It doesn't beg the question "how is this news except..."
What you mean to say is, "The real question is how is this news except..."
It begs the question: Is everyone afraid that google will know too much about you by what you search?
The article doesn't presuppose anything, but to someone who is slightly paranoid, they might have been suddenly reminded that google tracks them once that counter appeared, hence the explanatory piece.
Of course, google always tracks everything. That information is used to improve the relevancy metadata they use for providing "similar results", ad-word placement, etc.
The trick is to leverage it in a way to improve the value of google. Maybe a subscription service would expose more of the trends they pick up on as a value added service. I would surely pay for something cool like that.
Here's the rub, bub. Buzzwords fill in dead air.
on
The Smart Sensor Web
·
· Score: 1
any global sensor net will need all of this features (and more) implictly.
No need to state the obvious.
A brief overview of the various technologies and protocols which endeavor to tie everything we have together, and lay groundwork for future developments would have been taken less negatively.
It sounds like work is ongoing, according to the article, and the article pointed to by that article, but no leads or pointers to see the progress for yourself are provided.
At least I can get the name of some field experts, so now I'd have to cross-check them against citation indexes.
Whee, fun. I thought slashdot was supposed to minimize the effort needed to learn and play about new, cool, things, instead of copying speculation in blogs and telling me "trust me, its out there".
if this sensor web ACTUALLY EXISTED, your software might be of interest for someone who was looking to implement a smaller version of this, or to collect and personally mine captured ata.
But, as it stands, these are ethereal wet dreams. And I'm sure your software could improve many times over by that time. So, it's a plug, and not even a very useful one at this time.
In fact, I fail to see how it's really that important. Can you explain why your software is an improvement over any other generic RDBMS using time as an index?
Moreover, time is not the only dimension that requires correlation in a geographically large sensor net. Sensors can move, so a query may need to be split over sensors who passed through a region in question. Time, space, value ranges, set membership, these are all important datum selection criterion.
Of course, you CAN have an equally equipped machine if you're willing to go shopping.
You can do a lot better for a few hundred over $3000 too, like by doubling RAM, or getting the 9800 Pro. It gets much cheaper if you compare an Apple 1.8 to the dual Opteron 244 (246's are disproportionally expensive)
Also Monarch Computers will sell you a dual Opteron workstation, if you want it built and shipped. They're priced about the same (a little more expensive).
Tyan Thunder K85 (S2885) 1GB (2x512) DDR333 (ECC, registered) 2xOpteron 246 (2GHz) DVD-R/RW/CD-R/CD-RW 160GB S-ATA HD Radeon 9600 PRO, even 9700 Pro (misc. retail tower case/supply, optical mice, keyboard...) Windows XP Pro, and by having a copy you can get the 64-bit beta if you sign up.. or Redhat WS, or maybe GinGin, which is free.
All this for less than the base price of $3000, and it's backwards compatible with your older stuff for IA32.
it's not much less, but if they're going to compare the speed or feel of a G5 to anything, it has to be AT LEAST to a rig like that. The trouble is having the same apps with AMD64 support to compare speed of photoshop filters, etc. If that's the basis they want to compare across.
Anyway, there's a bunch of apple haters here talking about how much more it is, and who could afford that.
But the G5 is a huge leap. it really is, deal with it. It's got fucking DDR400 out of the box.
AND YET! For the same price, there's competetion. Where are the _fair_ benchmarks between two equally equiped setups?
-ss works by guessing how many frames to skip by the framerate in the header (but this can change in realtime, and often does, especially during playback if you're using pullup filters).
Meanwhile, -endpos only counts "output" time during playback or encoding.
This means segments corresponding to -ss 0 -endpos 60, -ss 60 -endpos 60, and -ss 120 -endpos 60 are not necessarily back to back 1 minute long chunks, but can arbitrarily overlap.
-ss positioning can be precise, in that you can specify a frame or byte amount to skip. Unfortunately, -endpos is always measured in units of time. This means you can never guarantee a seamless segmentation using those options unless you recode the video into a constant fps format that is de-interlaced FIRST, then segment it.
It would be nice if you could do -endpos as a "secondary" -ss where encoding when stop when a certain point in the input stream is reached. At least, using this method, the segments would be back to back, and guaranteed monotonic no matter how the parameter is used.
