Either I need indemnification, or I don't
I think you misunderstand insurance. Either I am going to crash my car, or I am not going to crash my car. If I won't crash my car, then I don't need insurance. You know that's not how it works. I just don't _know_ if I are going to crash my car or not. If we knew who was going to crash their cars, we would prevent it!
The point of insurance is that it spreads the cost of Bad Things (tm) over the whole population of people (and corporations) that are at risk of Bad Things, proportionately to the level of their risk.
No. It's far better to have both tracks available. Some people want the benefits of Linux-based OS, but want to keep the powerful/ easy-to-use features of some Windows programs (like Photoshop, newer MS Excel, MS Access), but some people just want cheaper, alternatives to MS Office, and Photoshop etc. but want to keep windows cos they know it (they use OO.o and Gimp). Different ppl want different things...
Google is available pretty much where phone is (POTS or GSM available), and in a library (in the West they always have broadband or some sort). So assuming we're not talking sub-saharen Africa, google is at least as available as the other sources.
Fair point tho'.
Exactly. The point of the article is (as someone else pointed out) is that clear, high-level code is easy to optimize, since it is easy to understand, and thus it's easy to reason about the code.
It doesn't sound like any low level work is going to get you anywhere in your simulation. The best bet is to buy lots of hardware to brute-force it... ... or get smart! Reason about the problem. Is it important to evaluate the function for all possible combinations of all possible values all the variables and parameters? Is there any hidden constraint or relationship between those variables? The economic model which provides those variables may well have made a distinction between two variables, or an assumption of the independence of two variables, which is not relevant to your modelling.
Or might statistics/ heuristics help? Picking the most likely region(s) (based on other theory) for computation, calculating that (those) first, and then working out into (assuming probability is smooth) less likely region(s).
As the article points out, these kinds of optimizations (very similar conceptually to those in the article) are those easiest to do in a very high-level language.
In fact, one big problem is the way that the loader performs relocations on C++ libraries. Google it. It's why KDE apps take a few seconds to load (and used to be even worse). IIRC the main problem is that many objects (function, variable etc.) need to be copied into the address space of each application using it, so that the sharing never happens in practise.
A fix in the source is a fix for every distro, where as a fix in a package fixes only a single release of a single distro.
This is also a weakness (for some use-cases) - it prevents the users from easily selecting fixes. Debian has a very crude way of doing this: run stable and you only get critical fixes, run experimental and get them all.
What Gentoo could do is run branches in CVS (if it doesn't already - I don't know) and have all fixes in HEAD, only the most critical in STABLE, and some grades in between. But then you've got BSD, and you're supporting your own packages....
From here:
"[M]y new machine wasn't very stable.
Obviously my first reaction was to go back down to 2x366Mhz. But now I experienced an even stranger problem. As long as my machine kept the CPUs chugging away, the machine didn't lock up. But if I left the machine idle overnight, there was a good probability that the system would lock up completely. Yes, an idle bug -- argh!"
And thus Gentoo was born: as a way to prevent idle bugs by keeping the CPU active 24/7!
The big claim that he makes is that MS Word succeeded because they gave users what they wanted.
Quote from TFA: "Some people ask why we don't make [MS Word's file format] public, often in a tone that implies we are somehow required to do this ethically. We don't do that because it is our intellectual property. People who want to work with us can get it by contacting us; people who want to compete with us need to work harder. That's business. We might change our minds about that if it seems that making the format public would be of most benefit, but really it is our prerogative."
Firstly, it is in no way clear what 'intellectual property' has to do with secret file formats. The format is a secret, that is all. He uses 'property' to suggest that if given away, the original owner loses, but does not support that assertion.
Secondly, the problem with this attitude is that if the format that a user's document is stored in is the property of the software supplier, then the user's document is not really their own. It is partly their own, but in practical terms, its _use_ is partly owned by the software supplier.
Thirdly, I think that one big thing that many users want (myself included) is to own their own documents. If Word won its position by giving users what they want, let's hope it keeps it only by doing the same. Why the hell is it written into so many laws that reverse-engineering is allowed for the purposes of inter-operability?
...except as others have pointed out Linux doesnt stop working without the GPL license marker, it just gets tacked onto bug reports. So the analogy is not that good.
These kind of programs on MSDN are more documentation than programs. The _real programs_ (like Rotor, the.NET engine) apart from this new one AFAIK are all shared-source (ie you _cannot_ re-use the code, only look). Yes it's useful, but snippets of code on how to implement hash tables are not the same thing as MySQL (for example).
Re:It's the same as in computers in general...
on
Beyond Megapixels
·
· Score: 1
I'd disagree that "good optics in the lens is still damn expensive". Yes optics have improved in the last say 50 years, but good quality optics have been around that long, and a lot of it is floating around second hand. The latest pro gear has bells and whistles that mean that pros can use it in harsh conditions, in tricky photographic conditions, and it will last for a decent length of time. That doesn't mean that a modern pro lens is significantly optically superior, it just retains is (normally small) optical superiority over a very wide range of parameters. Also some modern 'features' on optics are like in computers, hifis or any other fields: gimmicks that sound good but do nothing of value.
