Even if you could eliminate the problems with the latency, the asymmetric routing that exists in the internet will kill this technique. This communication technique depends on the forward and the reverse path being identical - something which is not true when asymmetric routing is used.
The point the original poster was trying to make is that you should have fun doing your job - not by watching TV, playing games, or whatever it is that floats your boat while on the job. To take the example you quoted : Brett Favre may be enjoying what he is doing - but he is doing his job. Now, if he were enjoying playing football because he gets to pee on the field (as a perk) while playing, then your analogy comes close to matching your argument.
In areas of computation, where computers are provably faster and more reliable that human beings, its makes sense to use computers for tasks like payroll processing, weather forecasting, drug dosage administration, etc. When computers and image recognition software running on computers are empirically much worse than humans for facial recognition (not that humans are very good either, but they beat the pants off of a computer any day), why would you want to use such technology?
This is why the Network Dispatcher product from IBM is so good (the free software equivalent of the technology is called something that I cant remember - sorry). What you have in that case is a solution that only handles the initiation of the connection (SYN packet in the case of TCP) and the ACKs and the rest of the protocol exchange is directly between the client and the origin server. The overhead is minimal.
OTOH, DNS round-robin has the flaws that you mentioned. It is more likely that the DNS server will fail, since the entire connection lifetime is routed through the DNS server. However DNS servers are typically run will one or more secondary servers - which might mitigate that solution.
Have you taken a look at the AIX toolbox(aka the additive that transforms AIX 4.3.3 to AIX5L)? This includes the official IBM blessed version of rpm as an lpp, and once rpm is installed, you can install GCC rpm's from the GNUPro directory.
Of course, since the IBM patches to gcc havent yet been rolled into the main source tree of GCC, the compilers are available as-is. But still this is a vast improvement from the Bull supplied lpp's that had to be installed in/usr/local/bin. Try it - you'll most probably be pleasantly surprised.
I havent tried building those rpm's from source, but the source is also provided (hurrah for the GPL!)
From the description that I read, the principle feature of NOW was a networked filesystem that supported disconnected operation too. Now isnt that exactly what Coda does? Why the desire to re-invent the wheel?
Surely you are joking Mr. V?
on
IBM Wants Linux
·
· Score: 1
Yes, AIX is solid, reliable, and has capabilities that Linux currently does not match yet. But surely you jest when you praise the ODM? This Windows-registry like mess is precisely that : a mess. Let me count the ways in which it ruins the whole experience:
Its a departure from the what you see is what you get configuration scheme of Unices everywhere. As was mentioned by someone else already, changes that you made to the configuration files can disappear after reboots. You and IBM may call this a feature, but to me it is a bug.
It introduces yet another secret initiation ritual that you must learn before you can administer the darn thing.
It is not flexible enough. Even if you go through the secret initiation ritual, you still cant use the ODM to do everything that is needed. Complex tasks frequently require you to edit the boot scripts anyway. So why have the ODM at all?
Another gripe that I have about AIX is the documentation. Have you seen the documentation website? It is slower than mud. Yeah, the documentation can be installed on the system - but good luck to actually finding what you need. The documentation has to be the shoddiest written piece of manuals that I have ever come across. I frequently read the Linux man pages to figure out how to use a command on AIX.
And let me address another point you brought up : support. Yes, IBM support positively rocks! Whats to stop them from supporting Linux the same way?
That would be true for TCP, but not for UDP. TCP promises reliability, whereas UDP applications have to make their own. If wireless protocols had been based on UDP, what would have been the problem? A protocol built on top of UDP could interpret lost packets however it chose to.
Surely you are not stating that their claims are consistent with the facts? I think the word that you are want is "constant". One would expect better English from the editorial staff.
