In the case of the GPL software, they are violating the terms of the GPL by not including the text of the license and not providing a link to source code or a written offer to send it.
In the case of the shareware utilities, they are probably violating those licenses as well, since most shareware can only be distributed for "reasonable cost of the media". Plus, they are being mighty deceptive since if you buy DRDOS you are still obligated to also pay the authors of these shareware utilities if you use them.
This is sort of like in the mid/late 90s when you would find those CDROMs of "1200 free programs!" in the Best Buy bargain bin. But it turned out they were all just shareware, which you were expected to register and pay for each individually if you used any of them past the trial. The difference is those were generally close to "cost of the media" and they had disclaimers stating as such. Here DRDOS is trying to foist this shareware off as their own commercial product.
Finally, it appears that they may be violating copyright law if they changed the copyright statement in the output of any of these utilities to say "Copyright DRDOS" instead of "Copyright ".
Slashdot uses the InnoDB engine and not MyISAM2. I'm pretty sure wikipedia uses InnoDB as well.
InnoDB is the one that actually supports transactions, row-level locking, foreign key constraints, triggers, etc. The "produce wrong answers quickly" engine would be MyISAM2, which is much faster but generally does not come close to ACID compliancy. The typical MySQL line is that for a lot of uses "you don't need all that fancy stuff." I guess I agree to a certain extent that for crap like guestbooks and blog entries and light-duty web apps it's appropriate.
And no I don't consider slashdot to be the paradigm of uptime and robustness.
What's sad is that this is a reaction to the sad state of accelerated 3d drivers in linux. Doesn't it seem a bit outrageous that a linux gaming company has to resort to reimplementing all the hardware accelerated features in software -- functionality for which happily dump hundreds of dollars into the latest video cards -- because the current state of linux video drivers is either "Free but no 3D supported" or "Non-free and unstable and hard to install". It's like 1996 all over again... the conversation went something like, "Hey that new Voodoo card looks nice, but who needs dedicated 3D hardware? You can render everything just fine in software using the main CPU." Doesn't this seem like a huge step backwards? Haven't the last 8 or 9 years of gaming shown that the argument that "you can do it all with the CPU" is bunk?
Are you f*cking kidding me? Ask anyone that has ever had to write software for reading and writing mp3 tags (and not just utilizing someone else's id3 tag library) and they will tell you that it is a cess pool of pain.
The original ID3 tag format (v1) was HORRIBLE: - Artist/Title fields limited to 30 characters. - The genre was just a fixed number, an index into a list. This meant that everyone had to agree that 136 is "Christian gangsa rap". This list of genres is done differently by various programs, which means that you really can't reliably use anything but the most common genres like "Rock" that everyone agrees on. - The tag is at the END of the file, so if you are streaming or downloading the file the tag is the very last thing you get.
ID3v1 was a quick and dirty hack made by one person (Eric Kemp) that has a ton of serious problems but since there was nothing better it caught on.
But, people realized that ID3v1 sucks so badly, so they made a whole new standard for tags called ID3v2. This format is the complete opposite: it is intricate and extraordinarily complex to read and parse. It has issues with character encoding (1252 or UTF-8?) so god help you if you have a track name with an accented character. These issues were only really addressed in the later versions of the standard, but hardly any software supports that version. Delightful.
On top of that, because ID3v1 and v2 are completely seperate, it means that they can contain different things and you have to keep both of them in sync. Some software only reads v1, some only writes v2 leaving the v1 tags blank. Some software writes both but only reads one. And because of the above issues of the 30 character limit and the genre issue, it can be impossible to "roundtrip" data between the two so sometimes it's impossible to keep them in sync.
This all ends up meaning that the state of tag technology is just a giant cess pool. You end up with stuff like iTunes that just stores all the metadata in an external database. The fact that mp3 tags work AT ALL is a miracle.
If someone had sat down FRIST and THOUGHT about how to write a tag format and wrote a sensible and practical standard to begin with, it would have eliminated a ton of pain. But instead it's a giant house of cards built on hacks, limitations, and bad design. Mp3 tags are a perfect example of how "de facto" standards can really make programmers' lives terrible. It is almost the poster child argument for why standards bodies are GOOD and SERVE A PURPOSE.
