Maybe you *should* study "EE", or rather software engineering
This is an Ad Hominem argument, a logical fallacy. And for the record, the 1.8 million downloads of my top100 project at sourceforge say that I'm doing just fine as a software engineer. Moving on, you claim that:
My entire post is about a redesign to get it to be, or rather, to get something similar to BT to be that protocol.
However, the introductory sentence to the original post reads:
The real breakthru for distributed P2P tech will come when someone publishes a BitTorrent content distributor that can be plugged transparently in front of an HTTPD
And I've simply explained why it's a bad idea to use BitTorrent. I didn't say it was a bad idea to use a custom system, it's just that isn't what I got from reading your original post. Free advice: miscommunications happen, and you really should try not to take things so personally, you sorta come off as a raving loon.
Maybe you don't even need EE training - just (dare I say it) some marketing experience.
I've been first to market on a number of new technologies (mp3, p2p, etc..) and millions of people have used and continue to use my products. My marketing is working out just for me...
PS: Bram Cohen is the software engineer, as far as I'm concerned he's the world's leading authority on p2p right now. I try to follow along, understand how it all works, and release some useful tools for the public.
Disclaimer: When i'm not studying EE, IAABCD (I _am_ a BitTorrent Client Developer).
Why does this keep coming up? no, No and NO!
3 reasons why BT is not suited to be a web cache protocol:
- BT is designed and optimized for LARGE amounts (on the order of hunderds of MB) of data. Web sites are comparatively tiny.
- BT will only work with STATIC content.. files are hashes at.torrent creation time, and those hashes are now set in stone. Change those, and you've created a new swarm. It's stupid to even think about trying to distribute any kind of dynamic content over BT.
- BT relies on the fact that some people will keep their transfers going to upload to others.. and users never stay on a particular website for long.
When one researches using tent material for pants, one does not label the study as one about tents.. and this is what's happened here. This is research into ONE particular website's implementation of a tracker and load distribution system (there are many other sites that do it differently), but yet they have labelled it as a "BitTorrent" study.
The main problem here is that some of the conclusions the author comes to about "BitTorrent" have nothing to do with BitTorrent the protocol.. and most of the problems are due to having to find webhosts to host high-demand semi-illegal material (there was a figure of something like ~250 torrent hosts over the period of research..)
A much more informative study (such as the one going on over at MIT, sorry can't find a link atm) would examine BitTorrent's effectiveness as applied to the field it was designed to be used in.. reducing server load when distributing large files.
Something from the article rubbed me the wrong way:
"Spyware usually gets on your computer through human error," said Marc Maiffret of eEye Digital Security Inc., which regularly discovers serious Windows flaws.
First.. a confession: My name is kRYPT, and I used to use Internet Explorer. I used to keep it patched, and updated. I browsed on High Security. I ran Spybot S&D and Adaware regularly, and TeaTimer always.
Spyware STILL got in. Every Spybot scan would regularly reveal something nasty (normally DSO or other IE Exploits).
Perhaps it's true that most Spyware is the result of user action (such as installing shady "free" smiley-enhancing software), but _lots_ of the Spyware out there is simply a direct result of using IE.
PS: I see the spyware people are trying to attack Firefox too.. see cracks.am for an example. However, in Firefox, a nice dialog pops up, makes it perfectly clear the code that's being requested to run is unsigned and unvalidated, and makes you wait for 2 seconds before you have the chance to accept or deny installing it.
If you're going to pass copyright off as bullshit, then you need to pass ownership of physical goods off as an artificial construct.
Hmm.. I don't think there's anything artifical about ownership of physical things. People have things that they consider theirs, and if anyone challenges them then they will fight for it. That's about as natural as it gets..
Our entire system of laws is based around artificial constructs. For example, "rights" are an artificial construct.
Ah, yes. I completely agree with you. Some people way back when decided that hey, lets not go around fighting with each other, and instead agree on some things that we all shouldn't do! Rights are just what we came up with,as a list of things that you shouldn't do to someone..
Copyright law benefits the copyright owners. That's abundantly clear. However, in the United States, commerce benefits us all.