While I totally agree with you that this is a very nasty problem, I dont understand why you impute a "cash cow" motive to not fixing this. Fixing this problem is non-trivial. The fundamental problem with virtual hosting and SSL is that the ip-address (from the ip-level information) is used by the webserver to derive the decryption key that needs to be used to decode the payload. However, with virtual hosting, the ip-address points to many domain names, and thus is not sufficient information to determine the decryption key. The Host header, which would be useful for this, is encrypted within the payload.
Now, you may argue that the SSL protocol could be "fixed" by taking this information outside of the SSL encapsulation and placing it in a plaintext form (as part of an un-encrypted SSL "header"). This would break the HTTP protocol. If this information is merely duplicated, the problem would be solved - but only for HTTP. Remember that SSL can be associated with any TCP based protocol, and for this solution to work, the SSL client would have to determine which protocol is being used, then determine on the basis of this info, which information would have to be duplicated outside of the encrypted payload. This is not a good engineering solution.
Is that "American" as in nationality, or as in currently residing on US soil. How does registration apply to the case of a US citizen creating a copyrightable work in, say, Germany? What about non-US citizens creating copyrightable works within the US?
The link in question describes a method used to do run-time optimization. Clearly runtime optimization is something that statically compiled programs can benefit from too - not just JIT compiled ones. Thus I feel that the static compilation approach will still retain a speed advantage.
Thanks for the link, BTW! It was very educational.
I went to the link referenced in the story about the SSH architecture, but couldnt figure out whether the document was talking about SSH v2 or SSH v1. Or maybe I am asking a stupid question?
I learned programming by reading Niklaus Wirth's classic introduction to programming via Pascal. In that he differentiates between two activities : programming and coding. Programming,
as defined by Wirth, is the art / technique of
finding a solution to a problem using computers. Coding is the implementation of that solution using a programming language. IMHO, most of the problems with software today stems from too many coders (who think incorrectly that they are programmers) and too few programmers.
Informal hacks do have their place. If I am trying to debug a production problem by sifting through 100's of megabytes of data, I am not going to go through a design, code and review process for any data analysis scripts that I write. This is where languages like PERL shine. Unfortunately, coders have decided that the same (absence of) engineering rigour they use in creating informal hacks can be used when developing entire applications. This is why a lot of applications today look and feel like informal hacks. Sadly, most such developers do arise from the PERL community - but PERL itself is not to blame. If engineered correctly, you can write quality software even in PERL.
People who state loudly that a lack of formal training in CS disciplines matters not in application development, havent understood that the biggest payoff from such an education comes from a sense of engineering instilled into the student. Without that, quality software is a random event.
Every time someone makes the statement "hardware is getting cheaper / faster" everyday I see red. Yes it is getting cheaper and faster. But apparently not fast enough to keep up with the lazy coders who make such statements. Because it is people who make such statements that dont care too much about making the best of existing resources and have the Microsoft-like attitude of "who cares about the end-user - using technology x makes my life easier"?
1. This problem has nothing to do with *IX or WinXX. It is OS-independant
2. *IX have more trusted copies because of multiple sources. It would be hard to backdoor every single trusted copy because you would have to do it at the FSF, at Redhat, Slackware, Debian,....you get the idea. This assumes of course, that the original FSF version from which all the other vendors got their trusted copies was good. In other words, if today, someone had to introduce a backdoor, they would have to backdoor all these copies. As opposed to just backdoor-ing the Microsoft trusted copies.
Yes, backdooring of this nature is certainly possible. However, it is fixable too. Just replace all the compilers and linkers and libraries on your system with signature verified binaries available from a trusted vender and then recompile your apps. This is equivalent to getting an authentic CD of the product from your vendor (Microsoft, or RedHat, or the FSF or whoever you choose).
The tough part is detecting the backdoor. I dont believe that there is any good way to do that unless one audits the assembler code with a trusted disassembler.
However, this is really not a disadvantage for *IX. Microsoft can fall prey to the same problem too. In fact, they would probably be in more trouble if this occurred. If by any chance all copies of Micrcosoft's build compiler got hit by the "Thompson poison", they are faced with an uphill task of purification - there is no other trusted source for the compiler.
