Wouldn't surprise me. Take the case of my University, which has X.Y/16 (no, I'm not dumb enough to specify what X and Y are). The physics department alone takes up the X.Y.1.0/24 to X.Y.10.0/24 subnets (IIRC), and these subnets are almost full up. That's around 4% of the whole class B allocation just for one department.
Suppose you have a 1024-node Beowulf cluster (had to get that in somewhere!), and you need a public address for each node, then you have already used approx 4/256 = 1/64th of your total address space. Hence, nowadays it's not that difficult to fill up a class B.
Of course, why you want every node of your cluster to have a public address is a totally different matter...
Sure, Miller and Urey (Univ. Chicago) did the first of this type of experiment, but Sagan did indeed repeat it later on, with a mixture of gasses more reprasentative of Earth's early atmosphere. Like the original Miller-Urey experiment, amino acids were produced, demonstrating that the first stage of Oparin's chemosynthesis hypothesis is possible and probable.
I think that, out of all the missions the article mentions, Mars Express is the most exciting. This mission, which is backed by the European Space Agency (rather than NASA, as the article implies), carries the British-built Beagle 2 lander, targeted at looking for evidence for Martian life, past and present. Beagle 2 (named after Charles Darwin's ship) is far more sensitive than the old Viking Missions, which were the first (and so far, the only) missions
to look for life. It's worth noting that the more-recent Pathfinder
mission was a proof-of-concept for the two upcoming Mars Exploration Rovers, which are for geological surveys rather than life searches.
One partcularly cool feature of Beagle 2 is its "Mole", which can crawl across the surface (at 1cm/s) and burrow imto the ground or under boulders. The Mole will be able to take samples from locations which the Viking landers couldn't reach; these samples may provide conclusive evidence that life once existed on Mars.
Mars Express, carrying Beagle 2, is due to blast off from the Baikonur Cosmodrome in Kazakhstan on June 2. Fingers crossed!
Bang underside of keyboard very hard for 5 seconds over desk
Sweep up little pile of paper clips, food droppings and cigarette ash which has collected on desk
If keyboard still malfunctions, buy new one, and repeat the following mantra 5 times: 'When playing Team Fortress Classic on-line, neither smoking or drinking make me faster or cooler'.
CDs are no good for backup anymore--way too small. Get yourself an external USB2 drive for backup and just copy over everything, then put it away in a safe and secure place, far away from the computer.
How fast is USB2? I would have thought that an external firewire drive is faster, and firewire seems to be more and more common on PCs today (even my crappy little laptop has a port). Of course, if speed is not important (i.e., you are using the drive for occasional backup rather than day-to-day use), then USB2 may be cheaper.
This isn't directly related to the post, but I'm wondering whether electronic voting stops people from deliberately spoiling their ballot.
In the UK, we still use a paper voting system in general elections, and I (and a number of friends) have deliberately spoilt our ballot papers in past elections, to indicate a RON vote (Re-Open Nominations -- basically, we believe that all of the candidates listed are total wankers, and want other people to stand instead.)
It would be a damn shame if the ability to vote RON is lost, since there will be no other way for people to register their disgust with the slime presenting itself for election.
Who was it who said that the best person for King/President/Emperor was the one who didn't want the job?
Re:Choices, Intel C/Fortran, GCC...
on
GCC 3.3 Released
·
· Score: 2, Informative
At the moment, I wouldn't bother switching to the Intel F95 compiler. Although it is a very good compiler (in terms of efficient code generation), it lags behind other compilers in terms of features supported. By this, I mean the TR-15580 and TR-15581 extensions to the FORTRAN 95 language, which most vendors support, and which are linguistically-important extensions which fix mistakes made in the original FORTRAN 90 language spec. Both extensions are endorsed by J3 (the ISO body which publishes FORTRAN standards) and WG5 (the working group which develops new FORTRAN standars -- at the moment, they are working on FORTRAN 2000).
I am eaglerly awaiting the completion of g95, the GNU gcc-based frontend for FORTRAN 95. At the moment, I use a combination of Intel, Lahey and NAG compilers for my FORTRAN 95 needs, but projects I am programming are specifically geared towards what g95 will support. If you want to program with the future in mind, sure, switch to the Intel F95 compiler; but have in mind that the FORTRAN language itself is advancing, and it might be better to (temporarily) spend some cash and get a compiler which is moving with the the language.