Well.. to recap, it was Disney that lobbied to have the copyrights extended as far as it has been. Disney based a large portion of their work on works from the Public Domain, and then lobbied to have it killed.
Logicly, the point of extending copyrights was to keep those works in the marketplace -- it creates jobs and tax revenues and allows for government funded infrastructure.
Was it? Really? Or was it to prevent those ideas from becoming free, as they should have been?
I've used Ultr@VNC extensively from school (OC3) to home (cablemodem).
Even with my cablemodem's poor upload (45k/sec), it was very usable. If really you think "VNC sends bitmaps", then you're about 2 years behind.. VNC will fall back to bitmaps (JPEGs actually) only if the region of the screen won't compress better using any of it's other built-in compression algorithms (which work great most of the time).
"Copyright Infringement" is an artifical construct. Copyright was originally created to give the original author the opportunity to benifit from their work, before releasing it to the public domain in 15 years time to benifit everyone, not just the person who created it.
The current legislated "Life+70" essentially guarantees that you will not see any of the currenttly available content enter the public domain within your lifetime.
Thus, the "benifit everyone" part of Copyright has been lost. And if it doesn't benifit everyone, then you have to ask who does it really benifit?
I agree with you, it's an issue of right and wrong. And Copyright is wrong.
If you sum an infinite (very large) number of terms, the Taylor/MacLaurin/Laurent series of a function will converege exactly at the point that you're expanding it around.
The problem is that the infinite-series expantions of trig functions given in textbooks (sin being the sum of odd powers of x, cos being the sum of even powers of x, alternating sign on each term, and each term divided by x's power factorial) are expanded around x=0.
However, as long as you're using the right approxomation (centered around a point near the one you're interested in; this would likely require you knew at least 3 series: centered at 0, -pi/2 and pi/2) they're good enough;)
(Math exam is on saturday, complex algebra course, deals heavily with infinite series expansions of functions)
I think 512x384 is the minimum resolution a portable DivX/XVid player needs to support to be useful.. as far as 640x480 would be optimal. A recent build of XVid at around 1mbit looks great at those resolutions, or around 141 kb/sec with 128kbit audio. This gives roughly 157.6 hours in 80GB or 80-100 movies.
Acrobat 6 would lock up Firefox hard every time I tried to close or exit the page with it embedded (closing the tab, going back to another page, and closing the window all had the same effect.. 100% cpu use until I killed the acrobat thread).
Next time I re-installed the system, I installed Acrobat Reader 5 (and FreePDF to make up for the lack of publishing ability) and all has been well since.
It's not. Trackers, and especially popular ones, are (almost) never run on windows machines due to reliability and speed issues, and I was under the impression PG was a windows only thing..
I don't see how this is useful in a BitTorrent context.
The *AA can scrape the tracker to get all the info about you they need. There is no technical reason they need to connect to you, so unless there's a legal one, I don't see how PG would be useful here.
Just how difficult is it to block out this spyware, anyway? Can't you just patch the source, or edit the Makefile or whatever Windows uses in place of that, so the spyware portions don't even get compiled? Or do Windows downloads work somehow totally different to Linux and BSD ones?
Yes, they work totally different.
As the author of a P2P client, I have now been approached 3 times by these sleezeballs:
1) The first time, some "media company" had taken my software, renamed it "KaZaa and Morpheus Enhancer", bundled their insane spyware that would play full-screen-full-motion-video ads and tried to submit it to download.com! Fortunately, the nice folks at download.com did their research, saw my name and e-mail in the About box, and contacted me. I received an e-mail from CEO of said company aplogizing, saying that his "R&D" people said they had "invented" the technology.. and how much could he pay me to let me go ahead with it? I told him that my users would lynch me, and I wasn't interested.
(Summary: Source was basically unaltered, spyware shop replaced my NSIS installer with their own)
2) Got an e-mail from a spyware company directly, making me an offer to bundle their garbage for $100/mo..