Perhaps call it endpos, and call the old, time-based endpos -maxtime?
the MPEG2 video (720x480 resolution at 9 megabits per second) is closer to the digital uncompressed stock at a least-squares error rate an order of magnitude better than MP3s at 128kbps, to use your example.
Have you looked at paused frames from a DVD on a progressive scan TV?
Looks nice, doesn't it.:-)
And (joke approaches) I'll wager the quality of such a DivX is much higher than one taken from some other method, say a telesynced camcorder in a movie theater.
I write command-line utils, no wait, CGI SCRIPTS in C.
because of this, the hodge-podge nature of the command line, and the maddening spottiness of the manpage DRIVES ME INSANE!!!! moreso than it might someone who was comfortable with a command line, and cuts and pastes from examples, maybe with a little experimentation.
But I'm a power user, and I'm grating up against it, and I don't like it.
How can I know that a script that I kick off that encodes a DVD chapter by chapter into ump-teen formats, with offsets the cut off the trailers (but of course -ss and -endpos use DIFFERENT UNITS) and subtitle redirection, and capture AC3 in one place, while stereo second langauge track goes somewhere else, and the video is noisy, so I have to apply video filtering...
Well, christ, I mean, I'll never know if it's working right until I do a few dry runs, and then sometimes I have to tweak the scripts. With ecasound, at least I can structure my command line into sections that I can move around and comment out without so much fuss (even if I have to tweak the scripts)
My biggest gripe is how there's three different de-telecine filters, and each of them has some quirk that always fucks me up.
pullup is nice, but you can't set the output frame rate, nor use 3-pass encoding because the audio gets out of sync (you MUST encode in pass2, and you have to let it pick the output framerate)
detc never works reliably, but at least you can set the framerate.
ivtc, I have no idea what that does, but I never got it to work.
Eh, I'm done bitching.
Using it (to it's fullest extent) is hard. Building it is easy in comparison.
Is not anti-spam vendors, or people who make money on the few hits garnerned by replies (other than Symantec and Learning Tree, but that's a story for another time)
Rather, the majority of spam comes from suckers who bought into get-money-quick and be-your-own-boss internet marketing schemes. These poor schmoes in the US and Asia buy these kits, which may even come with rented rackspace out of the US to mailbomb from and proceed to splatter their wares to these double-opt-in lists in the hopes of making a return on their investment.
Of course, no one is dumb enough to buy any significant amount from one person. They'll keep hammering that list, getting more desperate, trying to "build a customer base" until finally they default on their hosting contract or whatever.
Meanwhile those marketing "gurus" walk away laughing all the way to the bank.
They get joe-credit-card-debt-schmoe to do the dirty work for them.
They don't have to spam or advertise. They just need good placement in google, which isn't too hard to come by these days. The lazy, the "entrepeneurs" will find them, and a fool and his money are soon parted.
And everybody else has to suffer.
It's not as simple as just ignore it, or don't buy the stuff.
Sleazeball marketing gurus will sell you the Brooklyn Bridge and promise you the moon, a 50% response rate if you just use THEIR NEW, IMPROVED SYSTEM
THAT'S THE PROBLEM!!!
And if anyone knows how to fix this, they get the Nobel Peace Prize, I swear to fucking god.
If it sucks, I'd rather have NO gui, then one that makes me want to punch puppies.
So is there a tui_earthstation5? COME ON!
and this bullshit too:
on
Mplayer Revisited
·
· Score: 2, Insightful
want a dvd? You need to use the "dvd" URL
dvd://title#
But this syntax
dvd://title#/chapter#
Doesn't work, you need this: dvd://title# -chapter chapter#
Which is more typing than: -dvd title# -chapter chapter#
And filters for -vop are applied IN REVERSE ORDER.
How about this malarky:
-vop detc=dr=1:ar=0,denoise3d
commas distribute over colon, colon over equals, except for the first equals that shows a filter has options.
urrggghh...
Oh, and the syntax is horrible just in general. Some options only take effect when they come before or after certain things, certain ones depend on other options in weird ways (video filters, codecs, and -fps/-ofps hell).
Still, I love mplayer. Who am I kidding. I just way too much time trying to figure out how to do things I KNOW should work, I just can't get a handle on it.
ecasound, while having also having an insanely rich command line, is more logical.