If anyone has root on ANY system or there are ANY non-unix systems, forget it.
Actually, there are two sides to this problem:
I mount someone else's disk
Someone else mounts my disk
Problems caused by case 1 can be handled using the root squash functionality: basically, making all root-UID files on a filesystem which I mount lose their root ownership. This stops Joe Sixpack (who is root on a machine which hosts a remote disk) from copying a SUID-root program onto the remote disk, and then logging into my machine as a normal user, and running the program as a means to get root priv.
Case 2 can be handled by a sensible security policy: only export disks via NFS to machines where you are root. Otherwise, Joe Sixpack can mount your disk, and with his root priv look at anything he wants. However, he still won't be able to compromise your machine (although he may get information which leads to a compromise).
Problems with NFS are at more of a nuts and bolts level - portmap is legendary for having all manner of holes in it, and it remains vulnerable to packet sniffing. Tunneling NFS through an SSH link helps with the sniffing problem, but it creates security problems of its own (see the NFS HOWTO for details).
A few years ago, I was working out in Ghana (Africa), and I was struck by the fact that the highest-denomination banknote was only 5,000 cedis (around $2 at exchange rates back then). This struck me as an inevitable consequence of the high inflation experienced by the country. However, one of my local friends pointed out a secondary factor.
The banknotes currently in circulation in Ghana were printed by the Royal Mint, in the UK. As well as the 1,000, 2,000 and 5,000 cedi denominations, a large number of 10,000 cedi notes were printed. They were shipped out to Ghana in two maritime containers - the huge bastards you see on freight ships.
Unfortunately, one of the containers went missing, probably due to collusion between local crime gangs and corrupt officials. As a consequence, the Government were unable to issue the other container-load of notes. Hence the absence of anything larger than a 5,000 cedi note in Ghana.
Just imagine if the theives had been able to spend the 10,000 cedi notes. Anyone know how many notes you can fit inside, say, a 20-ft container?
I was once working for a leading financial information provider (>16,000 employees), and I was tasked with setting up a listserver, at minimal cost (read no expenditure beyond the cost of my own labour). I ended up opting for an open source SMTP-based listserver, which was far cheaper than the closest closed-source equivalent (free as in beer vs. over 11,000 pounds). I also used Mhonarc to provide a web interface to the message archive, with Perl bits bolted on to add functionality.
I had to put the listserver on the far side of the SMTP gateway, since the company was using some really fucked-up mail system. When I told one of the company software architects that SMTP played a role in the listserver functionality, he told me that SMTP was forbidden on the internal network. He then (very helpfully) pointed out that I should go ahead anyhow, since by the time the PHBs found out, the listserver would be up, running and proving its worth.
I left the company five years ago, but as far as I know the listserver still sees a great deal of use. The moral of this story is this: if the PHBs tell you to solve a problem, don't start evangelising about open source. Just implement the solution in open source, and after six months the software will have proven its worth. Hell, in my case I'm not even sure whether the PHBs realized the listserver was a) using SMTP, or b) using open-source software.
In one of Cordwainer Smith's stories ("Golden the ship was"), there was a spaceship the size of a solar system. Most of it was made from low-density foam; therefore, the overall weight of the ship was minimal, meaning that it was agile as fuck.
The cool thing is that the ship was totally defenseless. However, enemy fleets didn't know this, and were thereby scared shitless by the ship. Great story!
but this is posted every time the topic comes up. NVidia can't release the drivers because of legal reasons. There are things in the code that they do not own, thus cannot release.
OK, the thing to do then is release the code that they do own as GPL, and then package the stuff that they don't own into a binary module -- and tell us who owns this Top Secret(tm) stuff. We can then stop lobbying NVidia and lobby the people who own the secret stuff. Everybody wins -- we get more source code, NVidia stops getting blacklisted, and there is a chance of getting all of the source.
NVidia have been "supporting" Linux for a few years so far. Unfortunately, their drivers have been closed-source, binary-only -- a fact which has caused quite a bit of grief for kernel developers, since it makes it impossible to trace the cause for a kernel oops when using the NVidia drivers.
