The man at the top needs to go as well. Seriously, the oath of office for the POTUS is:
I do solemnly swear (or affirm) that I will faithfully execute the office of President of the United States, and will to the best of my ability, preserve, protect, and defend the Constitution of the United States.
This president apparently cares not for our constitution. Between wanting to add amendments to prevent flag burning and gay marriage, to trampling on the rights of the people in the name of security, he's never seemed to understand that his primary duty is to protect the ideals in that document upon which this nation is based.
As I see it, this program will never be legal as long as the fourth amendment protects me from unreasonable search and seizure. The government has no right to invate my privacy to further their investigations unless I provide them with enough reason for them to obtain a warrant. They can pass their law. I don't care. It doesn't change a damn thing since the constitution is the standard against which it will ultimately be measured.
As for this POTUS, he belongs in jail. His actions in office are detrimental to the long-term health and safety of our nation. If the dems somehow take control of the senate (I know, fat chance) then there should be an immediate movement to either censure or impeach him.
Re:Very interesting, but a little late
on
Gzip on a PCI card
·
· Score: 1
Installing a consistent software image across "hundreds of web servers" would be necessary regardless of the presence of mod_gzip. I fail to see how eliminating the mod_gzip module would really help in this regard.
I would also think "hundreds of webservers" would be able to gzip at a composite rate higher than 50MB/sec, even while serving content. After all, with a hundred webservers, each webserver would only have to compress at an average rate of 0.5MB/sec to achieve 50MB/sec performance.
Look, I can see certain scenarios where it would be beneficial to do the compression in the load balancer. I just don't think the scenario you give is a valid one.
Wow - there's some new OS that was written in the last 6 weeks?? Cool!
But seriously, I think the argument that UNIX is old technology is pretty lame. The fundamental design is that old, but it's also very elegant and flexible. That flexibility has allowed it to stay in step with the times, so that there are few (if any) missing useful features in current implementations.
As far as having a GUI for every UNIX command - I for one would not like to give up on the CLI. To paraphrase a Van Jacobsen statement from a USENIX keynote some years ago: Windows gives you sentences and paragraphs while UNIX gives you words.
Having the ability to address needs that were not forseen by the authors of the software is a huge win for "30-year-old" UNIX. It would be near impossible to duplicate that flexibility in a graphical environment. Even if possible, it'd me more time-consuming to build that pipe.
Let's see your tar gui do this:
tar cf - . | ( cd/newdisk && tar xpf - )
That little "|" is the key. In the UNIX world it's a true pipe - not a hidden temporary file. Until that functionality can be built into a GUI we'd be doing the UNIX community a disservice by eliminating the CLI...
Hmm. They could order ISPs to implement filters blocking the well-known ports or, with time, the p2p protocols via stateful inspection. They could even restrict contact between customers' machines, thereby forcing the use of a registered "server", or at least a host on another ISP...
Steps like these could dampen the spread of P2P protocols fairly quickly without breaking into any homes. Of course the impact would be greater for the casual users. The die-hards would constantly modify the port/protocol to stay ahead of any such filters...
Well, the idea is that they densely pack customers onto their fiber. Using current state-of-the-art technology they can have up to 100 gigabit ethernet channels per strand of fiber. That translates into roughly 1000 customers or a potential of $1M/mo revenue per strand of fiber.
I believe current pricing of established ISPs is not deflating enough to match technology trends. This is in part because the established carriers have a harder time deploying the new DWDM equipment into their existing networks as compared to a new company doing their initial build with the same equipment. After all, they do have to pay for their existing equipment before buying the new stuff...
Cogent has the advantage of building a new network from the ground up using SOTA DWDM equipment. It is concentrating on high-density markets in order to achieve a high density of customers (to better use their infrastructure). This makes economic sense since their model will fail if they don't achieve a good concentation of customers per fiber. So it's a pick and choose deal for them to some degree, and given their target price they can appeal to smaller customers which should work to improve their customer density.