Could you enlighten me please? I alway thought that "ground" was a reference potential. By definition, and convention, this reference point is always treated as 0. So what is a positive ground and what is a negative ground?
The older generation may remember him - but before there was Michael Crichton, there was Desmond Bagley.
Bagley's books were never very thick. But his writing style was great, and he did have a tendency to introduce science and engineering ideas as key elements of his plots - which is what attracted me to his writing. Its a shame that he is not more well known. Unfortunately, it has been so long since I've read any of his books that I cant remember the title of a single one that I read. (I found a number of titles on Amazon, and I think I read "Night of Error" and "The Enemy" there is no plot summary to confirm that)
CICS is not an operating system. It is a program that runs under operating systems like MVS, VM (not sure), AIX, and (once upon a time) OS/2.
The best of capitalism
on
Disconnected
·
· Score: 1
You seem to be under the impression that American corporations need to operate under some sort of socialistic principles. I'm sorry, but that doesnt compute. No corporation owes you anything - unless you work for it, and then all it owes you is a paycheck. This is capitalism 101.
The reason IBM is in India and improving the technology there is out of pure self-interest. It can tap into the highly skilled cheap labour force to offload projects to India. How the heck can IBM do this if the telcommunications and the computing infrastructure arent upto the task? There is no socialistic motive here, and if IBM chooses to ignore the "massive social problems right here at home" while doing stuff in India, you'll just have to deal with it, wont you?
I believe one of the conditions of the H1-B process is that your sponsoring company needs to show proof to INS that they can and are prepared to pay for your ticket back home. That is why companies like TCS send people to the US with an open return ticket in hand.
Of course, if you want to stay back, it doenst help that the cost of the return trip is taken care of - you will still end up home...
Slashdot Mirror
Even if you could eliminate the problems with the latency, the asymmetric routing that exists in the internet will kill this technique. This communication technique depends on the forward and the reverse path being identical - something which is not true when asymmetric routing is used.
Sounds to me like you are way off base.
The point the original poster was trying to make is that you should have fun doing your job - not by watching TV, playing games, or whatever it is that floats your boat while on the job. To take the example you quoted : Brett Favre may be enjoying what he is doing - but he is doing his job. Now, if he were enjoying playing football because he gets to pee on the field (as a perk) while playing, then your analogy comes close to matching your argument.
In areas of computation, where computers are provably faster and more reliable that human beings, its makes sense to use computers for tasks like payroll processing, weather forecasting, drug dosage administration, etc. When computers and image recognition software running on computers are empirically much worse than humans for facial recognition (not that humans are very good either, but they beat the pants off of a computer any day), why would you want to use such technology?
This is why the Network Dispatcher product from IBM is so good (the free software equivalent of the technology is called something that I cant remember - sorry). What you have in that case is a solution that only handles the initiation of the connection (SYN packet in the case of TCP) and the ACKs and the rest of the protocol exchange is directly between the client and the origin server. The overhead is minimal.
OTOH, DNS round-robin has the flaws that you mentioned. It is more likely that the DNS server will fail, since the entire connection lifetime is routed through the DNS server. However DNS servers are typically run will one or more secondary servers - which might mitigate that solution.
Have you taken a look at the AIX toolbox(aka the additive that transforms AIX 4.3.3 to AIX5L)? This includes the official IBM blessed version of rpm as an lpp, and once rpm is installed, you can install GCC rpm's from the GNUPro directory.
Of course, since the IBM patches to gcc havent yet been rolled into the main source tree of GCC, the compilers are available as-is. But still this is a vast improvement from the Bull supplied lpp's that had to be installed in /usr/local/bin. Try it - you'll most probably be pleasantly surprised.
I havent tried building those rpm's from source, but the source is also provided (hurrah for the GPL!)
From the description that I read, the principle feature of NOW was a networked filesystem that supported disconnected operation too. Now isnt that exactly what Coda does? Why the desire to re-invent the wheel?