Yeah, and Centos should be able to say that they are "based on the freely available SRPMS of Redhat Enterprise Linux". Oh wait, they can't either. Trademark's a bitch.
IMDB: Built and maintained by volunteers. Uses ads and value added service (like headshots and agent info) to make money. Completely free to use. Has never "pulled the rug out" of below users by radically changing policy. Raw DB dumps havs always available for download in several formats.
Gracenote: Built and maintained by volunteers. Now requires that software developers that query the DB pay for a license. No raw database available. Pulled a massive 180 by declaring that only licensed clients may access the DB, where it used to be free.
As a result, people still contribute prolifically to IMDB, and it remains the undisputed gem in its category. In contrast, most contributors jumped the Gracenote ship long ago and formed FreeDB. Gracenote slowly circling the drain, trying to remain relevent with their idiotic licensing.
You say that IMDB makes a profit as if that's somehow evil. How the heck are they supposed to run a site in the alexa top 100 without a revenue stream? Is slashdot also evil for plastering every page with ads, even though the content is 99% user-contributed? Hell, you can't even download a DB dump of slashdot.
You may have all those problems if you don't qualify for PI space and are stuck with PA, but everything you describe above is precisely the reason why PI allocations are available, with justification. You don't need NAT to achieve any of that.
Re:Why are we hiding from the police, daddy?
on
Vim 6.4 Released
·
· Score: 1
Oh please. That's a load of crap. Whoever convinced you that a program couldn't be both powerful and easy to use was just covering for lack of design.
UltraEdit has almost all of the advanced features that you would ever want in a programmer's editor. And yet, it is a standard windows program - hey look, ^C copies! ^V pastes! Select with the mouse! Move with the arrow keys! No inane modal shit. Easy to use and extremely powerful at the same time.
Fragmentation is one of Linux' strongest features.
Disagree.
You need an easy to use desktop OS? SUSE Linux is your friend, as is Fedora Core. You need an ultra-stable OS for use in the server room? Debian Linux is your friend. You need an extremely customizable OS that can be shaped into anything? Gentoo Linux is you friend. You need somtheing that runs off a CD without requiring any kind of installation? Knoppix is your friend. You need a distro that comes with professional support? Enterprise Linuces are your friends. You need a realtime OS? Something that runs on low-powered legacy systems? Something to help you with system recovery? Whatever you need to do, whichever itch there is to scratch, Linux is your friend. Wherever you want to go today, Linux will go there with you.
Ugh. But they're all mostly the same thing, really. All the major linux distros use glibc for the C library and coreutils for a lot of the userspace. Almost all of the software is from the same upstream sources, so it's not like there are substantial differences in terms of software. What differs is the way it's integrated, the way it's packaged, and what the defaults are. And in that respect, it varies significantly. Consider how vastly different the networking config is in Debian compared to FC. This just makes the end user experience terrible because the system is always subtly different from distro to distro.
Compare the 500 linux distros out there to FreeBSD. FreeBSD is by most people's standards a great server OS as well as a great desktop OS. You can install binary packages or you can compile everything from source with the ports system. But it's all one unified environment. This means for example that you can have a FreeBSD Handbook that comprehensively lists how to do just about anything, with specific commands. You don't have to bother looking anywhere else for a HOWTO or to the packager's READMEs of your distro. It's all in one place.
Compare this to situation with linux. When you want to accomplish something with linux and you don't know how, there are a number of places you have to look. The first is the upstream documentation for the particular package. but these are necessarily distro-neutral. They can give general methods of accomplishing things but they often don't have specifics. And they can't because all the major distros apply local patches, change default file layouts, enable non-stock options and so on. So you must be reliant on your distro to document the quirks of its packaging system, and often times this is not comprehensive. There are HOWTOs, but they too must either devote a lot of effort to explaining how the various distros do it, or they generalize and give generic instructions. So after reading both of those you have to then search mailing lists, READMEs, and so on to find out that the packager decided to do such-and-such and that the config file is actually located in XYZ.
As a long time linux user, the first time I tried FreeBSD I was completely blown away by the quality of the FreeBSD Handbook. It was just night and day. Ever since then I've been disgusted by the fact that every linux disto has to do things just a little bit differently. Having 13,000 different distros might be great for you, but I think it's a terrible thing and consolidation would improve the overall user experience.
