I am so impressed with the caliber of the reporting. This artical implies that the USA will become a net exporter of oil complements of its left over turkey parts. hahahaha.
Imagine 4 billion barrels of light turkey crude each year. Presently the US imports about 14 million barrles oil and equivalents per day (about 2/3 of its guttonous appetite). This amounts to about 5 billion barrels. So to hell with the middle east eh?
I agree with lobbying to make software patents invalid. The problem is that in the interm patents are used as trading cards between large companies. In general we little guys have no patents and no clout and not enough money to fight a patent suit even if we are in the right.
As for money to throw around... well - there are over 1/2 million developers registered in source forge alone and if each of us buys a membership for $100 bux that works out to $50 million. With a war chest of $50 million we can register quite a large number of patents and still have funds left over to fight suits.
Furthermore if we do end up with a war chest then we'll end up with some large corporations wanting to cross license with us and this will pull their patents into the association.
I figure $100 bux per year is quite reasonable if it eliminates the legal minefield that we presently face. Furthermore, an association like this can lobby to have patents eliminated.
The validity of this patent is not really the issue here. Regardless of whether there is prior art the problem is that programmers cannot generally afford to fight invalid patents. M$ accomplishes its objective which is to prevent programmers from doing their jobs.
This patent is just another example of why WE NEED TO ORGANISE an OPEN SOURCE PATENT ASSOCIATION and each of us needs to throw in $100 bux or $1000 or whatever it takes to finance an organisation that can both patent and fight for us. As a member of an organzation like this we would have the right to use any patents that we hold and we _CAN_ prevent M$ and TI and IBM and everyone else from using these patents. If _our_ organisation simply picks the best ideas we come up with and patents them in very short order we'll have a rather mean shief of patents up our collective sleeves.
It might be useful if you would educate yourself about the physics involved.
Reactors are quite safe. Furthermore mankind will either enjoy a nuclear future or freeze in the dark. Fossil fuel energy resources are quite limited.
The US DOE for instance forecasts that by 2020 the consumption of natural gas will be up about 489%. They actually forecast that much of this gas will come from Canada.
Well completions have doubled in the last few years and the result of this was a rather modest supply increase in 2001. In 2002 the supply dropped slightly. There is just no way on earth that the Exploration and Production industries can increase gas supplies by any significant amount.
American companies are welcome to come up here and look. Many are. Many are also buying reserves, companies like Burlington for instance who just bought Canadian Hunter Exploration Limited are an example.
The issue is that there is a supply side crunch on its way and we are totally unprepared for it.
So, nuclear will find its way back in rather soon I think. But - I do expect that it will be a ways past 2015 before this happens. Also - I do expect that before nuclear starts comming back there are going to be some rather sharp supply problems and some rather panicy people sitting in rather long line ups.
I expect there will be backouts due to insufficient gas supplies to co-generators as well. This could even start to happen say about 2005 and it is always possible that it will happen sooner. But I think 2005-2010 is the most likely time frame that these ugly problems start to be visible over the horizon.
Granted this is civil... but look at what Adobe and the US criminal justice system tried to pull off with Dimitris Sklyrov. IMHO these issues are related because the legal system is now being employed to harass and threaten programmers. Any one of us can be a target. 10 years ago we could pursue our careers with very little threat of a law suit. Today - if one has a success then the question becomes how many times over will we be sued.
The whole issue illustrates how fucking preposterous the US legal system has become and other countries are planning to follow suit. Of course we also have countries like Norway and the issue of the DeCSS and Joh Johanson and I have no idea what label should be stapled to this mess. It would be simpler to just take the lawyers involved out behind the barn and get rid of them! But the horrible thing is that the victims of this perverted system are expected to finance it. Next time you are in a courtroom ask yourself of all the people in there - which ones are not being paid?
Here we have a threat to sue an individual (Linus) because he used his own ideas... ideas that apparently an unrelated individual manages to patent in a country (USA) that the person (Linus) doesn't live in.
Then after this flight of stoopidity - people come forth and suggest they will donate to the defense fund. Of course - this simply subsidies the US lawyers who collectively created the problem in the first place.
The bottom line is that this is getting right fucking crasy! Somehow we need to figure out how to counter this.
There are two sets of laws here that are working against us. First is patent law which as it is currently implemented has the following consequences. 1) if you own a valid patent and a large company wants to use what you invented - they will simply claim your patent is invalid and bankrupt you in the courts. 2) if they own an invalid patent then you cannot afford to fight them in the courts. Thus - you cannot do your job. You cannot pusue your career. Here we have intellectual feudalism where the sherrif of cyber notingham tries to turn you into a peasant.
