So, if the ISPs do traffic shaping "to improve the service" it's bad, but we admit that on the small scale (when it affects ourselfs) there is a real need for traffic shaping!
I don't mind traffic shaping at all, anywhere. QoS is a good thing, even when the ISPs do it. What I mind a whole awful lot is traffic blocking, ala Comcast.
That's not how traffic shaping works, it's an instantaneous decision. If you have a link that can do 10 Gbit/sec, and 12Gbit/sec is trying to get through, you can't just deprioritize the bulk transfer stuff and have it all somehow get through.
Sure you can, unless you're really so oversold that the baseline interactive traffic is greater than your capacity, in which case nothing will help but extra capacity. After all, traffic is prioritized everywhere; the only difference is that the default metric is packet arrival order.
I still think this is infinitely preferable to shutting down P2P altogether. Shaping the traffic at least lets everyone keep using their Internet connection any way they want. The only difference is that huge P2P downloads don't interfere with VOIP or other more time-sensitive traffic.
I would have a huge problem with that, as I know for a fact many others would.
Why? Do you also run every process on your system at nice=0, including stuff like Folding@home, or do you prioritize so that your desktop stuff isn't unresponsive while still allowing the background tasks to run at full speed?
The distinguishing characteristic is interactivity. Stuff that a human is waiting for should go as fast as possible. Everything else should run unhindered after the latency-sensitive packets get through.
Impeding isn't "blocking". You could try a dictionary:
impede To retard or obstruct the progress of.
Take your own advice. A definition of "block":
obstruct: block passage through; "obstruct the path"
Sounds damn near synonymous. Anyway, QOS is not the same as blocking or impeding in any way. With QOS, all the packets get through, just not at the expense of other traffic.
And then, as I said, very quickly P2P apps will start to mimic or run over "highly interactive traffic" so as not to be slowed down.
What would the advantage be? Why would the Bittorrent (or whoever) devs want to do that? There's a huge difference between bandwidth and latency, and optimizing bulk transfers for the best latency would be completely pointless.
You could prevent that with shaping based on volume.
No, no, no. That's what Comcast is doing right now. What we want to do is leave everything unlimited, but prioritized. If it's 4AM and nothing's happening, why shouldn't a torrent get the full bandwidth possible? For that matter, if it's 4PM and everyone's out playing golf or something, why shouldn't that same torrent run at full speed then?
It's called phasing out old stuff, which is something you have to do when the old behavior is so widespread.
Webmasters have been phasing in standards for most of the last of the decade. The ones that haven't? Screw 'em. Seriously. Anyone so incompetent and misinformed that this is a surprise needs the cluebat application which they are about to receive.
The only ones I feel sorry for are the people who've had to maintain essentially two sites: the defective one for IE, and the good one for everyone else. At least this beta beta will give them time to update their browser checks so that they can serve the appropriate site to IE8.
Then tell all the big site owners to cut out all the tube clogging, virus riddled advertisements.
Very good point. All those ads are currently served for free by the Bandwidth Fairy Guild, and it's unfair that Comcast has to pony up to carry that subsidized content.
If other protocols were impeded, soon, all P2P would look like HTTP.
What do you mean by "impeded"? I'm not advocating blocking anything in the slightest. However, you can prioritize highly interactive traffic (IM, HTTP, SSH) over bulk data like FTP or P2P transfers. This lets all the packets through, but doesn't make browsing impossible just because a tenth of an ISP's customers are downloading screengrabs of the new Indiana Jones.
And philosophically, could the Open Source community support one side in a competition such as this? What other issues does this raise?
I don't like Obama in the slightest, but wouldn't hesitate to contribute to a FOSS project his folks were hosting. After all, maybe my guy (whoever it may be) could benefit from using it. If MS released a useful project under the GPL, would people actually avoid contributing because it might help MS?
Honestly, mixing outside influences with FOSS in this way is dumb. Everyone is free to use Free Software and a rising tide lifts all boats.