I did a search through the article for the word "open". I found "OpenGL", but no "open source". So, IMHO, this news release is just PR bullshit (apart from the BSD bit, which may be new) -- there appears to be no move whatsoever for NVidia to open up their source.
I wonder what implications the continuing close-source approach of NVidia will have, what with the upcoming abolition of binary-only modules in Linux kernel 2.6?
They don't keep their jobs if they don't patent or copyright anything. Publish or perish. So rather than continue researching/experimenting until they arrive at the truth they will just strive to create something patentable or copyrightable whether it works or not
For those scientists working for Universities (i.e., not in a private corporation), this comment couldn't be further from the truth. Firstly, the majority of scientists have no interest in (or are actively opposed to) patenting their research results, since it goes totally against the open nature of science. In fact, you will find that the vast majority of "IP" created in Universities is freely-available -- I myself have written a (moderately) succesfull code for modelling stellar seismic activity, which I've published under the GPL. Had I not published it, but patented it instead, I sincerely doubt that it would have achieved the uptake it now has.
Secondly, the first thing I have to do when one of my papers is accepted, is to sign a waiver passing the copyright to the publishers. This is essentially to ensure that all of the papers in a given volume fall under the same copyright. Therefore, although I retain the right to be recognised as the author of my papers, I don't own the copyright over them.
Of course, I'm not claiming that the majority of scientists work for the greater good of humanity. I certainly don't -- I do astronomy because I enjoy it. However, I would like to think that my net contribution to the world, when I die, will be in the positive due to my small additions to the body of human knowledge.
Ok, maybe the people holding the purse strings, but that is about it. The pressure has always been there to "publish or perish" in the academic scientific community.
I think that this hits the nail on the head, and gets to the bottom of the real rot at the heart of physics and related subjects.When a research project is applying for funding, the basic criterion which decides whether that funding is granted is the publication count of those scientists listed as part of the project.
This tends to mean that a result, which 20 years ago would be published in a single paper, now is strung out into 3 or 4 papers:
Introduction
Experiment
Results
Conclusions
Profit! (actually not -- every time I submit a paper to a journal, I sign a form passing the copyright over to the journal publishers. However, I retain the right to be recognised as the originator of the ideas contained in the paper).
What this publication dilution means is that many (if not the majority of) papers published today contain far less content/originality/new ideas than those published 20 or more years ago. This dilution can be traced directly to the manner in which research funding is doled out -- more papers (although maybe less content) == more funding. Just look how
the same thing happened with Enron; the management, fixated with the share price, let every other aspect of the company go to hell.
This drive for publication gluttony is especially penalising to those individuals who publish single-author papers. This is because, typically, the publication count of a given author scales in proportion to the number of co-authors they work with. An example: a group of N collaborators each write a paper, but all include the other N-1 members of the collaboration on the publication list. Net result: N people get N papers where they are a first author or co-author.
Clearly, if N is large, each individual gets lots of papers; if N is small, then your publication count gets screwed. This scaling effect is especially problematical for people who work on theoretical topics: the typical theoretical project has a single scientist working on it (Einsten, Heisenberg, Schrodinger, Heisenberg, Dirac, Feynman etc etc), while some experimental projects can have large numbers of collaborators. One example is particle physics; many of the papers coming out of CERN (the European accelerator located near Geneva) have over 100 authors on them, and the author list takes up the first few pages!
This sort of situation provides the impetus to form large collaborations, since it ups the publication count of each member of the collaboration. However, a secondary property of these large collaborations is that they have a lot of political clout; it is difficult to kill the funding for 50 scientists than it is to kill the funding for one.
So, what's the upshot of the drive to get a large publication list? IMHO, it pushes scientists into collaborations so large that
it penalizes theoreticians, since they naturally work in small groups or on their own,
it becomes impossible to kill funding for large but crap projects, even if these projects are taking funding away from smaller, more-productive projects,
it stifles imagination and progress -- how many profound scientific theories have been developed by a comittee?
So, as long as the culture of funding scientists with large publication lists continues,
we will increasingly move towards a culture of shoddy, comittee-based mediocrity.
Of course, the only way out of this is to judge science on its own merits, rather than by counting numbers of papers published. This is, of course, the ideal of science itself, enshrined in practices such as peer review. However, the ultimate holders of purse string are not scientists -- they're politicians. And those scientists who whisper in these peoples' ears have, more often than not, got where they are because they are more familiar with Machiavelli than Mendeleyev.