Yes, AIX is solid, reliable, and has capabilities that Linux currently does not match yet. But surely you jest when you praise the ODM? This Windows-registry like mess is precisely that : a mess. Let me count the ways in which it ruins the whole experience :
Its a departure from the what you see is what you get configuration scheme of Unices everywhere. As was mentioned by someone else already, changes that you made to the configuration files can disappear after reboots. You and IBM may call this a feature, but to me it is a bug.
It introduces yet another secret initiation ritual that you must learn before you can administer the darn thing.
It is not flexible enough. Even if you go through the secret initiation ritual, you still cant use the ODM to do everything that is needed. Complex tasks frequently require you to edit the boot scripts anyway. So why have the ODM at all?
Another gripe that I have about AIX is the documentation. Have you seen the documentation website? It is slower than mud. Yeah, the documentation can be installed on the system - but good luck to actually finding what you need. The documentation has to be the shoddiest written piece of manuals that I have ever come across. I frequently read the Linux man pages to figure out how to use a command on AIX.
And let me address another point you brought up : support. Yes, IBM support positively rocks! Whats to stop them from supporting Linux the same way?
Does anyone know when support for the C++ "export" keyword is expected? Is someone working on this already?
That would be true for TCP, but not for UDP. TCP promises reliability, whereas UDP applications have to make their own. If wireless protocols had been based on UDP, what would have been the problem? A protocol built on top of UDP could interpret lost packets however it chose to.
Surely you are not stating that their claims are consistent with the facts? I think the word that you are want is "constant". One would expect better English from the editorial staff.
While I totally agree with you that this is a very nasty problem, I dont understand why you impute a "cash cow" motive to not fixing this. Fixing this problem is non-trivial. The fundamental problem with virtual hosting and SSL is that the ip-address (from the ip-level information) is used by the webserver to derive the decryption key that needs to be used to decode the payload. However, with virtual hosting, the ip-address points to many domain names, and thus is not sufficient information to determine the decryption key. The Host header, which would be useful for this, is encrypted within the payload.
Now, you may argue that the SSL protocol could be "fixed" by taking this information outside of the SSL encapsulation and placing it in a plaintext form (as part of an un-encrypted SSL "header"). This would break the HTTP protocol. If this information is merely duplicated, the problem would be solved - but only for HTTP. Remember that SSL can be associated with any TCP based protocol, and for this solution to work, the SSL client would have to determine which protocol is being used, then determine on the basis of this info, which information would have to be duplicated outside of the encrypted payload. This is not a good engineering solution.
Perhaps you have thought of a better solution?
Is that "American" as in nationality, or as in currently residing on US soil. How does registration apply to the case of a US citizen creating a copyrightable work in, say, Germany? What about non-US citizens creating copyrightable works within the US?
I am specifically referring to the "Dynamo" link.
The link in question describes a method used to do run-time optimization. Clearly runtime optimization is something that statically compiled programs can benefit from too - not just JIT compiled ones. Thus I feel that the static compilation approach will still retain a speed advantage.
Thanks for the link, BTW! It was very educational.
I went to the link referenced in the story about the SSH architecture, but couldnt figure out whether the document was talking about SSH v2 or SSH v1. Or maybe I am asking a stupid question?
Anyone who actually makes changes (good or bad) after breaking in isnt a white hat. That is definitely black hat country.
Hear, hear!
I learned programming by reading Niklaus Wirth's classic introduction to programming via Pascal. In that he differentiates between two activities : programming and coding. Programming, as defined by Wirth, is the art / technique of finding a solution to a problem using computers. Coding is the implementation of that solution using a programming language. IMHO, most of the problems with software today stems from too many coders (who think incorrectly that they are programmers) and too few programmers.
