I think a log based file system would be really good for FLASH device based file systems, where erasing can typically only be achieved for a block of the device at a time. Essentially most flash file systems end up copying data out of a block to some new block before erasing the old block and starting again, much like how a log based file system works.
Given that a lot of FLASH based devices are embedded and maybe portable using battery power (think PDA/Mobile etc...), this technique could add extra safety to data stored on the device if the battery were to drain during an erase or write cycle. Nice.
If this is the case, it's great that such a feature should go into an OS by default. I personally love anything that gives me confidence in the implementation of any applications I write, especially as this type of technique makes debugging much easier.
I don't know why monitors don't just double scan things to reduce the flicker to make it bearable to view. I'd be happy with a 50Hz screen refresh if the monitor was displaying it at 100Hz, heck, when I'm not gaming a 25Hz refresh would probably be good enough if quadrupled in the monitor.
I think cinemas do something similar where each frame is displayed twice such that the flicker is less noticable while the frame rate remains the same. Maybe it's more expensive/difficult for a CRT.
The basic problem is that the length of the hash is always much less than that of the data being hashed. If you compress a 9 bit message into a mere 8 bits, you have to appreciate that there is a 50% chance of a collision i.e. two input messages having the same hash because there are twice as many hashs values as possible messages.
All the hash algorithms are basically up against this problem, and on a much greater scale. The defense is that they use various techniques to make it such that if I were to produce a meaningful message, it is very difficult for an attacker to produce a different message with the same hash value.
To make matters worse, it has already been pointed out elsewhere that many message formats (email, PDF, PS, Word Docs etc...) already contain lots of redundant data that can be manipulated to reach some desired hash value in a way that is not easily observable by the user. Given this, and fast algorithms to find collisions, I think such research is quite signifiant.
Re:Too many keyboard layouts
on
Blank Keyboard
·
· Score: 1
Not to mention laptop keyboards. I just moved from a Toshiba to an HP laptop and was horrified to find the home and end keys are placed in the top right of the keyboard, miles away from the arrow keys. This makes it more difficult to navigate around a document via the keypad alone:(
I wouldn't mind so much, but there are just two blank spaces above the arrow keys now - marked by the XX in the following diagram.
Which you should left unchecked if you're not entirely stupid
I always thought that one of Apple's selling points was that they are made for non-experts. So giving users an option to potentially shoot their foot off seems to be a little unfortunate. Almost by definition, few people are experts.
It seems that most people have corrected you on this point already.
When you move between WiFi base stations, you first must discover your new DHCP server, then get a new local IP address, then reconnect to the VOIP server.
This think that this aspect of mobility is a good concern. GSM networks go to great lengths to co-ordinate handovers such that the delay is minimised as much as possible. I'm not sure that wifi has been designed with the same concerns, and it could present a problem.
It's cool to see robots driving around and doing things, but at this level it's much easier to write a program to simulate the simple environment and then use a much bigger population and experiment faster than realtime. This also overcomes the annoyances of a real system where flat batteries, imperfect sensors and the need for the 'bots to correctly align in order to mate are significant to the problem, but not the investigation of the GA/GP.
When I was interested in this sort of thing I wrote simple simulations of bots moving around a field avoiding circular obsticles. Each bot had 8 sensors and encoded a byte that would index into a lookup table (the genetic material) to give the next action (move forward, turn left or turn right). Fitness was judged on distance moved div collisions and after a couple of hundred generations of 20 candidates, the virtual bots (rendeded as coloured circles) really could be seen to have improved.
Maybe I'm being harsh as this site looks like someone having a lot of fun and raising awareness of GA's, but with a population of 2 basic Lego 'bots, I don't think much about GAs is really being seen. I guess I see less point when [poorly funded?] University departments make little robots to play with these things.
Why is open source software always so memory-hungry?
I really don't know the answer, but would expect that is is partly due to different apps and code blocks being written by lots of different people in some amount of isolation. This could lead to a limited view of the project/distribution and as such the various inefficencies in the system may get missed.
I cant write code myself, so obviously there is a lot that i dont know. But is it really that hard to write code that is portable?