(Summary: Source was again unaltered, spyware shop asked maintainer to add their spyware to installer)
3) Got an e-mail from a guy who was interested in "purchasing the code" for my software.. he said even "old versions" were acceptable. I told him I wasn't intersted, and didn't mention that fact that it's GPL and source is freely available on the homepage.
(Summary: This one would be some combination of source-violation and installer-violation).
There you have it. Spyware is never "compiled", it's always provided as blackbox.exe or.dll, and most applications that willingly bundle it (my scneario #2 above) will check for it's presence at startup, and refuse to work w/o it.
"I think a killer feature, that does a somewhat decentralized bittorrent, would be for the protocol to dynamicly failover if the tracker goes down, so that a seeder would assume reponsibility(or several would)."
This is simply not possible; let me explain:
A.torrent contains 3 things: a) tracker, b) filenames, c) hashes. Without all 3 of the above, BT cannot operate.
The purpose of the tracker is to give a new peer joining the swarm a __random subset__ (-- very important to BT's speed) of all known peers' IPs.
Even if you could somehow figure out a way to create a new.torrent when the tracker goes down, and push this new torrent (with the new "tracker") to ALL the peers (this is required so that NAT'd people can join the swarm), no peer could ever know about every other peer in a relatively large (200+ peers) swarm.
Thus, you would inevitably end up, at BEST, segmenting the swarm into 2+ pieces.. that is assuming you can find a way to push these new torrents out to clients already transferring the file.
"But kRYPT", you say,"why not just broadcast a message every time the tracker changes?"
Becuase you cannot trust clients. Who decided which client becomes a tracker? What's to prevent half the clients in the swarm from being "trackers"? On slow connections? What's to prevent a rogue client-gone-tracker from giving out false IPs? (I can see the RIAA being all over this idea)
To recap, two things are essential to the way BT operates:
- The there is a central, reliable, TRUSTED, source of IPs in each swarm - and that it knows about -EVERY- peer involved in the swarm.
Any and all attempts to "decentralize" BitTorrent will fail.. and will, at best, end up implementing eDonkey.
"With trackers, you can not. Not yet. Because the protocol doesn't support it."
You're wrong on four counts:
1) Yes, you CAN do this with trackers. 2) Right now 3) Because the protocol (unofficially) supports it 4) Some groups (such as BTEfnet) are doing this now, without relying on the above spec.
They're using the exact same technique used by http servers, DNS load balancing. A tracker is just a special case of an http server.
Pfft, the whole low-noise thing is over rated if you ask me.
I have an Xbox with the 12V fan mod done (I put in a larger, faster, hotter HD and was worried about airflow). It sounds noticably loader then the stock xbox fan, particularly when starting up.. but once the movie is on, nobody can hear it.
Mind you, the xbox is right by the (HD)TV.. those cables don't have very good range. The whole setup is a good 6-8 ft away from the couch, depending on where you are.
"A paid membership will allow for access to the range viewing camera(s) at any time. Members can then schedule a reserved session time which allows exclusive control of the shooting system to fire at a choice of various reactive targets."
"At all times during a shooting session, someone is at the shooting station and is available to answer questions (e-mail, instant messaging), provide assistance, and ensure a quality experience. This person has the ability to override the firing mechanism of the system to minimize the chance of a dangerous or illegal discharge from occurring."
The problem here is that yes, in hindsight, what these people have accomplished is obvious, and not really a big deal.
But hindsight, as they say, is 20/20.. foresight, not so much. It's very difficult to come out with the best thing since sliced bread, and these people deserve their share of credit for shaping the internet as we know it.
This is OT, but you may want to look at FreePDF XP.
Don't let the german download page scare you, the program is in English.. Adds a printer just like Acrobat, and works very well (sometimes better then Acrobat itself).
I think you're looking at it all wrong.. Right Now is the time to buy the computer you wanted 2 years ago.. which would be the exact same computer that was twice as fast as 4 years ago!
Slashdot Mirror
Maybe you *should* study "EE", or rather software engineering
This is an Ad Hominem argument, a logical fallacy. And for the record, the 1.8 million downloads of my top100 project at sourceforge say that I'm doing just fine as a software engineer. Moving on, you claim that:
My entire post is about a redesign to get it to be, or rather, to get something similar to BT to be that protocol.