[read up on Leo Farinsworth if you doubt this - he invented television and died a broken man - bankrupt as well - because RCA fucked him over in the courts]
Then the second set of laws are in the same group as the DMCA where we sometimes face criminal charges because perhaps someone wants to play a CD or a DVD and does not want to use software from Microsoft to do it.
-----------
Patents are only valuable to large companies and they are only valuable because they can be used to restrain trade. Given this - large companies pool their patents in a defacto free patent zone. Those on the inside are more or less protected and do not run the risk of litigation. Anyone on the outside is fair game. What a wonderful little oligopoly eh?
Maybe "we" need to start playing this game. Suppose we organised an Open Source Patent Association and paid a feee like $100 bux to join it. This would create a pool of funds whereby the "best" ideas in the open source community could be patented. All members of the association would recieve protection and access to any and all patents. Any closed source shop would be billed or face court action -or- have to pool their patents in order to join.
Since most of the great ideas are invented in the open source community - in short order this association might have a rather wicked sheaf of patents and this could be used to ensure that members of the open source community cannot become victims of bad faith litigation.
I live within walking distance of this university and I am a professional developer and have been for a number of years. Last fall I contacted their IT people and asked if they have any courses on C++ cross platform development. (Rightly or wrongly I elected to use wxWindows and C/C++ from now on - but I still ahve a lot of legacy code of course).
I was suprised at the raw nerve I seemed to have hit with the prof I was speaking to because she became somewhat defensive.
My position is that if we for instance go to sourceforge and check the projects that we will find that C/C++ is perhaps the most popular language for these projects. If I look at my development requirements my conclusion is that C/C++ is THE ONLY viable languge I would even consider using! In my career I have programmed on over 13 platforms and I have used over 13 languages - many of which are now obsolete. I don't think I am biased towards C/C++ or say biased away from say Java. I have my career and at this point in my life I am managing it! I encourge all other programmers to do likewise. What this means is that for me - if a client asks me to program in VB, Java, etc. my answer is that I will NOT take on the job.
Given my strong feelings that C/C++ will be here for the foreseeable future - I find it totally ironic that the U of "C" doesn't even teach "C".
As such - I consider them rather irrelevant.
Furthermore as it turns out I was at the OpenBSD hackathon BBQ last weekend and made the point of asking the hackers how much Java there is in OpenBSD. They laughed. When I asked about C++ they were a little more serious and consided that perhaps there is some somewhere.
So I commented to them that the Uof"C" doesn't teach "C" and was actually quite surpised to hear one chap pipe up that his company doesn't hire UofC IT grads.
I think this is a really sad testiment to the department actually. My opinion is that they have a strong Java / M$ bias and I think this is rather sad. Just MHO...
--------------
BTW - these comments should not be construed to critisize Ruby, Python, Perl, Bash, PHP etc. These langages all have their place and I use some of them. My comments are about the use of C/C++ for general purpose applications development where you might end up with 50,000+ lines of code.
Did he disable these caches? Seems to me if not then he really may not have tested anything. If he did then I can't see why there is so little difference between some of the tests.
Of course, I've never run these benches so I don't know. I'm just asking.
Transmission standards in the U.S., Canada, Korea, Taiwan, and Hong Kong (ANS) ads the rest of the world (ITU-T) evolved from different basic-rate signals in the non-synchronous heirarchy. ANSI Time Division Multiplexing (TDM) combines twenty four 64 kbit/s channes (DS0) into one 1.544 DS1 signal. ITU multiplexes thirty 64bit/s chennels (E0) into one 2.038 Mbit/s E1 signal (an extra two channels provide frame alignment and signalling, making 32 total).
This is from the Tektronix SONET Telecommunications Standard Primer. Copies can be requested from the Tektronix offices, ir visit the website at http://www.tektronix.com
Yup - I agree! The head end can easily be scaled up.
They need to figure out a way to install underground fiber. Someone else posted that Anexiter was selling 6 conductor for about $1 bux per foot. Those were the same people I was talking to about 4 years ago.
Still - this is going to cost them about $7,000+ per mile when they factor in the cost of a cat to bury it. Probably it is money well spent but they will need the revenues from the cable and voice services to help pay for it. At a reasonable ROI and low interest rates they can expect $700 per year or more in financing costs per subscriber.