I suppose it is a bit of a longshot. But then again, we're talking about the main database engine of the crown jewel of their flagship product. If MySQL AB loses the rights to commercially license the InnoDB engine, that would be an extremely serious dent in their business model. Thus I'm going under the assumption that their contract gives them perpetual rights.
It's kind of like when SCO tried to revoke IBM's right to sell AIX, but IBM's claim was that their contract gave them a perpetual and irrevokable license to the SysV code in AIX (or whatever the details were.)
Maybe you should try reading the second paragraph of my response.
They most certainly could release the modified InnoDB as GPL, because it's already dual licensed GPL+commercial. So, no matter what, it will always live on in GPL form and anyone can fork it. The question is whether their agreement allows them the right to continue distributing past versions under the commercial license after the contract expires. If they have a perpetual right to do this then they can fork and continue the dual license, since any modifications/features they add will be their own work, and they are free to release that work under the same dual license.
it's still not, nor will it likely ever be (never say never) an enterprise solution. Prove to me that mySQL is robust enough to be the backend service for a major bank's mortgage application, for example.
I hope the irony of this being a comment on slashdot is not lost on anyone else.
(That would be slashdot, the site which is powered by a mysql database, the site that serves around 50 million page views per day and is consistently in the alexa top 1000. And yes, I am aware that the usage pattern of a forum/web log site is very different than that of a bank. I'm no mysql fanboy, I don't pretend that it's ready for critical data of an enterprise. I just find it funny that these threads always seem to take place on the very database that everyone loves to hate.)
(5) InnoBase doesn't renew contract with MySQL AB. MySQL developers maintain a fork of the last licensed version of the InnoDB engine before the contract expired, fixing and improving as necessary. They spend the money they saved on the InnoBase licensing to hire more programmers to develop their fork of InnoDB. Neither commercial nor GPL users are affected in the slightest, except perhaps the pace of InnoDB refinement slows. InnoBase scratches their heads wondering what the hell they're supposed to do now with their main meal ticket gone. Oracle execs then inform InnoBase that they were purchased to be dismantled and fire everyone.
From the GPL standpoint, they are absolutely allowed to do this, since the GPL guarantees it. On the commercial side of things it's not a given that they would have the right to do this, as there is a chance that the contract they signed causes them to lose all rights to sell InnoDB commercially when the contract expires. However, if they were smart they would have had language in their contract that allows them perpetual commercial licenses to past versions, such that they could fork it and continue to dual license it without needing further contracts with InnoBase.
Both the Mythbusters and the MIT team devoted a few days or perhaps a week or two to the task. Archimedes presumably had years, maybe decades, to work on this thing. Sure, it takes a long time to manufacture and polish thousands of mirrors -- but they had time and they had plenty of manual labor available. And yes it's very tricky to align all the mirrors correctly, but you can use a rigid structure, and once you get the alignment worked out you can lock the positions down so they can't move.
Granted, the technology and readily available parts of the modern age means that a week of MIT students' time is probably equivalent to 6 months of Greek time.
They did the "chicken gun" myth a couple times because they kept doubting their setup. I didn't get to catch the final conclusion, but I would say that by the time they were done, they had tried everything available to them to see what would happen.
If I recall correctly it took them four setups to test the chicken myth:
The first time they shot at a windshield of a small plane, but they only discovered later that it was not rated for birdstrike at all, so it was a poor target. Conclusion: no difference between frozen and thawed.
Running low on time they decided to shoot the chickens into a hard metal surface and attempt to guage the impact by looking at the high-speed camera footage and counting the number of frames that each took to deform. Conclusion: no difference.
In a later episode, they revisited the myth and this time constructed large foam blocks that were used as targets. However they had problems with deflection when the chicken struck the target, which caused their results to be erratic. One some trials it would stike the center and then veer off to the side, instead of going straight through the foam. Conclusion: no difference.
Finally, they constructed targets out of square sheets of glass, which were layered much like how a martial artist tries to break through a stack of boards. When shot these targets did a good job of absorbing the force, and they could count the number of sheets penetrated to judge the impact. There was a clear difference between frozen and thawed, with the frozen chicken blasting through the entire stack of glass, where the thawed only made it partially through. Their results were repeatable so they concluded that the frozen chickens were indeed better at penetrating during bird strike.