However, the introductory sentence to the original post reads:
The real breakthru for distributed P2P tech will come when someone publishes a BitTorrent content distributor that can be plugged transparently in front of an HTTPD
And I've simply explained why it's a bad idea to use BitTorrent. I didn't say it was a bad idea to use a custom system, it's just that isn't what I got from reading your original post. Free advice: miscommunications happen, and you really should try not to take things so personally, you sorta come off as a raving loon.
Maybe you don't even need EE training - just (dare I say it) some marketing experience.
I've been first to market on a number of new technologies (mp3, p2p, etc..) and millions of people have used and continue to use my products. My marketing is working out just for me...
PS: Bram Cohen is the software engineer, as far as I'm concerned he's the world's leading authority on p2p right now. I try to follow along, understand how it all works, and release some useful tools for the public.
Disclaimer: When i'm not studying EE, IAABCD (I _am_ a BitTorrent Client Developer).
.torrent creation time, and those hashes are now set in stone. Change those, and you've created a new swarm. It's stupid to even think about trying to distribute any kind of dynamic content over BT.
Why does this keep coming up? no, No and NO!
3 reasons why BT is not suited to be a web cache protocol:
- BT is designed and optimized for LARGE amounts (on the order of hunderds of MB) of data. Web sites are comparatively tiny.
- BT will only work with STATIC content.. files are hashes at
- BT relies on the fact that some people will keep their transfers going to upload to others.. and users never stay on a particular website for long.
When one researches using tent material for pants, one does not label the study as one about tents.. and this is what's happened here. This is research into ONE particular website's implementation of a tracker and load distribution system (there are many other sites that do it differently), but yet they have labelled it as a "BitTorrent" study.
The main problem here is that some of the conclusions the author comes to about "BitTorrent" have nothing to do with BitTorrent the protocol.. and most of the problems are due to having to find webhosts to host high-demand semi-illegal material (there was a figure of something like ~250 torrent hosts over the period of research..)
A much more informative study (such as the one going on over at MIT, sorry can't find a link atm) would examine BitTorrent's effectiveness as applied to the field it was designed to be used in.. reducing server load when distributing large files.
Something from the article rubbed me the wrong way:
"Spyware usually gets on your computer through human error," said Marc Maiffret of eEye Digital Security Inc., which regularly discovers serious Windows flaws.
First.. a confession: My name is kRYPT, and I used to use Internet Explorer. I used to keep it patched, and updated. I browsed on High Security. I ran Spybot S&D and Adaware regularly, and TeaTimer always.
Spyware STILL got in. Every Spybot scan would regularly reveal something nasty (normally DSO or other IE Exploits).
Perhaps it's true that most Spyware is the result of user action (such as installing shady "free" smiley-enhancing software), but _lots_ of the Spyware out there is simply a direct result of using IE.
PS: I see the spyware people are trying to attack Firefox too.. see cracks.am for an example. However, in Firefox, a nice dialog pops up, makes it perfectly clear the code that's being requested to run is unsigned and unvalidated, and makes you wait for 2 seconds before you have the chance to accept or deny installing it.
If you're going to pass copyright off as bullshit, then you need to pass ownership of physical goods off as an artificial construct.
Hmm.. I don't think there's anything artifical about ownership of physical things. People have things that they consider theirs, and if anyone challenges them then they will fight for it. That's about as natural as it gets..
Our entire system of laws is based around artificial constructs. For example, "rights" are an artificial construct.
Ah, yes. I completely agree with you. Some people way back when decided that hey, lets not go around fighting with each other, and instead agree on some things that we all shouldn't do! Rights are just what we came up with,as a list of things that you shouldn't do to someone..
Copyright law benefits the copyright owners. That's abundantly clear. However, in the United States, commerce benefits us all.
Well.. to recap, it was Disney that lobbied to have the copyrights extended as far as it has been. Disney based a large portion of their work on works from the Public Domain, and then lobbied to have it killed.
Logicly, the point of extending copyrights was to keep those works in the marketplace -- it creates jobs and tax revenues and allows for government funded infrastructure.