This reminds me of a coal mine serving a power plant. The railroad ran a loop where trains went round and round, back and forth between the mine and the power plant on a continous basis for years - and at very high hauling rates.
Eventually the coal mine and the power plant managed to get a right of way and started construction of a slurry pipleline. Still the rail rates remained high.
When the pipeline was finally put on stream the railroad slashed their rates to about 1/10th what they had been charging. They slashed the rates so much the slurry pipeline was no longer cost effective.
Funny how competition can bring prices and services into line. I suspect the same thing will happen in this community. Billing rates will remain high and service will remain poor to non-existant until they install an alternative infrastructure. Then magically things that were not possible before with happen.
Of course the telcos don't care because it was the taxpayers that build and paid for their infrastructure years ago. So they can milk it - and if it falls into disuse - so what - it was paid for years ago.
Damn I'm a cynic!
Re:Newest ver. still lock up on 1st website visite
on
Mozilla 1.4b Loosed
·
· Score: 1
I didn't know Intel made a 1.8 Ghz PIII
Maybe this is a debian woody problem. Mozilla actually locks up on a lot of websites.
It would be nice to get to the bottom of the problem and if it is working in unstable then maybe it is not a mozilla problem.
your 56kb actually runs in a DS0 which is 64KB/sec with inline switching so it is 24 lines.
Regardless how the lines are shared - normal browsing is highly asymetrical and so is web serving. The two co-exist quite nicely.
Games can place a very significant demand for bandwidth.
speed and memory management
on
Mozilla 1.4b Loosed
·
· Score: 1, Interesting
IMHO if the mozilla developers organized one thread or one fork per window - they would be better off. If they are interested in doing this - then they should change the way malloc() is handled. Maybe they already have!
Instead of malloc() going off to grab whatever bytes they need for the object in question - they should do two (2) things:
1) organize a [possibly hieracharical] logical identifier for an allocation group. This could be mapped 1:1 to the window in question (window=fork=thread depending on how the code is organized). This requires at least one extra parameter to the code layer that interfaces to malloc(). If they are using "new" then they will need to define their own constructors / destructors but they are still doing the same thing. In essance they write their own mymalloc() and myfree() and add an extra paramter that is their logical grouping identifier (which can be heirarchical)
2) allocate a page aligned block of multiple pages each time an actual "malloc()" is needed. Thereafter - malloc()s can chew into the free space. It is very simle. We call mymalloc() and mymalloc() checks if there is space - if so it returns it and if not it gets space with malloc() but at multiple pages at a time. The amount of memory mymalloc() gets is a tuning parameter.
The job mymalloc() has to do is very very simple. It typically may have zero hunting for holes because it might just operate like a stack.
This accomplishes several things.
a) since all malloc()'s are taking place within a logical memory "object" then you can't have leaks because when the object is no longer needed (as when the window gets closed) then they all get blown away at once.
b) Usually many many pages of physical memory are needed to support the window and underlying fork()/thread() anyways - so the issue of allocating only in multiples of a the machines page size and on a page frame is basically irrelavant. If the memory is not needed now - it will probably be needed very shortly anyways.
c) By allocating memory in this fashion, it is impossible for the operating system to somehow co-mingle memory for a unrelated process or object in a given physical page. This means that if a window goes idle its memory can be swapped out. Only one single reference into a page from something other than the idle window will prevent the page from being elegable to be swapped.
d) An uncontrolled malloc() / new will usually comingle shit from all over the system - severly limiting the systems ability to swap pages out of RAM.
e) malloc() algorithms are actually usually rather complex and thus they tend to go off hunting for holes of the proper size. This hunting is probably more time consuming than many programmers realise. By grouping memory allocations into a logical organization that the programmer KNOWS makes sense - then unnecessary work is eliminated with the side effect, that deallocations might not even be necesary. IE. If say 90% of the objects persist until say the window is closed - then who cares about the other 10%. Leave it. When the window closes, blow away all memory associated with it and all of this memory is multipage allocations anyways. This makes life for the memory manager rather easy. Again - you can't have leaks either.
f) The programmer loses nothing in flexibility because if certain objects he/she is using are not logically associated - then those memory requests can just be left with the existing malloc() or new operators.
g) the time required to write a logical layer over either malloc or new is trivial and can be done in 1/2 a day (poor boy solution) or slightly more time if more sophistication is required.
The benefits far outweigh the costs.
h) I think if the Mozilla developers have not done something like this - then they really need to think about it. Things that fall into a "logical" grouping include the memory for a window. The memory for an ssl connection. The memory for a dialogue such as when somone clicks on bookmarks.