By the time we have a decent language processing engine to do generic searches and reporting on documents with semantic relevance, we should have developed a nicer interface than voice recognition, which has lots of practical flaws.
I'm thinking iris or cerebral scans. You "think" about what you want of the computer, and it responsds accordingly. Voice output might be desirable, but voice input limits utility.
People should stop limiting their views of the future of technology to the realm of popular science fiction. I this particular instance was for the media, and the CTO is trying to be an effective communicator, so he has to put his future plans into context, but many of the replies to this article (on Slashdot!) are underwhelming.
It's all heavily geared towards Windows (DirectX 9 and all), so if you could afford the toolchain, you could probably afford to see the source code after signing an NDA or giving up your first child or something.
I don't see how the GPL is desirable or applicable here.
it understands all 4 Unicode encodings, can recognize east asian ideograms for digits and power of ten grouping, and even understands that whole french thing with the period and comma being switched (WTF IS WITH THAT ANYWAY???)
But you must remember, SYSV is quite old and was released into the public domain.
SVR4 is not. Since the code beared some resemblance to the ancestor of SVR4, it is a good idea to remove it to avoid any confusion, especially if it was redundant.
you should surround that system with a fork/forget, and add a timeout to the wget (or use curl with "-m #") so that you don't spawn off too many. (make sure to reap children in the spawning loop).
I'm sure a machine on a broadband connection could spawn 40 requests in the time it takes one to reply.
Finally, you should really consider not using the bare URLs in the request because all you really do is hammer your local DNS server.
instead, make fake requests DIRECTLY to sitefinder-idn.verisign.com. To see what the request ends up looking like, run wget with headers turned on with one of those fake queries.
Then duplicate the request but substitute that static IP address. Your ISP won't think you're DoS'ing their DNS.
Slashdot Mirror
He's got my vote.
for starters, you can talk to this guy
It doesn't beg the question "how is this news except..."
What you mean to say is, "The real question is how is this news except..."
It begs the question: Is everyone afraid that google will know too much about you by what you search?
The article doesn't presuppose anything, but to someone who is slightly paranoid, they might have been suddenly reminded that google tracks them once that counter appeared, hence the explanatory piece.
Of course, google always tracks everything. That information is used to improve the relevancy metadata they use for providing "similar results", ad-word placement, etc.
The trick is to leverage it in a way to improve the value of google. Maybe a subscription service would expose more of the trends they pick up on as a value added service. I would surely pay for something cool like that.
any global sensor net will need all of this features (and more) implictly.
No need to state the obvious.
A brief overview of the various technologies and protocols which endeavor to tie everything we have together, and lay groundwork for future developments would have been taken less negatively.
It sounds like work is ongoing, according to the article, and the article pointed to by that article, but no leads or pointers to see the progress for yourself are provided.
At least I can get the name of some field experts, so now I'd have to cross-check them against citation indexes.
Whee, fun. I thought slashdot was supposed to minimize the effort needed to learn and play about new, cool, things, instead of copying speculation in blogs and telling me "trust me, its out there".
if this sensor web ACTUALLY EXISTED, your software might be of interest for someone who was looking to implement a smaller version of this, or to collect and personally mine captured ata.
But, as it stands, these are ethereal wet dreams. And I'm sure your software could improve many times over by that time.
So, it's a plug, and not even a very useful one at this time.
In fact, I fail to see how it's really that important. Can you explain why your software is an improvement over any other generic RDBMS using time as an index?
Moreover, time is not the only dimension that requires correlation in a geographically large sensor net. Sensors can move, so a query may need to be split over sensors who passed through a region in question. Time, space, value ranges, set membership, these are all important datum selection criterion.
Of course, you CAN have an equally equipped machine if you're willing to go shopping.
You can do a lot better for a few hundred over $3000 too, like by doubling RAM, or getting the 9800 Pro.
It gets much cheaper if you compare an Apple 1.8 to the dual Opteron 244 (246's are disproportionally expensive)
Also Monarch Computers will sell you a dual Opteron workstation, if you want it built and shipped. They're priced about the same (a little more expensive).