Informal hacks do have their place. If I am trying to debug a production problem by sifting through 100's of megabytes of data, I am not going to go through a design, code and review process for any data analysis scripts that I write. This is where languages like PERL shine. Unfortunately, coders have decided that the same (absence of) engineering rigour they use in creating informal hacks can be used when developing entire applications. This is why a lot of applications today look and feel like informal hacks. Sadly, most such developers do arise from the PERL community - but PERL itself is not to blame. If engineered correctly, you can write quality software even in PERL.
People who state loudly that a lack of formal training in CS disciplines matters not in application development, havent understood that the biggest payoff from such an education comes from a sense of engineering instilled into the student. Without that, quality software is a random event.
Every time someone makes the statement "hardware is getting cheaper / faster" everyday I see red. Yes it is getting cheaper and faster. But apparently not fast enough to keep up with the lazy coders who make such statements. Because it is people who make such statements that dont care too much about making the best of existing resources and have the Microsoft-like attitude of "who cares about the end-user - using technology x makes my life easier"?
Not true. He demonstrated a compiler backdoor that would re-insert itself only if compiled with an infected compiler. Reread the ACM paper please.
You've missed the point
1. This problem has nothing to do with *IX or WinXX. It is OS-independant
2. *IX have more trusted copies because of multiple sources. It would be hard to backdoor every single trusted copy because you would have to do it at the FSF, at Redhat, Slackware, Debian,....you get the idea. This assumes of course, that the original FSF version from which all the other vendors got their trusted copies was good. In other words, if today, someone had to introduce a backdoor, they would have to backdoor all these copies. As opposed to just backdoor-ing the Microsoft trusted copies.
Yes, backdooring of this nature is certainly possible. However, it is fixable too. Just replace all the compilers and linkers and libraries on your system with signature verified binaries available from a trusted vender and then recompile your apps. This is equivalent to getting an authentic CD of the product from your vendor (Microsoft, or RedHat, or the FSF or whoever you choose).
The tough part is detecting the backdoor. I dont believe that there is any good way to do that unless one audits the assembler code with a trusted disassembler.
However, this is really not a disadvantage for *IX. Microsoft can fall prey to the same problem too. In fact, they would probably be in more trouble if this occurred. If by any chance all copies of Micrcosoft's build compiler got hit by the "Thompson poison", they are faced with an uphill task of purification - there is no other trusted source for the compiler.
Could you enlighten me please? I alway thought that "ground" was a reference potential. By definition, and convention, this reference point is always treated as 0. So what is a positive ground and what is a negative ground?
The older generation may remember him - but before there was Michael Crichton, there was Desmond Bagley.
Bagley's books were never very thick. But his writing style was great, and he did have a tendency to introduce science and engineering ideas as key elements of his plots - which is what attracted me to his writing. Its a shame that he is not more well known. Unfortunately, it has been so long since I've read any of his books that I cant remember the title of a single one that I read. (I found a number of titles on Amazon, and I think I read "Night of Error" and "The Enemy" there is no plot summary to confirm that)
CICS is not an operating system. It is a program that runs under operating systems like MVS, VM (not sure), AIX, and (once upon a time) OS/2.
You seem to be under the impression that American corporations need to operate under some sort of socialistic principles. I'm sorry, but that doesnt compute. No corporation owes you anything - unless you work for it, and then all it owes you is a paycheck. This is capitalism 101.
The reason IBM is in India and improving the technology there is out of pure self-interest. It can tap into the highly skilled cheap labour force to offload projects to India. How the heck can IBM do this if the telcommunications and the computing infrastructure arent upto the task? There is no socialistic motive here, and if IBM chooses to ignore the "massive social problems right here at home" while doing stuff in India, you'll just have to deal with it, wont you?
I believe one of the conditions of the H1-B process is that your sponsoring company needs to show proof to INS that they can and are prepared to pay for your ticket back home. That is why companies like TCS send people to the US with an open return ticket in hand.
Of course, if you want to stay back, it doenst help that the cost of the return trip is taken care of - you will still end up home...