Phones don't use uniform sampling they use u-law or a-law. Which samples in a way that compliments our hearing The reasion this isn't done with hi-fi is down to practical problems implienting the required DAC/ADCs at the required number of bits. (without making the quanization noise swamp any advantage).
Also see KL transform (AKA principle component anayisys) for removing statisticly redundent data. Not very practical for everyday use, but it has been demonstrated for storing faces on the 58 byte magnetic strip of a credit card. This works because faces span a very small subspace of the original images so the number of dimentions can be reduced.
Dictionary based compression is just one of the lossless methods. There are many other's like huffman coding (which requires a non-uniform distribution).
If they clame it only works with "small sequences" then maybe they are just compressing the predictability in the flawed random number sequence they are using as test data.
Maybe they need that patch that add's network generated entropy into/dev/random 9-)
You can cut the data rate by using motion detection and only recording when there is something interesting going on. This would make a HUGE difference in most security applications.
Hold on, the CD's include application software!
Are you trying to tell me that windows, office, painshop pro, visual C++ would fit on a single
CD!
Next you'll be saying SuSE crap because it comes
on 6 cds!
Slashdot Mirror
SMP and/or, god forbid, a decent non-x86 design running at a couple of GHzs.
Take me to your S390....
Have they tryed pressing the ESC key yet?
I'm not surprised!
Have you checked for nearby chronoton emisions?
Try a tatacion burst, that normaly helps...
I've just "ported" hangman form PDP-11 version of BSD to SuSE 7.2. Runs very nicely on my 1.4G athlon 8-).
It only takes a two line change to the source and a one line makefile tweek.
Maybe I should make it into an RPM and stick it on freshmeat 9-)
Errr thats why there's a reconstruction filter with a sinc function!!
Phones don't use uniform sampling they use u-law or a-law. Which samples in a way that compliments our hearing The reasion this isn't done with hi-fi is down to practical problems implienting the required DAC/ADCs at the required number of bits. (without making the quanization noise swamp any advantage).
Also see KL transform (AKA principle component anayisys) for removing statisticly redundent data. Not very practical for everyday use, but it has been demonstrated for storing faces on the 58 byte magnetic strip of a credit card. This works because faces span a very small subspace of the original images so the number of dimentions can be reduced.
Infact huffman coding is a bad example of "others" as it *is* a dictionary based method or al least part of one. (OOPs)
Dictionary based compression is just one of the lossless methods. There are many other's like huffman coding (which requires a non-uniform distribution).
It's a *huge* field.
If they clame it only works with "small sequences" then maybe they are just compressing the predictability in the flawed random number sequence they are using as test data.
/dev/random 9-)
Maybe they need that patch that add's network generated entropy into
Only an American would assume that because its dark out side, it must be night round the entire planet!
Store your mbox (or whatever) on an encrypted file system. Not perfect but easy.
Fuel cells produce water as one of their waste products right.
Do I *realy* want water dripping out from my mobile when I use it? - I think not.
Wet pactches in your pocket. - no thanks.
I suppose it could be stored in a little tank, so you could squirt it at your mates or drink it 9-).
8mm tape! LOL
How much cash would it take to make a programmer with access talk?
Very little cf. with the cost of the craft if there like anyone I know!
Crash it into something and watch it burn!
(This isn't a real question right?)
mpeg video codecs already compress the sound.
You can cut the data rate by using motion detection and only recording when there is something interesting going on. This would make a HUGE difference in most security applications.
It's that time of year again, time to post a slashdot rant.
What was the topic again?
Oh yes, hacking does this count?
Hold on, the CD's include application software! Are you trying to tell me that windows, office, painshop pro, visual C++ would fit on a single CD! Next you'll be saying SuSE crap because it comes on 6 cds!
True, Why would the 'backup key' survive but not the primary one? Whats more likely is that it will double the time it takes to crack both keys.
And lets face it, its going to be much more fun
cracking these keys cf. the RSA des/RC5 chalenges.
Maybe the bovine lot would care to host such a distributed microsoft attack?
The keys are probberbly copywrite anyway (can you copywrite a key?).
Has anyone extracted the keys, are they plain old des/rc5 or are they something MS/odd?