Newest ver. still lock up on 1st website visited!
on
Mozilla 1.4b Loosed
·
· Score: 1
I run debian woody and was running the default mozilla because I was too lazy to get a newer one.
The default version locks up on DOZENs of websites including www.wikipedia.org
So I tried 1.3.1 - it locks up on www.wikipedia.org
So I tried 1.4b - it locks up on www.wikipedia.org
-----------
Alas - it is not ready for prime time even now!
The newer version is faster mind you... that is until it locks up. After that all versions of Mozilla run at the same speed (zero). The only way out of the lockup that I know of is kill.
Funny how a program can just hang on something and thereafter not respond to ANY events! I wonder how they did that. As a programmer - I'd have to work pretty hard to create a deadlock to do that I think!
I just wish the Mozilla people would use one fork/thread per window the way the Konqueror people do it. At least then a person can kill the offending thread. (I've had to do this to konqueror too - because there are websites that Mozilla runs and konqueror locks up on).
Does anyone else have this experiance? BTW - someone else can post this as a bug - I'm still receiving emails from the last bug I reported almost 2 years ago - it is apparently still on the todo list.
Your biggest issue will probably be right-of-ways. If you can get everyone on the community to agree to a right of way then you have a number of options.
If not - look into incorporating your own telco and see if you can get the local authorities on side.
Physical installation will be expensive - but if _can_ look at a fiber link. Last I checked 6 conductor single mode fiber for overhead was only a wee bit more expensive than copper. Underground will probably be similarly expensize.
Last I checked there were ethernet to fiber drivers that ran 100 base-t (2/3 of a T1) for 50 miles and cost under $1000 USD. (allied telesyn for example). This issue here is that the capital cost is not out of line and the capacity is awesome.
With so much capacity you should be able to run local telephone dial up service and TV signals on the same fibre (but I haven't researched how). I just know there is a ton of bandwidth available.
Furthermore the infrastructure if it is put in properly will be viable for the forseeable future. I'd say over 100 years - but with technology who knows - maybe within 5 years something comes along. You have to take that chance. It is better to spend a little extra now and have something that is solid.
T1 will probably not be adequate for your users. But you can look at backbone links and if you do it right - other communities might join you and you can put the big ugly telcos out of their beauracratic misery.
Is it legal for you to ring your neighbours' doorbell? Technically it is trespassing. So when is trespassing not trespassing eh?
If you pop into one of my webservers are you accessing the computer in an "authorized" fashion? How do you know if I'm technically competant enuf to configure it so the people who should have access do have access and the ones who shouldn't have access don't?
If I have my winders file shares open - are you "authorized" to pop in for a look?
I say "YES". I know a person who deliberatly opened her shares because she wanted people to get at her music.
Yet another person who I called who had open shares claimed I hacked the computer. So much for trying to be a nice guy to these idjots.
Actually - on that phone call to tell them the shares were wide open - another person found out I did this and accused me of trying to get someone fired! I mean the bullshit factor is really deep sometimes.
Its like some people are so stupid that they will walk down the street with their damn dicks hanging out and if some one tells them their fly is open - that person is accused of being a peeping tom!
I think spamd is the way to go. Its in the new release of OpenBSD. Of course - spammers will react very quickly and blackhole any OpenBSD protected site.
But that is great for us - because we don't want to hear from them anyways.
This is just part of the evolution of the net. A new species pops up and slowly takes over.
Eventually uncompetative experiments die out completely.
Wouldn't it be funny if it turns out the code SCO is talking about was originally GPL and one of their programmers lifted it? I think this is more likely than an opensource developer taking SCO code.
A programmer working for SCO and under deadlines has a lot more incentive to snitch something than an open source developer.
Scientific Computer Applications incorporated out of Tulsa OK might be a candidate. They handle the MCS coutouring program and this is still one of the best applications of its type even today. I am pretty sure it was available before 1969. I tried to call then to confirm the date of its first release but they were not in. Friday aft - go figure eh?
Slashdot Mirror
Bullshit. Most hellium is harvested from gas wells.
Gawd!!!
I am so impressed with the caliber of the reporting. This artical implies that the USA will become a net exporter of oil complements of its left over turkey parts. hahahaha.
Imagine 4 billion barrels of light turkey crude each year. Presently the US imports about 14 million barrles oil and equivalents per day (about 2/3 of its guttonous appetite). This amounts to about 5 billion barrels. So to hell with the middle east eh?