Was it? Really? Or was it to prevent those ideas from becoming free, as they should have been?
I've used Ultr@VNC extensively from school (OC3) to home (cablemodem).
Even with my cablemodem's poor upload (45k/sec), it was very usable. If really you think "VNC sends bitmaps", then you're about 2 years behind.. VNC will fall back to bitmaps (JPEGs actually) only if the region of the screen won't compress better using any of it's other built-in compression algorithms (which work great most of the time).
"Copyright Infringement" is an artifical construct. Copyright was originally created to give the original author the opportunity to benifit from their work, before releasing it to the public domain in 15 years time to benifit everyone, not just the person who created it.
The current legislated "Life+70" essentially guarantees that you will not see any of the currenttly available content enter the public domain within your lifetime.
Thus, the "benifit everyone" part of Copyright has been lost. And if it doesn't benifit everyone, then you have to ask who does it really benifit?
I agree with you, it's an issue of right and wrong. And Copyright is wrong.
The squeeze theorem is taught in 2nd year calculus, used to evaluate limits of some functions..
Well.. yes, and no.
;)
If you sum an infinite (very large) number of terms, the Taylor/MacLaurin/Laurent series of a function will converege exactly at the point that you're expanding it around.
The problem is that the infinite-series expantions of trig functions given in textbooks (sin being the sum of odd powers of x, cos being the sum of even powers of x, alternating sign on each term, and each term divided by x's power factorial) are expanded around x=0.
However, as long as you're using the right approxomation (centered around a point near the one you're interested in; this would likely require you knew at least 3 series: centered at 0, -pi/2 and pi/2) they're good enough
(Math exam is on saturday, complex algebra course, deals heavily with infinite series expansions of functions)
I think 512x384 is the minimum resolution a portable DivX/XVid player needs to support to be useful.. as far as 640x480 would be optimal. A recent build of XVid at around 1mbit looks great at those resolutions, or around 141 kb/sec with 128kbit audio. This gives roughly 157.6 hours in 80GB or 80-100 movies.
Works fine for me? Fast too..
Acrobat 6 would lock up Firefox hard every time I tried to close or exit the page with it embedded (closing the tab, going back to another page, and closing the window all had the same effect.. 100% cpu use until I killed the acrobat thread).
Next time I re-installed the system, I installed Acrobat Reader 5 (and FreePDF to make up for the lack of publishing ability) and all has been well since.
It's not. Trackers, and especially popular ones, are (almost) never run on windows machines due to reliability and speed issues, and I was under the impression PG was a windows only thing..
I don't see how this is useful in a BitTorrent context.
The *AA can scrape the tracker to get all the info about you they need. There is no technical reason they need to connect to you, so unless there's a legal one, I don't see how PG would be useful here.
Just how difficult is it to block out this spyware, anyway? Can't you just patch the source, or edit the Makefile or whatever Windows uses in place of that, so the spyware portions don't even get compiled? Or do Windows downloads work somehow totally different to Linux and BSD ones?
.exe or .dll, and most applications that willingly bundle it (my scneario #2 above) will check for it's presence at startup, and refuse to work w/o it.
Yes, they work totally different.
As the author of a P2P client, I have now been approached 3 times by these sleezeballs:
1) The first time, some "media company" had taken my software, renamed it "KaZaa and Morpheus Enhancer", bundled their insane spyware that would play full-screen-full-motion-video ads and tried to submit it to download.com! Fortunately, the nice folks at download.com did their research, saw my name and e-mail in the About box, and contacted me. I received an e-mail from CEO of said company aplogizing, saying that his "R&D" people said they had "invented" the technology.. and how much could he pay me to let me go ahead with it? I told him that my users would lynch me, and I wasn't interested.
(Summary: Source was basically unaltered, spyware shop replaced my NSIS installer with their own)
2) Got an e-mail from a spyware company directly, making me an offer to bundle their garbage for $100/mo..