Correct me if I'm wrong, but this double-checking only applies to RST-ACKs (which an attacker is under no obligation to use). The attack can just use RSTs. (See W. R. Stevens, TCP/IP Illustrated, p227: "ACK The acknowlegement number is valid.").
It's called IPSEC, it's secure on the IP level up so TCP is encrypted over it.
Correct, for suitable levels of 'secure'. Schneier and Ferguson's evaluation of IPSec. It's no panacea... But nothing is:)
The best defence against this? Simply check for a stream of RST packets. They dont come in huge bundles with incrementing sequence numbers often. Detect that signature, block IP, sorted.
What would your preferred way of implementing this defence be? Is it easy to automate on linux (firewalls?)?
Correct. When a collegue of Einstein's suggested that it was impossible for an object with mass to reach light speed, Einstein felt compelled to point out that a photon has mass and it travels at light speed.
But a photon has no rest mass, unlike the objects that cannot reach light-speed.
the very fact that MS has not bought SCO suggests that MS believe that there is a substantial chance that SCO will lose and be liable for large damages to IBM, Redhat et al, which MS does not want to be liable for. But that's just one possible explanation...
"Last mile to the user is propietary" is the state of affairs the author invisages. Which is pretty much the case: Xandros, Lindows. They both are Free software underneath with an attractive veneer of easy to use propietary software on top.
The same could even be said of MacOS X (in places): KHTML underneath.
Basically: he makes a good point. Selling goods and services based on a Free use of commonly owned software basically means either doing to tricky things for users (consultancy) or writing apps so that users can do the tricky things themselves (distributions do this).
Distributions either make their nice configuration tool FOSS or propietary. Redhat do the former, Xandros and Lindows (AFAIAA) do the latter. Whose configuration tools are easier?
Slashdot Mirror
Either I need indemnification, or I don't
I think you misunderstand insurance. Either I am going to crash my car, or I am not going to crash my car. If I won't crash my car, then I don't need insurance. You know that's not how it works. I just don't _know_ if I are going to crash my car or not. If we knew who was going to crash their cars, we would prevent it!
The point of insurance is that it spreads the cost of Bad Things (tm) over the whole population of people (and corporations) that are at risk of Bad Things, proportionately to the level of their risk.
yeah - I have an 8x SCSI Plextor cd writer in the computer across the room from me. I can't remember it having made a coaster. damned good drive.
No. It's far better to have both tracks available. Some people want the benefits of Linux-based OS, but want to keep the powerful/ easy-to-use features of some Windows programs (like Photoshop, newer MS Excel, MS Access), but some people just want cheaper, alternatives to MS Office, and Photoshop etc. but want to keep windows cos they know it (they use OO.o and Gimp). Different ppl want different things...
obviously the catalogue is accessible electronically and the library user can walk fast :)
Google is available pretty much where phone is (POTS or GSM available), and in a library (in the West they always have broadband or some sort). So assuming we're not talking sub-saharen Africa, google is at least as available as the other sources.
Fair point tho'.
Exactly. The point of the article is (as someone else pointed out) is that clear, high-level code is easy to optimize, since it is easy to understand, and thus it's easy to reason about the code.
... or get smart! Reason about the problem. Is it important to evaluate the function for all possible combinations of all possible values all the variables and parameters? Is there any hidden constraint or relationship between those variables? The economic model which provides those variables may well have made a distinction between two variables, or an assumption of the independence of two variables, which is not relevant to your modelling.
It doesn't sound like any low level work is going to get you anywhere in your simulation. The best bet is to buy lots of hardware to brute-force it...
Or might statistics/ heuristics help? Picking the most likely region(s) (based on other theory) for computation, calculating that (those) first, and then working out into (assuming probability is smooth) less likely region(s).
As the article points out, these kinds of optimizations (very similar conceptually to those in the article) are those easiest to do in a very high-level language.
In fact, one big problem is the way that the loader performs relocations on C++ libraries. Google it. It's why KDE apps take a few seconds to load (and used to be even worse). IIRC the main problem is that many objects (function, variable etc.) need to be copied into the address space of each application using it, so that the sharing never happens in practise.
A fix in the source is a fix for every distro, where as a fix in a package fixes only a single release of a single distro.
This is also a weakness (for some use-cases) - it prevents the users from easily selecting fixes. Debian has a very crude way of doing this: run stable and you only get critical fixes, run experimental and get them all.
What Gentoo could do is run branches in CVS (if it doesn't already - I don't know) and have all fixes in HEAD, only the most critical in STABLE, and some grades in between. But then you've got BSD, and you're supporting your own packages....
