The Phase Tree algorithm is actually an improvement on this idea. Instead of recursively updating trees for every filechange, you keep 3 'phases' of the tree and cache which blocks have changed - then atomically write the metablock (et. al) after X changes or X milliseconds. This drastically reduces the amount of work that must be done and removes the constant recursion problem. This is better described in the info given on the Tux2 site.
>I'm no fan of trickle-down economics, but which >would you rather have:
>
>1) $1b paid by Cisco, employees get zip
>2) $7b in stock options "paid" out by Cisco, $2b >in taxes paid to the govt. Employees wind up >with $5b in their pockets.
The argument is not that Cisco and MS made bad decisions given the current tax law. And in fact, if I were Cisco - I would say #2 all the way, and expect everyone to be happy. The argument is that the tax law is outdated and unfair in todays economy. The real options are your #2 (the logical action w/ this tax law) or a #3 (if this law didnt exist):
3) $7b in stock options earned by Cisco employees (Cisco stops giving options + Cisco has no more employees, so this isnt going to change), $4.5b in taxes paid to the govt by Cisco, $2b paid by employees.
As the article points out, giving out lots of options will dilute earnings per share. But given that options are status quo in the new economy, and given that without them, employee retention rate is significantly lower (although options are definitely not everything) - I doubt that without this tax loophole options would disappear.
Lets get this straight - trading shares causes money to flow between shareholders. Shares are still just a piece of Cisco. Shareholders pay everything out of their pockets - Cisco does not pay these employees, they get paid by the stock market.
yeah - hose are valid issues. Butthe thing to remember with the PIII 1GHz and the P4 1X GHz is that it is not the processor that is the bottleneck right now.
If you bench the two next to ewachother in a similar config - the difference is slight - that is because the bottleneck is the PCI bus.
We now have fast procs - fast mem - great video (ala Gforce et al) but we have a slow ass 33/66mhz pci subsystem. We need PCIx at 133? to increase the overall bandwidth of the systems - the procs are thirsty for it.
oh - and not to mention some nicley optimized SW to really put all that tech through its paces....
I do know that there were plans to design a DSP module for the Quicknet PhoneJACK PCI cards that would do exactly this (they have powerful onboard DSPs that actually would work quite well with the signal processing used by SETI@Home). But, as you said, the SETI@Home guys stated that there were just too many people, and not enough data to need any client speed-up. Would certainly be cool though.
...but then again I've heard mysql described as a sql interface to a filesystem, so....
In fact you are exactly correct. Mysql is an SQL frontend to a filesystem. This is an incredible advantage, however. For instance, Mysql is written such that the format of the tables is not linked to the syntax of the interface. IE - there are multiple different table types available, each with their own pros and cons. But all of them are accessed in the same way from the user end - via SQL.
There are standard MyISAM tables which consist of 3 files in a binary portable format (index, data, and layout).
There are heap tables which are binary trees resident only in memory (EXTREMEMLY fast, but rather temporary).
And now there are Berkeley DB tables - which support transactions but are significantly slower than the other tables.
Unix/Linux is built on the premise that everything is a file, so it is rather ignorant to say that an SQL interface to a filesystem is a bad thing. Of course the database engine deals with things like locks, concurrency, threading, atomic operations, logging and more. But the real advantage here is if you want to create a new table type with your own design criteria, you can plug it right into the MySQL engine and boom - now you have an SQL interface, a Client Server app, a user security system, replication and more.
OpenH323 is an incredible project. It is more stable than most commercial grade H323 stacks when it comes to audio (H323 includes video also - think NetMeeting). Its really strong points are:
Crossplatform - runs on Windows and *nix (Im not sure about Max or Bez et al)
It has been tested and interoperates with more H323 stacks than you knew existed - Radware, Cisco...
Free Software - which is the point of this article, right?
There are some problems with the H323 specification in general though. For example:
Very complex... just take a look at the codebase
Control data is transmitted in binary form - most widely used protocols are based on ASCII (FTP, HTTP, SMTP)
It uses a port assignment process which is virtually impossible to use through a NAT firewall.
There are of course many options in the VoIP world right now - SIP is a protocol that works to simplify the processes of the H323 stack. As far as I know, there are a few different implementations of SIP and none of them work very well with each other. You can read more about it here.
A friend of mine has written some very good articles about Linux and Internet Telephony:
I personally think that the best solution right now in terms of interoperability, quality and Free-as-in-speech-ness is OpenH323 with OpenPhone. Our company uses a combination of Quicknet PhoneJACKs, OpenH323, and a few CIPE VPN tunnels to connect people from CA to Texas to Australia at their Linux boxes using real-live ringing phones - at essentially no cost. Quality is very very close to a typical old-guard phone call, even from San Francisco to Sydney over the Internet, _and_ encrypted. Blows my mind whenever I use this stuff. The Quicknet cards have GPL'd drivers and are in the current kernel tree. They seem to add a ton of power to the call by offloading alot of the work to hardware DSPs.