If they were serious about addressing congestion, they'd prioritize traffic flows and be done with it. I don't think anyone would have a problem with putting P2P at a lower priority to HTTP. Of course, that doesn't help their master plan of billing content providers for tiered service, so they don't do it.
"Counselor" == "lawyer". The implication is that the poster I was replying to was one of the RIAA's lawyers who was mad because they lost that decision. Now that I've explained it, I'm sure you're endlessly amused.
I used to think the Eee PC was a great idea until I actually tried to type on a 7" model at Best Buy.
I hated the Eee's keyboard for about the first hour, then adjusted enough that it wasn't that big of a deal, at least for what I use it for (web surfing, email, SSH, developing large applications in Python).
Seems to me moving loser pays system would make it a lot more difficult to abuse the system. Or potentially require the loser to pay the winner the cost of the winner's counsel plus the cost of the losers council as well.
Screw that. The RIAA sues me; they spend $1,000,000, and I can only afford $1,000. When they win, I'm in debt for an extra million on top of the judgment.
For everyone who says that it should only apply to plaintiffs: what do if you need to sue the RIAA? Try to outspend them and hope for the best, knowing your life with be over if you lose?
"Loser pays" sucks for the poorer party, no matter how cleverly you try to phrase the rules.
The US voted for approval from the start (big suprise: American company gets supported by an uninformed America) so we wouldn't be likely to protest.
Ummm, why would you automatically assume that the American delegates would vote against Google, IBM, Red Hat, Sun, and all the other American ODF Alliance members? This isn't "US vs. The World", it's "One US Company vs. The Rest".
Now, we know that M$ [1] stacked the deck here. In a hypothetical unbiased panel, though, voting for Microsoft isn't necessarily voting for the interests of America.
[1] When discussing the crap like they do like this, M$ is a perfectly reasonable abbreviation.
And all this time I thought Python was easy. I'd have to write all this to do the exact same thing:
class foo: . def __init__(self): pass # initialize foo . def function_foo(a, b): pass # do something foo().function_foo(1, 2.0) # use method
Oh, and inheritance is pretty tricky:
class bar(foo): . def function_foo(c, d, e): pass
"C: for people who love memorizing syntax."
Of course, some of my wind is stolen by the fact that Slashdot is the only forum I ever post to which is pathologically incapable of maintaining indention in Python code snippets.
Take *that* you young whippersnappers! If you want to program machines, you have to learn to think like 'em.
I'm not going to list all the assembly languages I'm comfortable in; suffice it to say that I could probably land a job doing either embedded or mainframe stuff.
But I think you're wrong.
Unless you're a CPU designed or are writing a compiler, you probably won't understand how they're thinking. From out-of-order execution to scheduling finickiness to microcode, there are so many outside variables that affect what your code is actually doing that it's almost pointless to bother trying to keep it straight.
Even the compiler geeks can at best hope to get the rules down correctly so that the machine can generate code that will run efficiently on itself. Analogy: just because you can write a fractal generator doesn't mean that you could actually hand-generate a fractal with any reasonable speed or accuracy.
It seems to me as I touched on before, that you may well not be familiar enough with modern editors and IDE's that do the heavy lifting for you if you are so bothered by code that requires braces.
I would say the same if you are so bothered by code that doesn't.
import crypt;_=r'KZ[i.KqX8jDtA6Uwv!V01d?&iQ#bs,aL\yP<hkAFJ:TE{';print ''.join([crypt.crypt(''.join(x),'ks')[2:5] for x in zip(_[::3],_[1::3],_[2::3])]).replace('/',' ')
There are few joys in life like using something that is the perfect expression of its intent. Each trade has its representative tools, and their common trait is quality, even if it's not obvious to the casual observer, and often counterintuitive. The best tools in a category are almost always the least flashy, and rarely the ones a new practitioner would choose.