From here:
"[M]y new machine wasn't very stable.
Obviously my first reaction was to go back down to 2x366Mhz. But now I experienced an even stranger problem. As long as my machine kept the CPUs chugging away, the machine didn't lock up. But if I left the machine idle overnight, there was a good probability that the system would lock up completely. Yes, an idle bug -- argh!"
And thus Gentoo was born: as a way to prevent idle bugs by keeping the CPU active 24/7!
The big claim that he makes is that MS Word succeeded because they gave users what they wanted.
Quote from TFA:
"Some people ask why we don't make [MS Word's file format] public, often in a tone that implies we are somehow required to do this ethically. We don't do that because it is our intellectual property. People who want to work with us can get it by contacting us; people who want to compete with us need to work harder. That's business. We might change our minds about that if it seems that making the format public would be of most benefit, but really it is our prerogative."
Firstly, it is in no way clear what 'intellectual property' has to do with secret file formats. The format is a secret, that is all. He uses 'property' to suggest that if given away, the original owner loses, but does not support that assertion.
Secondly, the problem with this attitude is that if the format that a user's document is stored in is the property of the software supplier, then the user's document is not really their own. It is partly their own, but in practical terms, its _use_ is partly owned by the software supplier.
Thirdly, I think that one big thing that many users want (myself included) is to own their own documents. If Word won its position by giving users what they want, let's hope it keeps it only by doing the same. Why the hell is it written into so many laws that reverse-engineering is allowed for the purposes of inter-operability?
...except as others have pointed out Linux doesnt stop working without the GPL license marker, it just gets tacked onto bug reports. So the analogy is not that good.
To see the blogs of those involved and commenting, go here.
See Antti-Juhani Kaijanaho and Joey Hess in particular. Anthony Towns (the Release Manager in question) has also blogged on the issue.
These kind of programs on MSDN are more documentation than programs. The _real programs_ (like Rotor, the .NET engine) apart from this new one AFAIK are all shared-source (ie you _cannot_ re-use the code, only look).
Yes it's useful, but snippets of code on how to implement hash tables are not the same thing as MySQL (for example).
I'd disagree that "good optics in the lens is still damn expensive". Yes optics have improved in the last say 50 years, but good quality optics have been around that long, and a lot of it is floating around second hand. The latest pro gear has bells and whistles that mean that pros can use it in harsh conditions, in tricky photographic conditions, and it will last for a decent length of time. That doesn't mean that a modern pro lens is significantly optically superior, it just retains is (normally small) optical superiority over a very wide range of parameters.
Also some modern 'features' on optics are like in computers, hifis or any other fields: gimmicks that sound good but do nothing of value.
Correct me if I'm wrong, but this double-checking only applies to RST-ACKs (which an attacker is under no obligation to use). The attack can just use RSTs. (See W. R. Stevens, TCP/IP Illustrated, p227: "ACK The acknowlegement number is valid.").
It's called IPSEC, it's secure on the IP level up so TCP is encrypted over it. :)
Correct, for suitable levels of 'secure'. Schneier and Ferguson's evaluation of IPSec. It's no panacea... But nothing is
The best defence against this? Simply check for a stream of RST packets. They dont come in huge bundles with incrementing sequence numbers often. Detect that signature, block IP, sorted.
What would your preferred way of implementing this defence be? Is it easy to automate on linux (firewalls?)?
Correct. When a collegue of Einstein's suggested that it was impossible for an object with mass to reach light speed, Einstein felt compelled to point out that a photon has mass and it travels at light speed.
But a photon has no rest mass, unlike the objects that cannot reach light-speed.
the very fact that MS has not bought SCO suggests that MS believe that there is a substantial chance that SCO will lose and be liable for large damages to IBM, Redhat et al, which MS does not want to be liable for.
But that's just one possible explanation...
actually it makes no difference if safe search is on or off.
"Last mile to the user is propietary" is the state of affairs the author invisages. Which is pretty much the case: Xandros, Lindows. They both are Free software underneath with an attractive veneer of easy to use propietary software on top.
The same could even be said of MacOS X (in places): KHTML underneath.
Basically: he makes a good point. Selling goods and services based on a Free use of commonly owned software basically means either doing to tricky things for users (consultancy) or writing apps so that users can do the tricky things themselves (distributions do this).
Distributions either make their nice configuration tool FOSS or propietary. Redhat do the former, Xandros and Lindows (AFAIAA) do the latter. Whose configuration tools are easier?
What I want to know is if we are going to have to put up with repeats every hour, on the hour, like last year...
Did anyone else parse that as 'MOD CHIP UP!' where chip must be a user?
GNOME 2.6 going into experimental. If it is stable (unlike GIMP 2.0, in my experience, and KDE 3.2.0, by report of the debian maintainer).
not to mention udev
(now in unstable debian).
I am impressed - you called yourself an arsehole (modulo spelling) and got moderated insightful - ouch!