They kept one of the glass targets that was shot with a frozen chicken, because it had a very nice round hole in the center all the way through. It's part of their "backdrop of crazy crap" and you can see it when they are sitting around their table with the blueprints.
The whole "it's based on science!" angle in the article really reminds me of the old claims of phrenology and other quakery. "We can map your brain!" Yeah, whatever.
Uhhh, well for one thing it makes it easy and simple for one user to have more than one Firefox profile. I would be highly annoyed if I had to create a whole seperate user account (and log off and on again) just to have another firefox profile. Imposing a 1:1 mapping of user accounts and Firefox profiles would suck.
The random directory name is done to mitigate the damage should there be some kind of exploit that would allow an attacker to read or write files on disk though Javascript (or whatever). If the profile directory is the same stock name for every installation, then the attacker can just hardcode that path in the exploit and for 99.9% of users he will be able to successfully read their cookies / stored logins / etc. If the directory name is random then it makes it somewhat harder. But it's a double edged sword, since as others have already pointed out, it makes the life of the sysadmin who wishes to do something in the profile via a script harder.
It would have to be the modified version, or whatever is "live" on the server at the time. If not, it would be directly analogous to saying "you can take a GPL program, distribute modified binaries, and ship the unmodified upstream source tarball to satisfy the GPL." That would never fly, and so you would have to assume that the same thing would apply to the source-download-link.
[...] I personally like bittorrent (just one P2P method) and hope to see it incorporated into many other programs, particularly online gaming. This would make downloading the typical 150mb patches much easier [...]
Well some people are trying to do that. The problem is that all the large gaming download sites just love the standard ftp mirror system -- it creates long lines when things are released which gives them the perfect opportunity to sell their "premium" / "gold" level access. If they poured half the resources and bandwidth into seeders on torrents as they do ftp sites, you would see a vast improvement.
Oh dear god, that "-exec rm {}" stuff is still horrible. What happens if you have a file with spaces in the filename? What if you had a file named "-r something"? And what if you're trying to delete 10,000 files, you really do NOT want to invoke rm 10,000 times. Better to use xargs
find "${DATADIR}" -type f -print0 | xargs -0 rm --
This will take care of any arbitrary number of files efficiently, and it does not break when filenames contain spaces or other weird characters.
Slashdot Mirror
Er, that was supposed to be "Copyright " at the end there but slashcode ate my < and >.
In the case of the GPL software, they are violating the terms of the GPL by not including the text of the license and not providing a link to source code or a written offer to send it.
In the case of the shareware utilities, they are probably violating those licenses as well, since most shareware can only be distributed for "reasonable cost of the media". Plus, they are being mighty deceptive since if you buy DRDOS you are still obligated to also pay the authors of these shareware utilities if you use them.
This is sort of like in the mid/late 90s when you would find those CDROMs of "1200 free programs!" in the Best Buy bargain bin. But it turned out they were all just shareware, which you were expected to register and pay for each individually if you used any of them past the trial. The difference is those were generally close to "cost of the media" and they had disclaimers stating as such. Here DRDOS is trying to foist this shareware off as their own commercial product.
Finally, it appears that they may be violating copyright law if they changed the copyright statement in the output of any of these utilities to say "Copyright DRDOS" instead of "Copyright ".
Slashdot uses the InnoDB engine and not MyISAM2. I'm pretty sure wikipedia uses InnoDB as well.
InnoDB is the one that actually supports transactions, row-level locking, foreign key constraints, triggers, etc. The "produce wrong answers quickly" engine would be MyISAM2, which is much faster but generally does not come close to ACID compliancy. The typical MySQL line is that for a lot of uses "you don't need all that fancy stuff." I guess I agree to a certain extent that for crap like guestbooks and blog entries and light-duty web apps it's appropriate.
And no I don't consider slashdot to be the paradigm of uptime and robustness.
What's sad is that this is a reaction to the sad state of accelerated 3d drivers in linux. Doesn't it seem a bit outrageous that a linux gaming company has to resort to reimplementing all the hardware accelerated features in software -- functionality for which happily dump hundreds of dollars into the latest video cards -- because the current state of linux video drivers is either "Free but no 3D supported" or "Non-free and unstable and hard to install". It's like 1996 all over again... the conversation went something like, "Hey that new Voodoo card looks nice, but who needs dedicated 3D hardware? You can render everything just fine in software using the main CPU." Doesn't this seem like a huge step backwards? Haven't the last 8 or 9 years of gaming shown that the argument that "you can do it all with the CPU" is bunk?