Part of the problem is that unless a developer tests their code on another platform, you don't know if it is truely portable (unless you are writing a pretty basic application that is compiled with -ansi -pedantic). As you can imagine, a lot of developers will not take the time to do this testing, maynot want to support other platforms, or maynot own other platforms on which to test their code.
Surely this is because MS search has Encarta, so it is in Googles interest for there to be a viable and free alternative - namely wikipedia.
I seem to remember some MS person comparing Wikipedia with a public toilet in recent times. Something about not being sure what state it will be in, and not knowing who's used it before... Personally I think Wikipedia is great though, and that's probably what really matters.
I presume that this is because the binaries are owned by root and permissions are 0755? Thus running it as someone else doesn't allow you to update it (a bit like trying to rm rm itself).
Can't you just change this? e.g. change the group for the firefox files to something like 'firefox_update' and then add the trusted users to this group? I agree it's not ideal, but it may work around your problem, although you are basically disolving some security. It's probably safer to just switch to root once in a while and update, running safely locked down at other times.
It's almost a benefit of Linux as most Windows users will always have the possibility of any old process hacking the binaries!
I saw a splashpad charing a couple of mobile phone batteries and gadgets when they were at a recruitment fair in Cambridge last year.
It's a very cool technology, impressively charging multiple devices on the pad at the same time. The product I observed seemed pretty advanced, I wonder what the holdup is in releasing it?
I think the only reason that the use Real is that the streams are more proprietary and harder to rip (for the novice in anycase), and it probably makes some copyright holders happier to let the BBC re-webcast certain content.
When the BBC began publishing audio and video content Real Media was the most secure form of streaming. Unfortunately Microsoft no longer supports Real content. Consequently, many of the later versions of the Windows Media Player will not play our clips. This may change in the future.
NB: Some World Service clips are also streamed for the Windows Player. "
The problem I have is that my compilations are almost always disk bound, and not CPU bound. This is because I do most of my work on a laptop which has a very slow 4200 rpm HDD....
Since distcc still requires the local machine to pre-process all the files and open all the include files (of which there are a great many for my work), I don't really expect that distcc would give a huge performance increase; certainly the gains with ccache have been show to only be around 180 seconds over a normal compile run of 660 seconds...
The specs to which GSM phones are implemented list a number of AT commands to allow you to send an SMS among other things. With a bit of scripting, it should be easy to automate the sending of SMS's repeatedly, should you want to!
It would be great, if instead, I could hook into the compiler and tell it exactly how it should handle vectors.
Well, great for you maybe. What if another programmer wants to look at your code? They would have to first try to understand all your compiler extensions and idiosyncrasies.
There are enough languages and standards already; enabling people to expand and modify them at their whim seems like a bad idea which could result in languages self-destructing!
Yeah, in the future there will probably be some scheme (DRM lite or something?) by which this is prevented such that you have to download ringtones over the air allowing the network to take its cut too.
I'm already aware of one network that blocked the ability for camera pictures taken with a handset from being downloaded direct to PC via a link cable; the network wants to see the data pass through it where it can be billed. Given that this handset was exclusive and branded to the network and heavily subsidised, I think this is quite reasonable.
If you use foreign key constraints you can't dump and restore tables unless you do so in a very specific order.
This has been fixed. You can dump them in any order using mysqldump, and then when re-importing them temporarily disable the foreign key checks with SET FOREIGN_KEY_CHECKS=0;. It's documented here
RedHat explain it here, and as a paying user of RHES3.0 in an enterprise environment, I think this is a good approach for them to have. The features they have left out feel to me to be the more risky sounding things that aren't essential like the new IO sub-system and scheduler tuning, while the things they have taken seem to be more applicable to the apps they may expect users to run e.g. O(1) scheduler, native POSIX library and Huge TLBFS
Interestingly on their page they also list 2.6 as not having Hyperthreading support, while their 2.4 does.
Slashdot Mirror
Oh. I only read the article linked from the story and not the thread linked from the story linked from the article...
Anyway, Perens himself comments on the original story mentioning efence, so I guess my post is redundant in any case. Ho hum.