As a closing thought, remember Louis De Broglie. His doctoral thesis was 24 pages long. Today, on the basis of such a publication record, De Broglie would be shown the door with little thought. Yet, in those 24 pages, he layed down a large part of the groundwork for quantum mechanics. Go figure.
Slashdot Mirror
Wouldn't surprise me. Take the case of my University, which has X.Y/16 (no, I'm not dumb enough to specify what X and Y are). The physics department alone takes up the X.Y.1.0/24 to X.Y.10.0/24 subnets (IIRC), and these subnets are almost full up. That's around 4% of the whole class B allocation just for one department.
Suppose you have a 1024-node Beowulf cluster (had to get that in somewhere!), and you need a public address for each node, then you have already used approx 4/256 = 1/64th of your total address space. Hence, nowadays it's not that difficult to fill up a class B.
Of course, why you want every node of your cluster to have a public address is a totally different matter...
Given the BSD core of OS X, I wonder:
Sure, Miller and Urey (Univ. Chicago) did the first of this type of experiment, but Sagan did indeed repeat it later on, with a mixture of gasses more reprasentative of Earth's early atmosphere. Like the original Miller-Urey experiment, amino acids were produced, demonstrating that the first stage of Oparin's chemosynthesis hypothesis is possible and probable.
I think that, out of all the missions the article mentions, Mars Express is the most exciting. This mission, which is backed by the European Space Agency (rather than NASA, as the article implies), carries the British-built Beagle 2 lander, targeted at looking for evidence for Martian life, past and present. Beagle 2 (named after Charles Darwin's ship) is far more sensitive than the old Viking Missions, which were the first (and so far, the only) missions to look for life. It's worth noting that the more-recent Pathfinder mission was a proof-of-concept for the two upcoming Mars Exploration Rovers, which are for geological surveys rather than life searches.
One partcularly cool feature of Beagle 2 is its "Mole", which can crawl across the surface (at 1cm/s) and burrow imto the ground or under boulders. The Mole will be able to take samples from locations which the Viking landers couldn't reach; these samples may provide conclusive evidence that life once existed on Mars.
Mars Express, carrying Beagle 2, is due to blast off from the Baikonur Cosmodrome in Kazakhstan on June 2. Fingers crossed!
I find the following technique works well for me:
CDs are no good for backup anymore--way too small. Get yourself an external USB2 drive for backup and just copy over everything, then put it away in a safe and secure place, far away from the computer.
How fast is USB2? I would have thought that an external firewire drive is faster, and firewire seems to be more and more common on PCs today (even my crappy little laptop has a port). Of course, if speed is not important (i.e., you are using the drive for occasional backup rather than day-to-day use), then USB2 may be cheaper.
This is something that serious ergs me as so many people believe it.
C'mon, man! This is the 21st Century -- you should be using SI units, a la: This is something that serious Joules me as so many people believe it.Sheesh!
This isn't directly related to the post, but I'm wondering whether electronic voting stops people from deliberately spoiling their ballot.
In the UK, we still use a paper voting system in general elections, and I (and a number of friends) have deliberately spoilt our ballot papers in past elections, to indicate a RON vote (Re-Open Nominations -- basically, we believe that all of the candidates listed are total wankers, and want other people to stand instead.)
It would be a damn shame if the ability to vote RON is lost, since there will be no other way for people to register their disgust with the slime presenting itself for election.
Who was it who said that the best person for King/President/Emperor was the one who didn't want the job?
At the moment, I wouldn't bother switching to the Intel F95 compiler. Although it is a very good compiler (in terms of efficient code generation), it lags behind other compilers in terms of features supported. By this, I mean the TR-15580 and TR-15581 extensions to the FORTRAN 95 language, which most vendors support, and which are linguistically-important extensions which fix mistakes made in the original FORTRAN 90 language spec. Both extensions are endorsed by J3 (the ISO body which publishes FORTRAN standards) and WG5 (the working group which develops new FORTRAN standars -- at the moment, they are working on FORTRAN 2000).