I dont think the point is that the cypherpunks have stopped believing that Big Brother is looming, but rather that a better way to combat 'Big Brother' is to understand the political and social power of circles of trust. Much like The Matrix... the agents are always looming, but learning to think and act like an agent is sometimes the most powerful tool for combat.
...In my feeling RedHat is the desktop distribution of Linux, Slackware closer to the production server distro....
I think SuSE is a much better desktop distribution than RedHat - its always on the cutting edge of XFree86 (Top execs are on the board) and KDE. And you also get 6 CDs of apps. I haven't found RedHat's niche yet... I do think that in the corporate US market, which is what SAS is targeting, RedHat has the strongest foothold in the Linux Server realm.
Consider this: Traditional voting rules dictate that there is a certain distance around voting booths within which there can be no campaigning or campaign signs. With internet voting, this becomes a very different problem. Net voting booths are set up in various communities (as has already been done). Since these are not traditionally sanctioned voting locations, do they have to abide by the same no-propaganda rules?
This is pure biship. I admit that moving elections online is a quite liberal => Democratic idea, but to imply both that it will be used primarily by special interest groups to manipulate underprivileged citizens and that underprivileged citizens have less of a right to be heard by their government is utterly ridiculous.
A. I am a fairly competent citizen of California working at an intense start up in a highly congested area aka San Francisco. I was unable to vote in our March 7th election for many reasons:
Local Voting Booths close at 8pm.
I work long hours, like many people in this industry.
Traffic is very congested and most people, including myself, live a long way from where they work (real estate prices in this area are VERY high) - thus it takes a long time to get home/to a voting booth.
I do not have my own method of transportation, so in many cases I am not in full control of how/when I can get home.
All of these are common problems of citizens of highly congested urban areas like myself. Now tell me that internet voting would not gain a significant number of voters from this demographic...
B. I am sorry to say that you have exactly as much right to vote as the homeless man you pass on your way to work everyday. In the eyes of the U.S. government, and the U.S. Constitution, you have exactly as much worth as he does. No more, no less. And they have as much right to say people who post ignorant comments on Slashdot should be denied the right to vote as they do to deny that right from any citizen of the United States.
Just as any software package gets refined with time, election.com's will inevitably improve. Maybe Java, maybe Perl, hell maybe even ASP.. Who knows.
I think, however, the fact that over 35,000 people managed to successfully vote online makes b****ing about implementation (security withheld, of course) a moot point.
Even if you ignore the ethics of it (pretend pulling the plug is ok-e-dok-e), the real question is what position does evolving/creating systems that "engineers" and humans in general don't understand put the world in? Society has shown us that the human race as a whole, (but maybe not all of us) loves systems. It can be governments, schools, churchs, whathaveyou. Whatever it is that lets us understand that big picture which is beyond our grasp. Well what happens when the actual functioning of these computers, the pinacle of human control, falls outside of all of our methods of understanding systems? We lose control. Oh well, gotta catch a plane. Oh and I just watched the Matrix for the 4th time. That'll probably do it.
Slashdot Mirror
The Phase Tree algorithm is actually an improvement on this idea. Instead of recursively updating trees for every filechange, you keep 3 'phases' of the tree and cache which blocks have changed - then atomically write the metablock (et. al) after X changes or X milliseconds. This drastically reduces the amount of work that must be done and removes the constant recursion problem. This is better described in the info given on the Tux2 site.
Put your hand in the puppethead
>I'm no fan of trickle-down economics, but which >would you rather have:
>
>1) $1b paid by Cisco, employees get zip
>2) $7b in stock options "paid" out by Cisco, $2b >in taxes paid to the govt. Employees wind up >with $5b in their pockets.
The argument is not that Cisco and MS made bad decisions given the current tax law. And in fact, if I were Cisco - I would say #2 all the way, and expect everyone to be happy. The argument is that the tax law is outdated and unfair in todays economy. The real options are your #2 (the logical action w/ this tax law) or a #3 (if this law didnt exist):
3) $7b in stock options earned by Cisco employees (Cisco stops giving options + Cisco has no more employees, so this isnt going to change), $4.5b in taxes paid to the govt by Cisco, $2b paid by employees.
As the article points out, giving out lots of options will dilute earnings per share. But given that options are status quo in the new economy, and given that without them, employee retention rate is significantly lower (although options are definitely not everything) - I doubt that without this tax loophole options would disappear.
Lets get this straight - trading shares causes money to flow between shareholders. Shares are still just a piece of Cisco. Shareholders pay everything out of their pockets - Cisco does not pay these employees, they get paid by the stock market.
yeah - hose are valid issues. Butthe thing to remember with the PIII 1GHz and the P4 1X GHz is that it is not the processor that is the bottleneck right now.
If you bench the two next to ewachother in a similar config - the difference is slight - that is because the bottleneck is the PCI bus.