I think a log based file system would be really good for FLASH device based file systems, where erasing can typically only be achieved for a block of the device at a time. Essentially most flash file systems end up copying data out of a block to some new block before erasing the old block and starting again, much like how a log based file system works.
Given that a lot of FLASH based devices are embedded and maybe portable using battery power (think PDA/Mobile etc...), this technique could add extra safety to data stored on the device if the battery were to drain during an erase or write cycle. Nice.
Didn't Bruce Perens use a similar method of page protection to implement Electric Fence?
http://perens.com/FreeSoftware/ElectricFence/
If this is the case, it's great that such a feature should go into an OS by default. I personally love anything that gives me confidence in the implementation of any applications I write, especially as this type of technique makes debugging much easier.
I don't know why monitors don't just double scan things to reduce the flicker to make it bearable to view. I'd be happy with a 50Hz screen refresh if the monitor was displaying it at 100Hz, heck, when I'm not gaming a 25Hz refresh would probably be good enough if quadrupled in the monitor.
I think cinemas do something similar where each frame is displayed twice such that the flicker is less noticable while the frame rate remains the same. Maybe it's more expensive/difficult for a CRT.
If money was no object, I'd get one of these puppies:
:)
http://www.superssd.com/products/ramsan-400/
That should make compiling nice and fast
The basic problem is that the length of the hash is always much less than that of the data being hashed. If you compress a 9 bit message into a mere 8 bits, you have to appreciate that there is a 50% chance of a collision i.e. two input messages having the same hash because there are twice as many hashs values as possible messages.
All the hash algorithms are basically up against this problem, and on a much greater scale. The defense is that they use various techniques to make it such that if I were to produce a meaningful message, it is very difficult for an attacker to produce a different message with the same hash value.
To make matters worse, it has already been pointed out elsewhere that many message formats (email, PDF, PS, Word Docs etc...) already contain lots of redundant data that can be manipulated to reach some desired hash value in a way that is not easily observable by the user. Given this, and fast algorithms to find collisions, I think such research is quite signifiant.
I'd rather there be little to no exposure of variables.
I guess you are a fan of functional languages then...
Not to mention laptop keyboards. I just moved from a Toshiba to an HP laptop and was horrified to find the home and end keys are placed in the top right of the keyboard, miles away from the arrow keys. This makes it more difficult to navigate around a document via the keypad alone :(
I wouldn't mind so much, but there are just two blank spaces above the arrow keys now - marked by the XX in the following diagram.
XX ^ XX<- V ->
Ho hum.
Which you should left unchecked if you're not entirely stupid
I always thought that one of Apple's selling points was that they are made for non-experts. So giving users an option to potentially shoot their foot off seems to be a little unfortunate. Almost by definition, few people are experts.
Last I checked, VOIP uses TCP sockets.
It seems that most people have corrected you on this point already.
When you move between WiFi base stations, you first must discover your new DHCP server, then get a new local IP address, then reconnect to the VOIP server.
This think that this aspect of mobility is a good concern. GSM networks go to great lengths to co-ordinate handovers such that the delay is minimised as much as possible. I'm not sure that wifi has been designed with the same concerns, and it could present a problem.
It's cool to see robots driving around and doing things, but at this level it's much easier to write a program to simulate the simple environment and then use a much bigger population and experiment faster than realtime. This also overcomes the annoyances of a real system where flat batteries, imperfect sensors and the need for the 'bots to correctly align in order to mate are significant to the problem, but not the investigation of the GA/GP.
When I was interested in this sort of thing I wrote simple simulations of bots moving around a field avoiding circular obsticles. Each bot had 8 sensors and encoded a byte that would index into a lookup table (the genetic material) to give the next action (move forward, turn left or turn right). Fitness was judged on distance moved div collisions and after a couple of hundred generations of 20 candidates, the virtual bots (rendeded as coloured circles) really could be seen to have improved.
Maybe I'm being harsh as this site looks like someone having a lot of fun and raising awareness of GA's, but with a population of 2 basic Lego 'bots, I don't think much about GAs is really being seen. I guess I see less point when [poorly funded?] University departments make little robots to play with these things.
Why is open source software always so memory-hungry?