You make a very good point. Thankyou.
I agree with lobbying to make software patents invalid. The problem is that in the interm patents are used as trading cards between large companies. In general we little guys have no patents and no clout and not enough money to fight a patent suit even if we are in the right.
As for money to throw around... well - there are over 1/2 million developers registered in source forge alone and if each of us buys a membership for $100 bux that works out to $50 million. With a war chest of $50 million we can register quite a large number of patents and still have funds left over to fight suits.
Furthermore if we do end up with a war chest then we'll end up with some large corporations wanting to cross license with us and this will pull their patents into the association.
I figure $100 bux per year is quite reasonable if it eliminates the legal minefield that we presently face. Furthermore, an association like this can lobby to have patents eliminated.
The validity of this patent is not really the issue here. Regardless of whether there is prior art the problem is that programmers cannot generally afford to fight invalid patents. M$ accomplishes its objective which is to prevent programmers from doing their jobs.
This patent is just another example of why WE NEED TO ORGANISE an OPEN SOURCE PATENT ASSOCIATION and each of us needs to throw in $100 bux or $1000 or whatever it takes to finance an organisation that can both patent and fight for us. As a member of an organzation like this we would have the right to use any patents that we hold and we _CAN_ prevent M$ and TI and IBM and everyone else from using these patents. If _our_ organisation simply picks the best ideas we come up with and patents them in very short order we'll have a rather mean shief of patents up our collective sleeves.
Cleary the concept of truth isn't applicable in the USA legal system.
It might be useful if you would educate yourself about the physics involved.
Reactors are quite safe. Furthermore mankind will either enjoy a nuclear future or freeze in the dark. Fossil fuel energy resources are quite limited.
The US DOE for instance forecasts that by 2020 the consumption of natural gas will be up about 489%. They actually forecast that much of this gas will come from Canada.
Well completions have doubled in the last few years and the result of this was a rather modest supply increase in 2001. In 2002 the supply dropped slightly. There is just no way on earth that the Exploration and Production industries can increase gas supplies by any significant amount.
American companies are welcome to come up here and look. Many are. Many are also buying reserves, companies like Burlington for instance who just bought Canadian Hunter Exploration Limited are an example.
The issue is that there is a supply side crunch on its way and we are totally unprepared for it.
So, nuclear will find its way back in rather soon I think. But - I do expect that it will be a ways past 2015 before this happens. Also - I do expect that before nuclear starts comming back there are going to be some rather sharp supply problems and some rather panicy people sitting in rather long line ups.
I expect there will be backouts due to insufficient gas supplies to co-generators as well. This could even start to happen say about 2005 and it is always possible that it will happen sooner. But I think 2005-2010 is the most likely time frame that these ugly problems start to be visible over the horizon.
Granted this is civil... but look at what Adobe and the US criminal justice system tried to pull off with Dimitris Sklyrov. IMHO these issues are related because the legal system is now being employed to harass and threaten programmers. Any one of us can be a target. 10 years ago we could pursue our careers with very little threat of a law suit. Today - if one has a success then the question becomes how many times over will we be sued.
The whole issue illustrates how fucking preposterous the US legal system has become and other countries are planning to follow suit. Of course we also have countries like Norway and the issue of the DeCSS and Joh Johanson and I have no idea what label should be stapled to this mess. It would be simpler to just take the lawyers involved out behind the barn and get rid of them! But the horrible thing is that the victims of this perverted system are expected to finance it. Next time you are in a courtroom ask yourself of all the people in there - which ones are not being paid?
Here we have a threat to sue an individual (Linus) because he used his own ideas... ideas that apparently an unrelated individual manages to patent in a country (USA) that the person (Linus) doesn't live in.
Then after this flight of stoopidity - people come forth and suggest they will donate to the defense fund. Of course - this simply subsidies the US lawyers who collectively created the problem in the first place.
The bottom line is that this is getting right fucking crasy! Somehow we need to figure out how to counter this.
There are two sets of laws here that are working against us. First is patent law which as it is currently implemented has the following consequences. 1) if you own a valid patent and a large company wants to use what you invented - they will simply claim your patent is invalid and bankrupt you in the courts. 2) if they own an invalid patent then you cannot afford to fight them in the courts. Thus - you cannot do your job. You cannot pusue your career. Here we have intellectual feudalism where the sherrif of cyber notingham tries to turn you into a peasant.