The Model M keyboard is like that: it's loud, ugly, heavy, and utterly lacking modern niceties like buttons to change your sound volume or check your email. And yet, it has that transcendent feeling that's hard to explain, that sense of rightness where you realize that you're using the best that's ever been made, that every change since then has been superfluous and cosmetic. With time, the loud clacking becomes the background music of your work, the harmony that tells you that your thoughts have become words. Its beige boxiness yields to elegant simplicity and the realization that true beauty is born of function, not appearance. The sheer weight of the thing turns to solidity and the confidence that it will stay where you put it. The dearth of features becomes the singleminded dedication to the parts that really matter and a proud disregard of unneeded distractions.
A tool attains its peak when a craftsman forgets that he's using it because it has become an extension of himself. Thus the humble Model M has become the iconic favorite of hackers everywhere, an ode to the engineers who grasped for excellence and acheived it.
Slashdot Mirror
I don't mind traffic shaping at all, anywhere. QoS is a good thing, even when the ISPs do it. What I mind a whole awful lot is traffic blocking, ala Comcast.
Sure you can, unless you're really so oversold that the baseline interactive traffic is greater than your capacity, in which case nothing will help but extra capacity. After all, traffic is prioritized everywhere; the only difference is that the default metric is packet arrival order.
I still think this is infinitely preferable to shutting down P2P altogether. Shaping the traffic at least lets everyone keep using their Internet connection any way they want. The only difference is that huge P2P downloads don't interfere with VOIP or other more time-sensitive traffic.
Why? Do you also run every process on your system at nice=0, including stuff like Folding@home, or do you prioritize so that your desktop stuff isn't unresponsive while still allowing the background tasks to run at full speed?
The distinguishing characteristic is interactivity. Stuff that a human is waiting for should go as fast as possible. Everything else should run unhindered after the latency-sensitive packets get through.
impede To retard or obstruct the progress of.
Take your own advice. A definition of "block":
Sounds damn near synonymous. Anyway, QOS is not the same as blocking or impeding in any way. With QOS, all the packets get through, just not at the expense of other traffic.
And then, as I said, very quickly P2P apps will start to mimic or run over "highly interactive traffic" so as not to be slowed down.What would the advantage be? Why would the Bittorrent (or whoever) devs want to do that? There's a huge difference between bandwidth and latency, and optimizing bulk transfers for the best latency would be completely pointless.
No, no, no. That's what Comcast is doing right now. What we want to do is leave everything unlimited, but prioritized. If it's 4AM and nothing's happening, why shouldn't a torrent get the full bandwidth possible? For that matter, if it's 4PM and everyone's out playing golf or something, why shouldn't that same torrent run at full speed then?
Webmasters have been phasing in standards for most of the last of the decade. The ones that haven't? Screw 'em. Seriously. Anyone so incompetent and misinformed that this is a surprise needs the cluebat application which they are about to receive.
The only ones I feel sorry for are the people who've had to maintain essentially two sites: the defective one for IE, and the good one for everyone else. At least this beta beta will give them time to update their browser checks so that they can serve the appropriate site to IE8.
Very good point. All those ads are currently served for free by the Bandwidth Fairy Guild, and it's unfair that Comcast has to pony up to carry that subsidized content.
What do you mean by "impeded"? I'm not advocating blocking anything in the slightest. However, you can prioritize highly interactive traffic (IM, HTTP, SSH) over bulk data like FTP or P2P transfers. This lets all the packets through, but doesn't make browsing impossible just because a tenth of an ISP's customers are downloading screengrabs of the new Indiana Jones.
I don't like Obama in the slightest, but wouldn't hesitate to contribute to a FOSS project his folks were hosting. After all, maybe my guy (whoever it may be) could benefit from using it. If MS released a useful project under the GPL, would people actually avoid contributing because it might help MS?
Honestly, mixing outside influences with FOSS in this way is dumb. Everyone is free to use Free Software and a rising tide lifts all boats.
If they were serious about addressing congestion, they'd prioritize traffic flows and be done with it. I don't think anyone would have a problem with putting P2P at a lower priority to HTTP. Of course, that doesn't help their master plan of billing content providers for tiered service, so they don't do it.
"Counselor" == "lawyer". The implication is that the poster I was replying to was one of the RIAA's lawyers who was mad because they lost that decision. Now that I've explained it, I'm sure you're endlessly amused.