Tyan Thunder K85 (S2885)
1GB (2x512) DDR333 (ECC, registered)
2xOpteron 246 (2GHz)
DVD-R/RW/CD-R/CD-RW
160GB S-ATA HD
Radeon 9600 PRO, even 9700 Pro
(misc. retail tower case/supply, optical mice, keyboard...)
Windows XP Pro, and by having a copy you can get the 64-bit beta if you sign up.. or
Redhat WS, or maybe GinGin, which is free.
All this for less than the base price of $3000, and it's backwards compatible with your older stuff for IA32.
it's not much less, but if they're going to compare the speed or feel of a G5 to anything, it has to be AT LEAST to a rig like that.
The trouble is having the same apps with AMD64 support to compare speed of photoshop filters, etc. If that's the basis they want to compare across.
Anyway, there's a bunch of apple haters here talking about how much more it is, and who could afford that.
But the G5 is a huge leap. it really is, deal with it. It's got fucking DDR400 out of the box.
AND YET! For the same price, there's competetion. Where are the _fair_ benchmarks between two equally equiped setups?
-ss works by guessing how many frames to skip by the framerate in the header (but this can change in realtime, and often does, especially during playback if you're using pullup filters).
Meanwhile, -endpos only counts "output" time during playback or encoding.
This means segments corresponding to -ss 0 -endpos 60, -ss 60 -endpos 60, and -ss 120 -endpos 60 are not necessarily back to back 1 minute long chunks, but can arbitrarily overlap.
-ss positioning can be precise, in that you can specify a frame or byte amount to skip. Unfortunately, -endpos is always measured in units of time. This means you can never guarantee a seamless segmentation using those options unless you recode the video into a constant fps format that is de-interlaced FIRST, then segment it.
It would be nice if you could do -endpos as a "secondary" -ss where encoding when stop when a certain point in the input stream is reached. At least, using this method, the segments would be back to back, and guaranteed monotonic no matter how the parameter is used.
Perhaps call it endpos, and call the old, time-based endpos -maxtime?
Love the nick. Are you from PRC?
the MPEG2 video (720x480 resolution at 9 megabits per second) is closer to the digital uncompressed stock at a least-squares error rate an order of magnitude better than MP3s at 128kbps, to use your example.
:-)
Have you looked at paused frames from a DVD on a progressive scan TV?
Looks nice, doesn't it.
And (joke approaches) I'll wager the quality of such a DivX is much higher than one taken from some other method, say a telesynced camcorder in a movie theater.
thanks for doing a much better job explaining than I could ever do.
I'm the hugest linux geek there is.
I write command-line utils, no wait, CGI SCRIPTS in C.
because of this, the hodge-podge nature of the command line, and the maddening spottiness of the manpage DRIVES ME INSANE!!!! moreso than it might someone who was comfortable with a command line, and cuts and pastes from examples, maybe with a little experimentation.
But I'm a power user, and I'm grating up against it, and I don't like it.
How can I know that a script that I kick off that encodes a DVD chapter by chapter into ump-teen formats, with offsets the cut off the trailers (but of course -ss and -endpos use DIFFERENT UNITS) and subtitle redirection, and capture AC3 in one place, while stereo second langauge track goes somewhere else, and the video is noisy, so I have to apply video filtering...
Well, christ, I mean, I'll never know if it's working right until I do a few dry runs, and then sometimes I have to tweak the scripts.
With ecasound, at least I can structure my command line into sections that I can move around and comment out without so much fuss (even if I have to tweak the scripts)
My biggest gripe is how there's three different de-telecine filters, and each of them has some quirk that always fucks me up.
pullup is nice, but you can't set the output frame rate, nor use 3-pass encoding because the audio gets out of sync (you MUST encode in pass2, and you have to let it pick the output framerate)
detc never works reliably, but at least you can set the framerate.
ivtc, I have no idea what that does, but I never got it to work.
Eh, I'm done bitching.
Using it (to it's fullest extent) is hard. Building it is easy in comparison.
Is not anti-spam vendors, or people who make money on the few hits garnerned by replies (other than Symantec and Learning Tree, but that's a story for another time)
Rather, the majority of spam comes from suckers who bought into get-money-quick and be-your-own-boss internet marketing schemes. These poor schmoes in the US and Asia buy these kits, which may even come with rented rackspace out of the US to mailbomb from and proceed to splatter their wares to these double-opt-in lists in the hopes of making a return on their investment.