Are you f*cking kidding me? Ask anyone that has ever had to write software for reading and writing mp3 tags (and not just utilizing someone else's id3 tag library) and they will tell you that it is a cess pool of pain.
The original ID3 tag format (v1) was HORRIBLE:
- Artist/Title fields limited to 30 characters.
- The genre was just a fixed number, an index into a list. This meant that everyone had to agree that 136 is "Christian gangsa rap". This list of genres is done differently by various programs, which means that you really can't reliably use anything but the most common genres like "Rock" that everyone agrees on.
- The tag is at the END of the file, so if you are streaming or downloading the file the tag is the very last thing you get.
ID3v1 was a quick and dirty hack made by one person (Eric Kemp) that has a ton of serious problems but since there was nothing better it caught on.
But, people realized that ID3v1 sucks so badly, so they made a whole new standard for tags called ID3v2. This format is the complete opposite: it is intricate and extraordinarily complex to read and parse. It has issues with character encoding (1252 or UTF-8?) so god help you if you have a track name with an accented character. These issues were only really addressed in the later versions of the standard, but hardly any software supports that version. Delightful.
On top of that, because ID3v1 and v2 are completely seperate, it means that they can contain different things and you have to keep both of them in sync. Some software only reads v1, some only writes v2 leaving the v1 tags blank. Some software writes both but only reads one. And because of the above issues of the 30 character limit and the genre issue, it can be impossible to "roundtrip" data between the two so sometimes it's impossible to keep them in sync.
This all ends up meaning that the state of tag technology is just a giant cess pool. You end up with stuff like iTunes that just stores all the metadata in an external database. The fact that mp3 tags work AT ALL is a miracle.
If someone had sat down FRIST and THOUGHT about how to write a tag format and wrote a sensible and practical standard to begin with, it would have eliminated a ton of pain. But instead it's a giant house of cards built on hacks, limitations, and bad design. Mp3 tags are a perfect example of how "de facto" standards can really make programmers' lives terrible. It is almost the poster child argument for why standards bodies are GOOD and SERVE A PURPOSE.
Yeah, and Centos should be able to say that they are "based on the freely available SRPMS of Redhat Enterprise Linux". Oh wait, they can't either. Trademark's a bitch.
Come on, that's a terrible comparison:
IMDB:
Built and maintained by volunteers.
Uses ads and value added service (like headshots and agent info) to make money.
Completely free to use.
Has never "pulled the rug out" of below users by radically changing policy.
Raw DB dumps havs always available for download in several formats.
Gracenote:
Built and maintained by volunteers.
Now requires that software developers that query the DB pay for a license.
No raw database available.
Pulled a massive 180 by declaring that only licensed clients may access the DB, where it used to be free.
As a result, people still contribute prolifically to IMDB, and it remains the undisputed gem in its category. In contrast, most contributors jumped the Gracenote ship long ago and formed FreeDB. Gracenote slowly circling the drain, trying to remain relevent with their idiotic licensing.
You say that IMDB makes a profit as if that's somehow evil. How the heck are they supposed to run a site in the alexa top 100 without a revenue stream? Is slashdot also evil for plastering every page with ads, even though the content is 99% user-contributed? Hell, you can't even download a DB dump of slashdot.
And by the way, 500 hosts is more than enough to qualify for PI space. As long as you multihome you can qualify for anything down to a /24.
You may have all those problems if you don't qualify for PI space and are stuck with PA, but everything you describe above is precisely the reason why PI allocations are available, with justification. You don't need NAT to achieve any of that.
There's no reason why anyone would need to view the kernel source code to do a high level type of comparison as in the article. Smart people have already done this, in fact.
Oh please. That's a load of crap. Whoever convinced you that a program couldn't be both powerful and easy to use was just covering for lack of design.
UltraEdit has almost all of the advanced features that you would ever want in a programmer's editor. And yet, it is a standard windows program - hey look, ^C copies! ^V pastes! Select with the mouse! Move with the arrow keys! No inane modal shit. Easy to use and extremely powerful at the same time.