(Summary: Source was again unaltered, spyware shop asked maintainer to add their spyware to installer)
3) Got an e-mail from a guy who was interested in "purchasing the code" for my software.. he said even "old versions" were acceptable. I told him I wasn't intersted, and didn't mention that fact that it's GPL and source is freely available on the homepage.
(Summary: This one would be some combination of source-violation and installer-violation).
There you have it. Spyware is never "compiled", it's always provided as blackbox
"I think a killer feature, that does a somewhat decentralized bittorrent, would be for the protocol to dynamicly failover if the tracker goes down, so that a seeder would assume reponsibility(or several would)."
.torrent contains 3 things: a) tracker, b) filenames, c) hashes. Without all 3 of the above, BT cannot operate.
.torrent when the tracker goes down, and push this new torrent (with the new "tracker") to ALL the peers (this is required so that NAT'd people can join the swarm), no peer could ever know about every other peer in a relatively large (200+ peers) swarm.
This is simply not possible; let me explain:
A
The purpose of the tracker is to give a new peer joining the swarm a __random subset__ (-- very important to BT's speed) of all known peers' IPs.
Even if you could somehow figure out a way to create a new
Thus, you would inevitably end up, at BEST, segmenting the swarm into 2+ pieces.. that is assuming you can find a way to push these new torrents out to clients already transferring the file.
"But kRYPT", you say,"why not just broadcast a message every time the tracker changes?"
Becuase you cannot trust clients. Who decided which client becomes a tracker? What's to prevent half the clients in the swarm from being "trackers"? On slow connections? What's to prevent a rogue client-gone-tracker from giving out false IPs? (I can see the RIAA being all over this idea)
To recap, two things are essential to the way BT operates:
- The there is a central, reliable, TRUSTED, source of IPs in each swarm
- and that it knows about -EVERY- peer involved in the swarm.
Any and all attempts to "decentralize" BitTorrent will fail.. and will, at best, end up implementing eDonkey.
"With trackers, you can not. Not yet. Because the protocol doesn't support it."
You're wrong on four counts:
1) Yes, you CAN do this with trackers.
2) Right now
3) Because the protocol (unofficially) supports it
4) Some groups (such as BTEfnet) are doing this now, without relying on the above spec.
They're using the exact same technique used by http servers, DNS load balancing. A tracker is just a special case of an http server.
Oh c'mon..if you're gonna make fun of my client (burst!) not having any market share, you may as well do it correctly ;)
Sounds like you need BugMeNot .. comes in IE, Firefox and anything-supporting-html flavours.
Pfft, the whole low-noise thing is over rated if you ask me.
.. but once the movie is on, nobody can hear it.
.. those cables don't have very good range. The whole setup is a good 6-8 ft away from the couch, depending on where you are.
I have an Xbox with the 12V fan mod done (I put in a larger, faster, hotter HD and was worried about airflow). It sounds noticably loader then the stock xbox fan, particularly when starting up
Mind you, the xbox is right by the (HD)TV
This has been brought up *so* many times on the BT mailing list, and always get shot down.
BT swarms are optmized for large amounts of static content... they are terrible for small amounts of dynamic data (ie, websites).
From site:
"A paid membership will allow for access to the range viewing camera(s) at any time. Members can then schedule a reserved session time which allows exclusive control of the shooting system to fire at a choice of various reactive targets."
"At all times during a shooting session, someone is at the shooting station and is available to answer questions (e-mail, instant messaging), provide assistance, and ensure a quality experience. This person has the ability to override the firing mechanism of the system to minimize the chance of a dangerous or illegal discharge from occurring."
The problem here is that yes, in hindsight, what these people have accomplished is obvious, and not really a big deal.
.. foresight, not so much. It's very difficult to come out with the best thing since sliced bread, and these people deserve their share of credit for shaping the internet as we know it.
But hindsight, as they say, is 20/20
This is OT, but you may want to look at FreePDF XP .
Don't let the german download page scare you, the program is in English.. Adds a printer just like Acrobat, and works very well (sometimes better then Acrobat itself).
I think you're looking at it all wrong..
Right Now is the time to buy the computer you wanted 2 years ago.. which would be the exact same computer that was twice as fast as 4 years ago!