which will destroy even good old quality paper in the long run.
But yes, I too prefer the genuine article over any of the modern extremely volatile e-replacements.
Basically, the stuff that I use to keep track of what is really important to me (some of those things are just memories "attached" to a book or document that anybody else will regard as "just a pile of old paper") should at least be able to outlast me. With the life expectancy of an average European being somewhere near 75 years, that means any technology that hopes to replace paper must reliably last 60 years before I will consider using it exclusively. And of course it needs to be as convenient as real paper as well.
Franckly, I really don't expect something that satisfies these criteria to show up during my lifefime (and in my family we have a tendency to reach grow very old).
Actually, I'm not sure you read his response correctly. There have been some high profile cases in which internal e-mails turned into embarassing evidence. Microsoft vs. DOJ comes to mind.
Instead of organizing a mass deletion once a year, why not delete them right away and enjoy having a managable inbox in which you won't loose track of an important mail by accident? It's just a click of the mouse after all (which you have to spend anyway hitting "next message").
The issue is not whether C++ is standardised or not, but how compatible the various compilers are
with that standard (or at the very least with each other). As one who has to fight with the C++ compilers from Sun, HP, and GNU on a daily basis I understand very well why they preferred C over C++.
Sun's compiler is very bad indeed. The HP one (I'm talking about aCC here, not about their old CC
compiler mock up) is OK. It's definitely much better than what Sun tries to push down our throat. It also easily beats the old gcc 2.95 series.
Concerning the advice to stop when one hits swap: that kills Linux for the use we currently make of our aging HP-UX servers. Of course we don't want them to swap. But they are big servers after all, shared by tens of potential users who 1) don't always know how to check these things; 2) don't sit right next to it (hence they get no audible warning from the disks going mad); and 3) run jobs that take days of weeks anyway (hence they do daily not look into why their results are slow to come in).
To make matters worse, we're a research organisation, so it is standard practice for some of us to start something of which the run time behaviour initially is unknown.
But if you really want to know why the early 2.4 VM is bad (it's finally improving a bit with 2.4.[89], fortunately), just go reading the Linux kernel mailing list archives. I scan the list in real time. And read a substantial part of it. Look for reports of machines that hang for minutes/hours while trying to sort out what to do with memory pages when the have hundreds of free megabytes of swap and other nasty things like that. Who cares about overcommit if the machine just locks up before even reaching the limit?
And now for another ad hominem attack: have you actually used 2.4.x like I do all day? Or have just just uttered an ad hominem attack against me for getting moderated up while mentioning a problem with Linux?
See also this history of linux.
No, this is not flamebait. It's an honest (and somewhat sadened) remark by a very long time Linux fan (cfr. my signature) who just finally managed to get the sysadmins at his office (very much against the desires of their NT-minded (blinded?) boss) to seriously consider Linux SMP servers with Gigabytes of RAM for some heavy E-CAD work as a replacement for our aging HP-UX boxes. In fact, one such Linux sweety will likely be ordered quite soon for evaluation purposes. And what happens? Precisely now the 2.4 kernels are taking over the various distributions while having major trouble with their VM in exactly the kind of conditions we want to use them for.
Just to short cut one kind of replies: Of course we can use an older distribution or build our own combination of things (heck, I don't even use a distribution at home and compile everything from scratch). But at work, I'm not a sysadmin, and we have to make do with a few UNIX and NT experts and lots of people who how how to fix NT problems, but whom one might suspect of fearing that a Linux box will explode if they push the wrong key. And no, the latter is not a reproach, its an simple observation. It's quite normal given their backgounds, but it's also a major problem for us Linux zealots.
I absolulety hate to say it, but Linux still has a few more years to go before really making it, even in certain non-desktop roles. It needs a several more improvements in the technical department. It definitely needs to loose its tendency to have stable versions that aren't stable at all until several months after initial release. But most of all, it needs a whole class of people who know a considerable amount about using computers (read as: Windows machines), but don't really understand them, to get used to something which, to them at least, is completely new and exotic.
If your e-mail quota are filling up, they should simply refuse to accept more mail, not delete old stuff. This scheme too is prone to denial of service, but at least your correspondents will know that their message to you was lost and that they should try again later.
--
--
But don't count on me explaining it here. The enemy might be reading... The day that I die, the algorithm will die with me. Especially since those who find the paper are unlikely to know what it is.
--
--
So I've had to write them down "somewhere" bloody safe. As it happens, I ended up encrypting the piece of paper, such that the only thing that I definitely have to remember is the non-trivial decryption scheme. Of course, I also remember the passwords that I need most often, but for the others my encrypted paper has occasionally worked miracles.
--