I really don't know the answer, but would expect that is is partly due to different apps and code blocks being written by lots of different people in some amount of isolation. This could lead to a limited view of the project/distribution and as such the various inefficencies in the system may get missed.
I cant write code myself, so obviously there is a lot that i dont know. But is it really that hard to write code that is portable?
Part of the problem is that unless a developer tests their code on another platform, you don't know if it is truely portable (unless you are writing a pretty basic application that is compiled with -ansi -pedantic). As you can imagine, a lot of developers will not take the time to do this testing, maynot want to support other platforms, or maynot own other platforms on which to test their code.
Surely this is because MS search has Encarta, so it is in Googles interest for there to be a viable and free alternative - namely wikipedia.
I seem to remember some MS person comparing Wikipedia with a public toilet in recent times. Something about not being sure what state it will be in, and not knowing who's used it before... Personally I think Wikipedia is great though, and that's probably what really matters.
The bundled ear buds make a pretty crappy sound and yet are almost a trademark of iPods. I'd say that is form over function...
I presume that this is because the binaries are owned by root and permissions are 0755? Thus running it as someone else doesn't allow you to update it (a bit like trying to rm rm itself).
Can't you just change this? e.g. change the group for the firefox files to something like 'firefox_update' and then add the trusted users to this group? I agree it's not ideal, but it may work around your problem, although you are basically disolving some security. It's probably safer to just switch to root once in a while and update, running safely locked down at other times.
It's almost a benefit of Linux as most Windows users will always have the possibility of any old process hacking the binaries!
I saw a splashpad charing a couple of mobile phone batteries and gadgets when they were at a recruitment fair in Cambridge last year.
It's a very cool technology, impressively charging multiple devices on the pad at the same time. The product I observed seemed pretty advanced, I wonder what the holdup is in releasing it?
All this GUI stuff is overrated - using a secure shell makes it a breeze to administer remote machines from anywhere!
I think the only reason that the use Real is that the streams are more proprietary and harder to rip (for the novice in anycase), and it probably makes some copyright holders happier to let the BBC re-webcast certain content.
See here:The problem I have is that my compilations are almost always disk bound, and not CPU bound. This is because I do most of my work on a laptop which has a very slow 4200 rpm HDD....
Since distcc still requires the local machine to pre-process all the files and open all the include files (of which there are a great many for my work), I don't really expect that distcc would give a huge performance increase; certainly the gains with ccache have been show to only be around 180 seconds over a normal compile run of 660 seconds...
The specs to which GSM phones are implemented list a number of AT commands to allow you to send an SMS among other things. With a bit of scripting, it should be easy to automate the sending of SMS's repeatedly, should you want to!
Info here.
It would be great, if instead, I could hook into the compiler and tell it exactly how it should handle vectors.
Well, great for you maybe. What if another programmer wants to look at your code? They would have to first try to understand all your compiler extensions and idiosyncrasies.
There are enough languages and standards already; enabling people to expand and modify them at their whim seems like a bad idea which could result in languages self-destructing!
Yeah, in the future there will probably be some scheme (DRM lite or something?) by which this is prevented such that you have to download ringtones over the air allowing the network to take its cut too.
I'm already aware of one network that blocked the ability for camera pictures taken with a handset from being downloaded direct to PC via a link cable; the network wants to see the data pass through it where it can be billed. Given that this handset was exclusive and branded to the network and heavily subsidised, I think this is quite reasonable.
If you use foreign key constraints you can't dump and restore tables unless you do so in a very specific order.
This has been fixed. You can dump them in any order using mysqldump, and then when re-importing them temporarily disable the foreign key checks with SET FOREIGN_KEY_CHECKS=0;. It's documented here
RedHat explain it here, and as a paying user of RHES3.0 in an enterprise environment, I think this is a good approach for them to have. The features they have left out feel to me to be the more risky sounding things that aren't essential like the new IO sub-system and scheduler tuning, while the things they have taken seem to be more applicable to the apps they may expect users to run e.g. O(1) scheduler, native POSIX library and Huge TLBFS
Interestingly on their page they also list 2.6 as not having Hyperthreading support, while their 2.4 does.