[read up on Leo Farinsworth if you doubt this - he invented television and died a broken man - bankrupt as well - because RCA fucked him over in the courts]
Then the second set of laws are in the same group as the DMCA where we sometimes face criminal charges because perhaps someone wants to play a CD or a DVD and does not want to use software from Microsoft to do it.
-----------
Patents are only valuable to large companies and they are only valuable because they can be used to restrain trade. Given this - large companies pool their patents in a defacto free patent zone. Those on the inside are more or less protected and do not run the risk of litigation. Anyone on the outside is fair game. What a wonderful little oligopoly eh?
Maybe "we" need to start playing this game. Suppose we organised an Open Source Patent Association and paid a feee like $100 bux to join it. This would create a pool of funds whereby the "best" ideas in the open source community could be patented. All members of the association would recieve protection and access to any and all patents. Any closed source shop would be billed or face court action -or- have to pool their patents in order to join.
Since most of the great ideas are invented in the open source community - in short order this association might have a rather wicked sheaf of patents and this could be used to ensure that members of the open source community cannot become victims of bad faith litigation.
I live within walking distance of this university and I am a professional developer and have been for a number of years. Last fall I contacted their IT people and asked if they have any courses on C++ cross platform development. (Rightly or wrongly I elected to use wxWindows and C/C++ from now on - but I still ahve a lot of legacy code of course).
I was suprised at the raw nerve I seemed to have hit with the prof I was speaking to because she became somewhat defensive.
My position is that if we for instance go to sourceforge and check the projects that we will find that C/C++ is perhaps the most popular language for these projects. If I look at my development requirements my conclusion is that C/C++ is THE ONLY viable languge I would even consider using! In my career I have programmed on over 13 platforms and I have used over 13 languages - many of which are now obsolete. I don't think I am biased towards C/C++ or say biased away from say Java. I have my career and at this point in my life I am managing it! I encourge all other programmers to do likewise. What this means is that for me - if a client asks me to program in VB, Java, etc. my answer is that I will NOT take on the job.
Given my strong feelings that C/C++ will be here for the foreseeable future - I find it totally ironic that the U of "C" doesn't even teach "C".
As such - I consider them rather irrelevant.
Furthermore as it turns out I was at the OpenBSD hackathon BBQ last weekend and made the point of asking the hackers how much Java there is in OpenBSD. They laughed. When I asked about C++ they were a little more serious and consided that perhaps there is some somewhere.
So I commented to them that the Uof"C" doesn't teach "C" and was actually quite surpised to hear one chap pipe up that his company doesn't hire UofC IT grads.
I think this is a really sad testiment to the department actually. My opinion is that they have a strong Java / M$ bias and I think this is rather sad. Just MHO...
--------------
BTW - these comments should not be construed to critisize Ruby, Python, Perl, Bash, PHP etc. These langages all have their place and I use some of them. My comments are about the use of C/C++ for general purpose applications development where you might end up with 50,000+ lines of code.
Did he disable these caches? Seems to me if not then he really may not have tested anything. If he did then I can't see why there is so little difference between some of the tests.
Of course, I've never run these benches so I don't know. I'm just asking.
From the SONET Reference
Transmission standards in the U.S., Canada, Korea, Taiwan, and Hong Kong (ANS) ads the rest of the world (ITU-T) evolved from different basic-rate signals in the non-synchronous heirarchy. ANSI Time Division Multiplexing (TDM) combines twenty four 64 kbit/s channes (DS0) into one 1.544 DS1 signal. ITU multiplexes thirty 64bit/s chennels (E0) into one 2.038 Mbit/s E1 signal (an extra two channels provide frame alignment and signalling, making 32 total).
This is from the Tektronix SONET Telecommunications Standard Primer. Copies can be requested from the Tektronix offices, ir visit the website at http://www.tektronix.com
Yup - I agree! The head end can easily be scaled up.
They need to figure out a way to install underground fiber. Someone else posted that Anexiter was selling 6 conductor for about $1 bux per foot. Those were the same people I was talking to about 4 years ago.
Still - this is going to cost them about $7,000+ per mile when they factor in the cost of a cat to bury it. Probably it is money well spent but they will need the revenues from the cable and voice services to help pay for it. At a reasonable ROI and low interest rates they can expect $700 per year or more in financing costs per subscriber.
This reminds me of a coal mine serving a power plant. The railroad ran a loop where trains went round and round, back and forth between the mine and the power plant on a continous basis for years - and at very high hauling rates.
Eventually the coal mine and the power plant managed to get a right of way and started construction of a slurry pipleline. Still the rail rates remained high.