So if I were an apartment/condo dweller in one of the supported cities, I'd be giving my super a call.;^) Heck, if they open a POP nearby the thought of a neighborhood cooperative LAN is really appealing to me - 20 homes @ $50/mo for 5Mbps full duplex connections... Sounds yummy!
I swear this repetition is becoming more common on/. Is it simply the number of articles being submitted that precludes doing verification of the uniqueness of a topic? Is real news hard to come by? Do the people in control not read the site?
I'm baffled since a simple search for "cogent" turned up the old article for me...
A few questions/comments on my mind that weren't touched were:
The web is becoming much more of a cached entity, such that the "1000 people in UK causing a page to cross the Atlantic 1000 times" scenario is becoming less true, especially for the Yahoo!s, CNNs, etc. Given that this is the way the web is going, besides adding encryption and anonymity how is FreeNet really different than a heirarchy of web caches?
Isn't replication of data the reason that we were supposed to have URNs (Universal Resource Names) in addition to URLs? My understanding was that a URN would address data in a location-independent way, and would resolve into a valid URL which the browser would use to retrieve the data. It seems to me that URNs were skipped because they posed a difficult problem and now we're paying the price for taking the easy path in the early days.
One of the great things about the advent of mosaic was that it placed a unified interface in front of what were, up until that time, separate services (gopher, nntp, http, and to some extent telnet). Why is it that we haven't seen more growth in the protocol (e.g. "http://") area of URLs so that newer technologies can leverage the public's acceptance of the browser interface? It seems to me that this spawning of applications will lead us back to the confusion of having to use specific applications to use specific services. In other words I'd like to see freenet://... (or gnutella, napster, etc.) URLs (er, URNs) in mozilla someday, if not the commercial browsers. Is there any chance of that happening?
Anyway - I hope these comments aren't complete drivel.:-)
I've heard that there is now a process for voting early, and I believe today (Oct 16) is the first day that you can do it. I suspect the details vary by community, but you can go to an early polling location, get a ballot, and be done with it.
To prove I'm not totally nuts, the Durham, NC information is (they'll hate me for this:)
here.
"No... No... No... Please remove me from your calling list."
My understanding is that a phone solicitor is allowed to press you until you've said "no" 3 times. Some are nice enough to stop at the first "no" and politely go away. (That actually is worse for me -- I tend to forget the "Please remove me" line in those situations.:)
Also if you ask them to remove you from their calling list, they are obligated by law to do so. By sticking to this regimine, we get far fewer phone solicitation calls than we used to...
True, but protocols such as MFTP (multicast file transfer protocol) address this issue. MFTP blindly transmits a number of blocks, then sends a query. The clients see the query, determine if they missed blocks, and then request retransmission. This is repeated until none of the clients complain...
I played with developing this a few years back, building a rough prototype of an MFTP-like multicast file distribution program in perl. It worked well in the test phase, but I never took the software anywhere beyond a simple test scenario... The amazing thing was that with the speed-up of xmitting to 20 hosts simultaneously, the prototype had acceptable speed.:-)
Wow. I wouldn't give this article much credit for being anything other than flame bait.
Hardly anyone running a large production service would use a product that had just been announced. The only reason to do so would be to get out of a bad situation (the fix is in the latest version.. really:^). So while such a comment might get MSes goat, it seems to be taken out of context.
I also think MS' pains in migrating their subsidiaries' services to MS products is to be expected. It would probably be a lot easier for them if they could build the service from the ground up. But they have to migrate an existing service with millions of customers (and who knows what kind of code base / cool hacks) to an entirely different platform. That's not an easy job by any means, and they're doing it while the service is live...
This sounds like a columnist that feeds off of disgruntled former employees to me. I don't think I'd trust his information very much, even if it is entertaining reading.
There are 2 primary sites for the Olympics - www.sydney.olympics.org and www.nbcolympics.com. AFAIK both sites will be doing live publishing of scores during the events, and their coverage will be extensive.