Of course, no one is dumb enough to buy any significant amount from one person. They'll keep hammering that list, getting more desperate, trying to "build a customer base" until finally they default on their hosting contract or whatever.
Meanwhile those marketing "gurus" walk away laughing all the way to the bank.
They get joe-credit-card-debt-schmoe to do the dirty work for them.
They don't have to spam or advertise. They just need good placement in google, which isn't too hard to come by these days. The lazy, the "entrepeneurs" will find them, and a fool and his money are soon parted.
And everybody else has to suffer.
It's not as simple as just ignore it, or don't buy the stuff.
Sleazeball marketing gurus will sell you the Brooklyn Bridge and promise you the moon, a 50% response rate if you just use THEIR NEW, IMPROVED SYSTEM
THAT'S THE PROBLEM!!!
And if anyone knows how to fix this, they get the Nobel Peace Prize, I swear to fucking god.
If it sucks, I'd rather have NO gui, then one that makes me want to punch puppies.
So is there a tui_earthstation5? COME ON!
want a dvd? You need to use the "dvd" URL
dvd://title#
But this syntax
dvd://title#/chapter#
Doesn't work, you need this:
dvd://title# -chapter chapter#
Which is more typing than: -dvd title# -chapter chapter#
And filters for -vop are applied IN REVERSE ORDER.
How about this malarky:
-vop detc=dr=1:ar=0,denoise3d
commas distribute over colon, colon over equals, except for the first equals that shows a filter has options.
urrggghh...
Oh, and the syntax is horrible just in general. Some options only take effect when they come before or after certain things, certain ones depend on other options in weird ways (video filters, codecs, and -fps/-ofps hell).
Still, I love mplayer. Who am I kidding. I just way too much time trying to figure out how to do things I KNOW should work, I just can't get a handle on it.
ecasound, while having also having an insanely rich command line, is more logical.
By the time we have a decent language processing engine to do generic searches and reporting on documents with semantic relevance, we should have developed a nicer interface than voice recognition, which has lots of practical flaws.
I'm thinking iris or cerebral scans. You "think" about what you want of the computer, and it responsds accordingly. Voice output might be desirable, but voice input limits utility.
People should stop limiting their views of the future of technology to the realm of popular science fiction. I this particular instance was for the media, and the CTO is trying to be an effective communicator, so he has to put his future plans into context, but many of the replies to this article (on Slashdot!) are underwhelming.
What's in there anyway? Server source code only?
It's all heavily geared towards Windows (DirectX 9 and all), so if you could afford the toolchain, you could probably afford to see the source code after signing an NDA or giving up your first child or something.
I don't see how the GPL is desirable or applicable here.
And I have signed you up for an embarassing number of pornographic magazines and movie clubs.
I do this so to help you fulfill your dreamland fantasy of being a playa.
it understands all 4 Unicode encodings, can recognize east asian ideograms for digits and power of ten grouping, and even understands that whole french thing with the period and comma being switched (WTF IS WITH THAT ANYWAY???)
SGI found a few bugs during the review process and turned them over to quality assurance.
:-)
It probably wasn't a total waste.
But you must remember, SYSV is quite old and was released into the public domain.
SVR4 is not. Since the code beared some resemblance to the ancestor of SVR4, it is a good idea to remove it to avoid any confusion, especially if it was redundant.
you should surround that system with a fork/forget, and add a timeout to the wget (or use curl with "-m #") so that you don't spawn off too many. (make sure to reap children in the spawning loop).
I'm sure a machine on a broadband connection could spawn 40 requests in the time it takes one to reply.
Finally, you should really consider not using the bare URLs in the request because all you really do is hammer your local DNS server.
instead, make fake requests DIRECTLY to sitefinder-idn.verisign.com. To see what the request ends up looking like, run wget with headers turned on with one of those fake queries.
Then duplicate the request but substitute that static IP address. Your ISP won't think you're DoS'ing their DNS.
Timely jokes that are insulting to the editor, homophobic in tone, and on-topic all at once.
This is the type of troll that makes browsing slashdot at -1 fun.
Therefore: BSD is dying, QED.
Wait, now I'm confused. What was this analogy about?