We now have fast procs - fast mem - great video (ala Gforce et al) but we have a slow ass 33/66mhz pci subsystem. We need PCIx at 133? to increase the overall bandwidth of the systems - the procs are thirsty for it.
oh - and not to mention some nicley optimized SW to really put all that tech through its paces....
ex intel here.
yes it is true - check with my "sources" in intel...
I do know that there were plans to design a DSP module for the Quicknet PhoneJACK PCI cards that would do exactly this (they have powerful onboard DSPs that actually would work quite well with the signal processing used by SETI@Home). But, as you said, the SETI@Home guys stated that there were just too many people, and not enough data to need any client speed-up. Would certainly be cool though.
In fact you are exactly correct. Mysql is an SQL frontend to a filesystem. This is an incredible advantage, however. For instance, Mysql is written such that the format of the tables is not linked to the syntax of the interface. IE - there are multiple different table types available, each with their own pros and cons. But all of them are accessed in the same way from the user end - via SQL.
Unix/Linux is built on the premise that everything is a file, so it is rather ignorant to say that an SQL interface to a filesystem is a bad thing. Of course the database engine deals with things like locks, concurrency, threading, atomic operations, logging and more. But the real advantage here is if you want to create a new table type with your own design criteria, you can plug it right into the MySQL engine and boom - now you have an SQL interface, a Client Server app, a user security system, replication and more.
Ok. There in. Are you happy now? 3.23.14
P.S. Things change rapidly in the OpenSource world, lest you forgot.
There are some problems with the H323 specification in general though. For example:
There are of course many options in the VoIP world right now - SIP is a protocol that works to simplify the processes of the H323 stack. As far as I know, there are a few different implementations of SIP and none of them work very well with each other. You can read more about it here.
A friend of mine has written some very good articles about Linux and Internet Telephony:
Linux Journal Article
SVLUG Presentation
I personally think that the best solution right now in terms of interoperability, quality and Free-as-in-speech-ness is OpenH323 with OpenPhone. Our company uses a combination of Quicknet PhoneJACKs, OpenH323, and a few CIPE VPN tunnels to connect people from CA to Texas to Australia at their Linux boxes using real-live ringing phones - at essentially no cost. Quality is very very close to a typical old-guard phone call, even from San Francisco to Sydney over the Internet, _and_ encrypted. Blows my mind whenever I use this stuff. The Quicknet cards have GPL'd drivers and are in the current kernel tree. They seem to add a ton of power to the call by offloading alot of the work to hardware DSPs.
I dont think the point is that the cypherpunks have stopped believing that Big Brother is looming, but rather that a better way to combat 'Big Brother' is to understand the political and social power of circles of trust. Much like The Matrix... the agents are always looming, but learning to think and act like an agent is sometimes the most powerful tool for combat.
dpk
...In my feeling RedHat is the desktop distribution of Linux, Slackware closer to the production server distro....
I think SuSE is a much better desktop distribution than RedHat - its always on the cutting edge of XFree86 (Top execs are on the board) and KDE. And you also get 6 CDs of apps. I haven't found RedHat's niche yet... I do think that in the corporate US market, which is what SAS is targeting, RedHat has the strongest foothold in the Linux Server realm.
Consider this: Traditional voting rules dictate that there is a certain distance around voting booths within which there can be no campaigning or campaign signs. With internet voting, this becomes a very different problem. Net voting booths are set up in various communities (as has already been done). Since these are not traditionally sanctioned voting locations, do they have to abide by the same no-propaganda rules?
A. I am a fairly competent citizen of California working at an intense start up in a highly congested area aka San Francisco. I was unable to vote in our March 7th election for many reasons:
All of these are common problems of citizens of highly congested urban areas like myself. Now tell me that internet voting would not gain a significant number of voters from this demographic...
B. I am sorry to say that you have exactly as much right to vote as the homeless man you pass on your way to work everyday. In the eyes of the U.S. government, and the U.S. Constitution, you have exactly as much worth as he does. No more, no less. And they have as much right to say people who post ignorant comments on Slashdot should be denied the right to vote as they do to deny that right from any citizen of the United States.
Just as any software package gets refined with time, election.com's will inevitably improve. Maybe Java, maybe Perl, hell maybe even ASP.. Who knows.
I think, however, the fact that over 35,000 people managed to successfully vote online makes b****ing about implementation (security withheld, of course) a moot point.
Even if you ignore the ethics of it (pretend pulling the plug is ok-e-dok-e), the real question is what position does evolving/creating systems that "engineers" and humans in general don't understand put the world in? Society has shown us that the human race as a whole, (but maybe not all of us) loves systems. It can be governments, schools, churchs, whathaveyou. Whatever it is that lets us understand that big picture which is beyond our grasp. Well what happens when the actual functioning of these computers, the pinacle of human control, falls outside of all of our methods of understanding systems? We lose control. Oh well, gotta catch a plane. Oh and I just watched the Matrix for the 4th time. That'll probably do it.