When the pipeline was finally put on stream the railroad slashed their rates to about 1/10th what they had been charging. They slashed the rates so much the slurry pipeline was no longer cost effective.
Funny how competition can bring prices and services into line. I suspect the same thing will happen in this community. Billing rates will remain high and service will remain poor to non-existant until they install an alternative infrastructure. Then magically things that were not possible before with happen.
Of course the telcos don't care because it was the taxpayers that build and paid for their infrastructure years ago. So they can milk it - and if it falls into disuse - so what - it was paid for years ago.
Damn I'm a cynic!
I didn't know Intel made a 1.8 Ghz PIII
Maybe this is a debian woody problem. Mozilla actually locks up on a lot of websites.
It would be nice to get to the bottom of the problem and if it is working in unstable then maybe it is not a mozilla problem.
typo. Sorry. I meant OC3 which should be 155mb/sec last I checked. T3 = 45mb/sec
Typically a telco likes to get $500 - $1000 per month for a T1 (24 "circuits" = 24 DS0's). In europe it is an E1 (30 DS0's).
T3 is 30xT1 so it would cost about $15,000 per month minimum and 100base-T would be billed at about $30,000 per month minimum.
One can quickly see that the cost of physical infrastructure is pretty small when faced with billing rates like this.
Actually they are not fucked.
your 56kb actually runs in a DS0 which is 64KB/sec with inline switching so it is 24 lines.
Regardless how the lines are shared - normal browsing is highly asymetrical and so is web serving. The two co-exist quite nicely.
Games can place a very significant demand for bandwidth.
IMHO if the mozilla developers organized one thread or one fork per window - they would be better off. If they are interested in doing this - then they should change the way malloc() is handled. Maybe they already have!
Instead of malloc() going off to grab whatever bytes they need for the object in question - they should do two (2) things:
1) organize a [possibly hieracharical] logical identifier for an allocation group. This could be mapped 1:1 to the window in question (window=fork=thread depending on how the code is organized). This requires at least one extra parameter to the code layer that interfaces to malloc(). If they are using "new" then they will need to define their own constructors / destructors but they are still doing the same thing. In essance they write their own mymalloc() and myfree() and add an extra paramter that is their logical grouping identifier (which can be heirarchical)
2) allocate a page aligned block of multiple pages each time an actual "malloc()" is needed. Thereafter - malloc()s can chew into the free space. It is very simle. We call mymalloc() and mymalloc() checks if there is space - if so it returns it and if not it gets space with malloc() but at multiple pages at a time. The amount of memory mymalloc() gets is a tuning parameter.
The job mymalloc() has to do is very very simple. It typically may have zero hunting for holes because it might just operate like a stack.
This accomplishes several things.
a) since all malloc()'s are taking place within a logical memory "object" then you can't have leaks because when the object is no longer needed (as when the window gets closed) then they all get blown away at once.
b) Usually many many pages of physical memory are needed to support the window and underlying fork()/thread() anyways - so the issue of allocating only in multiples of a the machines page size and on a page frame is basically irrelavant. If the memory is not needed now - it will probably be needed very shortly anyways.
c) By allocating memory in this fashion, it is impossible for the operating system to somehow co-mingle memory for a unrelated process or object in a given physical page. This means that if a window goes idle its memory can be swapped out. Only one single reference into a page from something other than the idle window will prevent the page from being elegable to be swapped.
d) An uncontrolled malloc() / new will usually comingle shit from all over the system - severly limiting the systems ability to swap pages out of RAM.
e) malloc() algorithms are actually usually rather complex and thus they tend to go off hunting for holes of the proper size. This hunting is probably more time consuming than many programmers realise. By grouping memory allocations into a logical organization that the programmer KNOWS makes sense - then unnecessary work is eliminated with the side effect, that deallocations might not even be necesary. IE. If say 90% of the objects persist until say the window is closed - then who cares about the other 10%. Leave it. When the window closes, blow away all memory associated with it and all of this memory is multipage allocations anyways. This makes life for the memory manager rather easy. Again - you can't have leaks either.
f) The programmer loses nothing in flexibility because if certain objects he/she is using are not logically associated - then those memory requests can just be left with the existing malloc() or new operators.
g) the time required to write a logical layer over either malloc or new is trivial and can be done in 1/2 a day (poor boy solution) or slightly more time if more sophistication is required.