They said that Ruby is slow, Smalltalk is better, and that SQLAlchemy can beat up ActiveRecord. Also, PHP sucks.
*sits back to watch the ensuing flamewar*
I hated the Eee's keyboard for about the first hour, then adjusted enough that it wasn't that big of a deal, at least for what I use it for (web surfing, email, SSH, developing large applications in Python).
Sour grapes, Counselor.
Screw that. The RIAA sues me; they spend $1,000,000, and I can only afford $1,000. When they win, I'm in debt for an extra million on top of the judgment.
For everyone who says that it should only apply to plaintiffs: what do if you need to sue the RIAA? Try to outspend them and hope for the best, knowing your life with be over if you lose?
"Loser pays" sucks for the poorer party, no matter how cleverly you try to phrase the rules.
Ummm, why would you automatically assume that the American delegates would vote against Google, IBM, Red Hat, Sun, and all the other American ODF Alliance members? This isn't "US vs. The World", it's "One US Company vs. The Rest".
Now, we know that M$ [1] stacked the deck here. In a hypothetical unbiased panel, though, voting for Microsoft isn't necessarily voting for the interests of America.
[1] When discussing the crap like they do like this, M$ is a perfectly reasonable abbreviation.
And all this time I thought Python was easy. I'd have to write all this to do the exact same thing:
Oh, and inheritance is pretty tricky:
"C: for people who love memorizing syntax."
Of course, some of my wind is stolen by the fact that Slashdot is the only forum I ever post to which is pathologically incapable of maintaining indention in Python code snippets.
I'm not going to list all the assembly languages I'm comfortable in; suffice it to say that I could probably land a job doing either embedded or mainframe stuff.
But I think you're wrong.
Unless you're a CPU designed or are writing a compiler, you probably won't understand how they're thinking. From out-of-order execution to scheduling finickiness to microcode, there are so many outside variables that affect what your code is actually doing that it's almost pointless to bother trying to keep it straight.
Even the compiler geeks can at best hope to get the rules down correctly so that the machine can generate code that will run efficiently on itself. Analogy: just because you can write a fractal generator doesn't mean that you could actually hand-generate a fractal with any reasonable speed or accuracy.
I would say the same if you are so bothered by code that doesn't.
Thanks! Getting the right values to feed into crypt() was the hardest part.
...he says by clicking on an HTML widget that causes JavaScript to turn his words into XML and then transmit them to an application written in Perl.
BASIC and Pascal are procedural.
as is SQLSQL is declarative.
LISP, SCHEME and SMALLTALK were all developed when space was at a premium. so, you kept the source file small by using as obtuse-as-possible a syntax.That's not why they're that way at all.
Python's a great language, but that's no reason to get sloppy about the details.
Until I quit developing with Notepad, I felt the same way.
(Note: the above loop can be short-circuited)
Reposted from my blog:
There are few joys in life like using something that is the perfect expression of its intent. Each trade has its representative tools, and their common trait is quality, even if it's not obvious to the casual observer, and often counterintuitive. The best tools in a category are almost always the least flashy, and rarely the ones a new practitioner would choose.
The Model M keyboard is like that: it's loud, ugly, heavy, and utterly lacking modern niceties like buttons to change your sound volume or check your email. And yet, it has that transcendent feeling that's hard to explain, that sense of rightness where you realize that you're using the best that's ever been made, that every change since then has been superfluous and cosmetic. With time, the loud clacking becomes the background music of your work, the harmony that tells you that your thoughts have become words. Its beige boxiness yields to elegant simplicity and the realization that true beauty is born of function, not appearance. The sheer weight of the thing turns to solidity and the confidence that it will stay where you put it. The dearth of features becomes the singleminded dedication to the parts that really matter and a proud disregard of unneeded distractions.
A tool attains its peak when a craftsman forgets that he's using it because it has become an extension of himself. Thus the humble Model M has become the iconic favorite of hackers everywhere, an ode to the engineers who grasped for excellence and acheived it.