The bit about "IBM... landed the gig for producing Olympics.com, a job that ordinarily in the sports world goes to an online media outfit" is (IMHO) totally wrong. IBM produces and hosts some of the biggest sporting-events sites out there, including www.wimbledon.org, www.usopen.org (tennis, not golf), www.pga.com, www.ausopen.org, www.rydercup.org, etc. In addition they hosted the official sites for the Atlanta and Nagano games. This comment ignores IBM's track record in hosting these sporting event sites during the events. (FWIW, this is the last Olympics that IBM will be hosting...)
Slashdot Mirror
This president apparently cares not for our constitution. Between wanting to add amendments to prevent flag burning and gay marriage, to trampling on the rights of the people in the name of security, he's never seemed to understand that his primary duty is to protect the ideals in that document upon which this nation is based.
As I see it, this program will never be legal as long as the fourth amendment protects me from unreasonable search and seizure. The government has no right to invate my privacy to further their investigations unless I provide them with enough reason for them to obtain a warrant. They can pass their law. I don't care. It doesn't change a damn thing since the constitution is the standard against which it will ultimately be measured.
As for this POTUS, he belongs in jail. His actions in office are detrimental to the long-term health and safety of our nation. If the dems somehow take control of the senate (I know, fat chance) then there should be an immediate movement to either censure or impeach him.
Installing a consistent software image across "hundreds of web servers" would be necessary regardless of the presence of mod_gzip. I fail to see how eliminating the mod_gzip module would really help in this regard.
I would also think "hundreds of webservers" would be able to gzip at a composite rate higher than 50MB/sec, even while serving content. After all, with a hundred webservers, each webserver would only have to compress at an average rate of 0.5MB/sec to achieve 50MB/sec performance.
Look, I can see certain scenarios where it would be beneficial to do the compression in the load balancer. I just don't think the scenario you give is a valid one.
Wow - there's some new OS that was written in the last 6 weeks?? Cool!
/newdisk && tar xpf - )
But seriously, I think the argument that UNIX is old technology is pretty lame. The fundamental design is that old, but it's also very elegant and flexible. That flexibility has allowed it to stay in step with the times, so that there are few (if any) missing useful features in current implementations.
As far as having a GUI for every UNIX command - I for one would not like to give up on the CLI. To paraphrase a Van Jacobsen statement from a USENIX keynote some years ago: Windows gives you sentences and paragraphs while UNIX gives you words.
Having the ability to address needs that were not forseen by the authors of the software is a huge win for "30-year-old" UNIX. It would be near impossible to duplicate that flexibility in a graphical environment. Even if possible, it'd me more time-consuming to build that pipe.
Let's see your tar gui do this:
tar cf - . | ( cd
That little "|" is the key. In the UNIX world it's a true pipe - not a hidden temporary file. Until that functionality can be built into a GUI we'd be doing the UNIX community a disservice by eliminating the CLI...
Ah well, I guess I'm just set in my old ways...
Hmm. They could order ISPs to implement filters blocking the well-known ports or, with time, the p2p protocols via stateful inspection. They could even restrict contact between customers' machines, thereby forcing the use of a registered "server", or at least a host on another ISP...
Steps like these could dampen the spread of P2P protocols fairly quickly without breaking into any homes. Of course the impact would be greater for the casual users. The die-hards would constantly modify the port/protocol to stay ahead of any such filters...
Well, the idea is that they densely pack customers onto their fiber. Using current state-of-the-art technology they can have up to 100 gigabit ethernet channels per strand of fiber. That translates into roughly 1000 customers or a potential of $1M/mo revenue per strand of fiber.
;^) Heck, if they open a POP nearby the thought of a neighborhood cooperative LAN is really appealing to me - 20 homes @ $50/mo for 5Mbps full duplex connections... Sounds yummy!
I believe current pricing of established ISPs is not deflating enough to match technology trends. This is in part because the established carriers have a harder time deploying the new DWDM equipment into their existing networks as compared to a new company doing their initial build with the same equipment. After all, they do have to pay for their existing equipment before buying the new stuff...
Cogent has the advantage of building a new network from the ground up using SOTA DWDM equipment. It is concentrating on high-density markets in order to achieve a high density of customers (to better use their infrastructure). This makes economic sense since their model will fail if they don't achieve a good concentation of customers per fiber. So it's a pick and choose deal for them to some degree, and given their target price they can appeal to smaller customers which should work to improve their customer density.