I am eaglerly awaiting the completion of g95, the GNU gcc-based frontend for FORTRAN 95. At the moment, I use a combination of Intel, Lahey and NAG compilers for my FORTRAN 95 needs, but projects I am programming are specifically geared towards what g95 will support. If you want to program with the future in mind, sure, switch to the Intel F95 compiler; but have in mind that the FORTRAN language itself is advancing, and it might be better to (temporarily) spend some cash and get a compiler which is moving with the the language.
If anyone has root on ANY system or there are ANY non-unix systems, forget it.
Actually, there are two sides to this problem:
Problems caused by case 1 can be handled using the root squash functionality: basically, making all root-UID files on a filesystem which I mount lose their root ownership. This stops Joe Sixpack (who is root on a machine which hosts a remote disk) from copying a SUID-root program onto the remote disk, and then logging into my machine as a normal user, and running the program as a means to get root priv.
Case 2 can be handled by a sensible security policy: only export disks via NFS to machines where you are root. Otherwise, Joe Sixpack can mount your disk, and with his root priv look at anything he wants. However, he still won't be able to compromise your machine (although he may get information which leads to a compromise).
Problems with NFS are at more of a nuts and bolts level - portmap is legendary for having all manner of holes in it, and it remains vulnerable to packet sniffing. Tunneling NFS through an SSH link helps with the sniffing problem, but it creates security problems of its own (see the NFS HOWTO for details).
A few years ago, I was working out in Ghana (Africa), and I was struck by the fact that the highest-denomination banknote was only 5,000 cedis (around $2 at exchange rates back then). This struck me as an inevitable consequence of the high inflation experienced by the country. However, one of my local friends pointed out a secondary factor.
The banknotes currently in circulation in Ghana were printed by the Royal Mint, in the UK. As well as the 1,000, 2,000 and 5,000 cedi denominations, a large number of 10,000 cedi notes were printed. They were shipped out to Ghana in two maritime containers - the huge bastards you see on freight ships.
Unfortunately, one of the containers went missing, probably due to collusion between local crime gangs and corrupt officials. As a consequence, the Government were unable to issue the other container-load of notes. Hence the absence of anything larger than a 5,000 cedi note in Ghana.
Just imagine if the theives had been able to spend the 10,000 cedi notes. Anyone know how many notes you can fit inside, say, a 20-ft container?
I was once working for a leading financial information provider (>16,000 employees), and I was tasked with setting up a listserver, at minimal cost (read no expenditure beyond the cost of my own labour). I ended up opting for an open source SMTP-based listserver, which was far cheaper than the closest closed-source equivalent (free as in beer vs. over 11,000 pounds). I also used Mhonarc to provide a web interface to the message archive, with Perl bits bolted on to add functionality.
I had to put the listserver on the far side of the SMTP gateway, since the company was using some really fucked-up mail system. When I told one of the company software architects that SMTP played a role in the listserver functionality, he told me that SMTP was forbidden on the internal network. He then (very helpfully) pointed out that I should go ahead anyhow, since by the time the PHBs found out, the listserver would be up, running and proving its worth.
I left the company five years ago, but as far as I know the listserver still sees a great deal of use. The moral of this story is this: if the PHBs tell you to solve a problem, don't start evangelising about open source. Just implement the solution in open source, and after six months the software will have proven its worth. Hell, in my case I'm not even sure whether the PHBs realized the listserver was a) using SMTP, or b) using open-source software.
In one of Cordwainer Smith's stories ("Golden the ship was"), there was a spaceship the size of a solar system. Most of it was made from low-density foam; therefore, the overall weight of the ship was minimal, meaning that it was agile as fuck.
The cool thing is that the ship was totally defenseless. However, enemy fleets didn't know this, and were thereby scared shitless by the ship. Great story!
but this is posted every time the topic comes up. NVidia can't release the drivers because of legal reasons. There are things in the code that they do not own, thus cannot release.
OK, the thing to do then is release the code that they do own as GPL, and then package the stuff that they don't own into a binary module -- and tell us who owns this Top Secret(tm) stuff. We can then stop lobbying NVidia and lobby the people who own the secret stuff. Everybody wins -- we get more source code, NVidia stops getting blacklisted, and there is a chance of getting all of the source.
NVidia have been "supporting" Linux for a few years so far. Unfortunately, their drivers have been closed-source, binary-only -- a fact which has caused quite a bit of grief for kernel developers, since it makes it impossible to trace the cause for a kernel oops when using the NVidia drivers.