The benefits far outweigh the costs.
h) I think if the Mozilla developers have not done something like this - then they really need to think about it. Things that fall into a "logical" grouping include the memory for a window. The memory for an ssl connection. The memory for a dialogue such as when somone clicks on bookmarks.
I run debian woody and was running the default mozilla because I was too lazy to get a newer one.
The default version locks up on DOZENs of websites including www.wikipedia.org
So I tried 1.3.1 - it locks up on www.wikipedia.org
So I tried 1.4b - it locks up on www.wikipedia.org
-----------
Alas - it is not ready for prime time even now!
The newer version is faster mind you... that is until it locks up. After that all versions of Mozilla run at the same speed (zero). The only way out of the lockup that I know of is kill.
Funny how a program can just hang on something and thereafter not respond to ANY events! I wonder how they did that. As a programmer - I'd have to work pretty hard to create a deadlock to do that I think!
I just wish the Mozilla people would use one fork/thread per window the way the Konqueror people do it. At least then a person can kill the offending thread. (I've had to do this to konqueror too - because there are websites that Mozilla runs and konqueror locks up on).
Does anyone else have this experiance? BTW - someone else can post this as a bug - I'm still receiving emails from the last bug I reported almost 2 years ago - it is apparently still on the todo list.
Your biggest issue will probably be right-of-ways. If you can get everyone on the community to agree to a right of way then you have a number of options.
If not - look into incorporating your own telco and see if you can get the local authorities on side.
Physical installation will be expensive - but if _can_ look at a fiber link. Last I checked 6 conductor single mode fiber for overhead was only a wee bit more expensive than copper. Underground will probably be similarly expensize.
Last I checked there were ethernet to fiber drivers that ran 100 base-t (2/3 of a T1) for 50 miles and cost under $1000 USD. (allied telesyn for example). This issue here is that the capital cost is not out of line and the capacity is awesome.
With so much capacity you should be able to run local telephone dial up service and TV signals on the same fibre (but I haven't researched how). I just know there is a ton of bandwidth available.
Furthermore the infrastructure if it is put in properly will be viable for the forseeable future. I'd say over 100 years - but with technology who knows - maybe within 5 years something comes along. You have to take that chance. It is better to spend a little extra now and have something that is solid.
T1 will probably not be adequate for your users. But you can look at backbone links and if you do it right - other communities might join you and you can put the big ugly telcos out of their beauracratic misery.
Good luck.
I think you are employable. You are the kind of person I would hire.
Yes it is a good question.
Is it legal for you to ring your neighbours' doorbell? Technically it is trespassing. So when is trespassing not trespassing eh?
If you pop into one of my webservers are you accessing the computer in an "authorized" fashion? How do you know if I'm technically competant enuf to configure it so the people who should have access do have access and the ones who shouldn't have access don't?
If I have my winders file shares open - are you "authorized" to pop in for a look?
I say "YES". I know a person who deliberatly opened her shares because she wanted people to get at her music.
Yet another person who I called who had open shares claimed I hacked the computer. So much for trying to be a nice guy to these idjots.
Actually - on that phone call to tell them the shares were wide open - another person found out I did this and accused me of trying to get someone fired! I mean the bullshit factor is really deep sometimes.
Its like some people are so stupid that they will walk down the street with their damn dicks hanging out and if some one tells them their fly is open - that person is accused of being a peeping tom!
So - this is a good question.
I think spamd is the way to go. Its in the new release of OpenBSD. Of course - spammers will react very quickly and blackhole any OpenBSD protected site.
But that is great for us - because we don't want to hear from them anyways.
This is just part of the evolution of the net. A new species pops up and slowly takes over.
Eventually uncompetative experiments die out completely.
Wouldn't it be funny if it turns out the code SCO is talking about was originally GPL and one of their programmers lifted it? I think this is more likely than an opensource developer taking SCO code.
A programmer working for SCO and under deadlines has a lot more incentive to snitch something than an open source developer.
Scientific Computer Applications incorporated out of Tulsa OK might be a candidate. They handle the MCS coutouring program and this is still one of the best applications of its type even today. I am pretty sure it was available before 1969. I tried to call then to confirm the date of its first release but they were not in. Friday aft - go figure eh?
Well - the US legal system is pretty bad too.
:-)
We are just dealing with the criminalization of the computer industry that is all
Ignorant people are doing ignorant things.
Hmm - this is puzzling. I went to thestreet.com and no ads. Hmm - no ads at all. There are no pop ups - nothing!
DO I have to disable privoxy and re-enable java perhaps?
Darn! Go figure!