So if I were an apartment/condo dweller in one of the supported cities, I'd be giving my super a call.
I swear this repetition is becoming more common on /. Is it simply the number of articles being submitted that precludes doing verification of the uniqueness of a topic? Is real news hard to come by? Do the people in control not read the site?
I'm baffled since a simple search for "cogent" turned up the old article for me...
A few questions/comments on my mind that weren't touched were:
:-)
The web is becoming much more of a cached entity, such that the "1000 people in UK causing a page to cross the Atlantic 1000 times" scenario is becoming less true, especially for the Yahoo!s, CNNs, etc. Given that this is the way the web is going, besides adding encryption and anonymity how is FreeNet really different than a heirarchy of web caches?
Isn't replication of data the reason that we were supposed to have URNs (Universal Resource Names) in addition to URLs? My understanding was that a URN would address data in a location-independent way, and would resolve into a valid URL which the browser would use to retrieve the data. It seems to me that URNs were skipped because they posed a difficult problem and now we're paying the price for taking the easy path in the early days.
One of the great things about the advent of mosaic was that it placed a unified interface in front of what were, up until that time, separate services (gopher, nntp, http, and to some extent telnet). Why is it that we haven't seen more growth in the protocol (e.g. "http://") area of URLs so that newer technologies can leverage the public's acceptance of the browser interface? It seems to me that this spawning of applications will lead us back to the confusion of having to use specific applications to use specific services. In other words I'd like to see freenet://... (or gnutella, napster, etc.) URLs (er, URNs) in mozilla someday, if not the commercial browsers. Is there any chance of that happening?
Anyway - I hope these comments aren't complete drivel.
To prove I'm not totally nuts, the Durham, NC information is (they'll hate me for this :)
here.
"No... No... No... Please remove me from your calling list."
:)
My understanding is that a phone solicitor is allowed to press you until you've said "no" 3 times. Some are nice enough to stop at the first "no" and politely go away. (That actually is worse for me -- I tend to forget the "Please remove me" line in those situations.
Also if you ask them to remove you from their calling list, they are obligated by law to do so. By sticking to this regimine, we get far fewer phone solicitation calls than we used to...
True, but protocols such as MFTP (multicast file transfer protocol) address this issue. MFTP blindly transmits a number of blocks, then sends a query. The clients see the query, determine if they missed blocks, and then request retransmission. This is repeated until none of the clients complain...
:-)
I played with developing this a few years back, building a rough prototype of an MFTP-like multicast file distribution program in perl. It worked well in the test phase, but I never took the software anywhere beyond a simple test scenario... The amazing thing was that with the speed-up of xmitting to 20 hosts simultaneously, the prototype had acceptable speed.
Hardly anyone running a large production service would use a product that had just been announced. The only reason to do so would be to get out of a bad situation (the fix is in the latest version .. really :^). So while such a comment might get MSes goat, it seems to be taken out of context.
I also think MS' pains in migrating their subsidiaries' services to MS products is to be expected. It would probably be a lot easier for them if they could build the service from the ground up. But they have to migrate an existing service with millions of customers (and who knows what kind of code base / cool hacks) to an entirely different platform. That's not an easy job by any means, and they're doing it while the service is live...
This sounds like a columnist that feeds off of disgruntled former employees to me. I don't think I'd trust his information very much, even if it is entertaining reading.
The bit about "IBM ... landed the gig for producing Olympics.com, a job that ordinarily in the sports world goes to an online media outfit" is (IMHO) totally wrong. IBM produces and hosts some of the biggest sporting-events sites out there, including www.wimbledon.org, www.usopen.org (tennis, not golf), www.pga.com, www.ausopen.org, www.rydercup.org, etc. In addition they hosted the official sites for the Atlanta and Nagano games. This comment ignores IBM's track record in hosting these sporting event sites during the events. (FWIW, this is the last Olympics that IBM will be hosting...)