I did a search through the article for the word "open". I found "OpenGL", but no "open source". So, IMHO, this news release is just PR bullshit (apart from the BSD bit, which may be new) -- there appears to be no move whatsoever for NVidia to open up their source.
I wonder what implications the continuing close-source approach of NVidia will have, what with the upcoming abolition of binary-only modules in Linux kernel 2.6?
They don't keep their jobs if they don't patent or copyright anything. Publish or perish. So rather than continue researching/experimenting until they arrive at the truth they will just strive to create something patentable or copyrightable whether it works or not
For those scientists working for Universities (i.e., not in a private corporation), this comment couldn't be further from the truth. Firstly, the majority of scientists have no interest in (or are actively opposed to) patenting their research results, since it goes totally against the open nature of science. In fact, you will find that the vast majority of "IP" created in Universities is freely-available -- I myself have written a (moderately) succesfull code for modelling stellar seismic activity, which I've published under the GPL. Had I not published it, but patented it instead, I sincerely doubt that it would have achieved the uptake it now has.
Secondly, the first thing I have to do when one of my papers is accepted, is to sign a waiver passing the copyright to the publishers. This is essentially to ensure that all of the papers in a given volume fall under the same copyright. Therefore, although I retain the right to be recognised as the author of my papers, I don't own the copyright over them.
Of course, I'm not claiming that the majority of scientists work for the greater good of humanity. I certainly don't -- I do astronomy because I enjoy it. However, I would like to think that my net contribution to the world, when I die, will be in the positive due to my small additions to the body of human knowledge.
Ok, maybe the people holding the purse strings, but that is about it. The pressure has always been there to "publish or perish" in the academic scientific community.
I think that this hits the nail on the head, and gets to the bottom of the real rot at the heart of physics and related subjects.When a research project is applying for funding, the basic criterion which decides whether that funding is granted is the publication count of those scientists listed as part of the project.
This tends to mean that a result, which 20 years ago would be published in a single paper, now is strung out into 3 or 4 papers:
What this publication dilution means is that many (if not the majority of) papers published today contain far less content/originality/new ideas than those published 20 or more years ago. This dilution can be traced directly to the manner in which research funding is doled out -- more papers (although maybe less content) == more funding. Just look how the same thing happened with Enron; the management, fixated with the share price, let every other aspect of the company go to hell.
This drive for publication gluttony is especially penalising to those individuals who publish single-author papers. This is because, typically, the publication count of a given author scales in proportion to the number of co-authors they work with. An example: a group of N collaborators each write a paper, but all include the other N-1 members of the collaboration on the publication list. Net result: N people get N papers where they are a first author or co-author.
Clearly, if N is large, each individual gets lots of papers; if N is small, then your publication count gets screwed. This scaling effect is especially problematical for people who work on theoretical topics: the typical theoretical project has a single scientist working on it (Einsten, Heisenberg, Schrodinger, Heisenberg, Dirac, Feynman etc etc), while some experimental projects can have large numbers of collaborators. One example is particle physics; many of the papers coming out of CERN (the European accelerator located near Geneva) have over 100 authors on them, and the author list takes up the first few pages!
This sort of situation provides the impetus to form large collaborations, since it ups the publication count of each member of the collaboration. However, a secondary property of these large collaborations is that they have a lot of political clout; it is difficult to kill the funding for 50 scientists than it is to kill the funding for one.
So, what's the upshot of the drive to get a large publication list? IMHO, it pushes scientists into collaborations so large that
So, as long as the culture of funding scientists with large publication lists continues, we will increasingly move towards a culture of shoddy, comittee-based mediocrity.
Of course, the only way out of this is to judge science on its own merits, rather than by counting numbers of papers published. This is, of course, the ideal of science itself, enshrined in practices such as peer review. However, the ultimate holders of purse string are not scientists -- they're politicians. And those scientists who whisper in these peoples' ears have, more often than not, got where they are because they are more familiar with Machiavelli than Mendeleyev.
As a closing thought, remember Louis De Broglie. His doctoral thesis was 24 pages long. Today, on the basis of such a publication record, De Broglie would be shown the door with little thought. Yet, in those 24 pages, he layed down a large part of the groundwork for quantum mechanics. Go figure.