Disagree.
Ugh. But they're all mostly the same thing, really. All the major linux distros use glibc for the C library and coreutils for a lot of the userspace. Almost all of the software is from the same upstream sources, so it's not like there are substantial differences in terms of software. What differs is the way it's integrated, the way it's packaged, and what the defaults are. And in that respect, it varies significantly. Consider how vastly different the networking config is in Debian compared to FC. This just makes the end user experience terrible because the system is always subtly different from distro to distro.
Compare the 500 linux distros out there to FreeBSD. FreeBSD is by most people's standards a great server OS as well as a great desktop OS. You can install binary packages or you can compile everything from source with the ports system. But it's all one unified environment. This means for example that you can have a FreeBSD Handbook that comprehensively lists how to do just about anything, with specific commands. You don't have to bother looking anywhere else for a HOWTO or to the packager's READMEs of your distro. It's all in one place.
Compare this to situation with linux. When you want to accomplish something with linux and you don't know how, there are a number of places you have to look. The first is the upstream documentation for the particular package. but these are necessarily distro-neutral. They can give general methods of accomplishing things but they often don't have specifics. And they can't because all the major distros apply local patches, change default file layouts, enable non-stock options and so on. So you must be reliant on your distro to document the quirks of its packaging system, and often times this is not comprehensive. There are HOWTOs, but they too must either devote a lot of effort to explaining how the various distros do it, or they generalize and give generic instructions. So after reading both of those you have to then search mailing lists, READMEs, and so on to find out that the packager decided to do such-and-such and that the config file is actually located in XYZ.
As a long time linux user, the first time I tried FreeBSD I was completely blown away by the quality of the FreeBSD Handbook. It was just night and day. Ever since then I've been disgusted by the fact that every linux disto has to do things just a little bit differently. Having 13,000 different distros might be great for you, but I think it's a terrible thing and consolidation would improve the overall user experience.
I suppose it is a bit of a longshot. But then again, we're talking about the main database engine of the crown jewel of their flagship product. If MySQL AB loses the rights to commercially license the InnoDB engine, that would be an extremely serious dent in their business model. Thus I'm going under the assumption that their contract gives them perpetual rights.
It's kind of like when SCO tried to revoke IBM's right to sell AIX, but IBM's claim was that their contract gave them a perpetual and irrevokable license to the SysV code in AIX (or whatever the details were.)
Maybe you should try reading the second paragraph of my response.
They most certainly could release the modified InnoDB as GPL, because it's already dual licensed GPL+commercial. So, no matter what, it will always live on in GPL form and anyone can fork it. The question is whether their agreement allows them the right to continue distributing past versions under the commercial license after the contract expires. If they have a perpetual right to do this then they can fork and continue the dual license, since any modifications/features they add will be their own work, and they are free to release that work under the same dual license.
I hope the irony of this being a comment on slashdot is not lost on anyone else.
(That would be slashdot, the site which is powered by a mysql database, the site that serves around 50 million page views per day and is consistently in the alexa top 1000. And yes, I am aware that the usage pattern of a forum/web log site is very different than that of a bank. I'm no mysql fanboy, I don't pretend that it's ready for critical data of an enterprise. I just find it funny that these threads always seem to take place on the very database that everyone loves to hate.)
You forgot:
(5) InnoBase doesn't renew contract with MySQL AB. MySQL developers maintain a fork of the last licensed version of the InnoDB engine before the contract expired, fixing and improving as necessary. They spend the money they saved on the InnoBase licensing to hire more programmers to develop their fork of InnoDB. Neither commercial nor GPL users are affected in the slightest, except perhaps the pace of InnoDB refinement slows. InnoBase scratches their heads wondering what the hell they're supposed to do now with their main meal ticket gone. Oracle execs then inform InnoBase that they were purchased to be dismantled and fire everyone.
From the GPL standpoint, they are absolutely allowed to do this, since the GPL guarantees it. On the commercial side of things it's not a given that they would have the right to do this, as there is a chance that the contract they signed causes them to lose all rights to sell InnoDB commercially when the contract expires. However, if they were smart they would have had language in their contract that allows them perpetual commercial licenses to past versions, such that they could fork it and continue to dual license it without needing further contracts with InnoBase.
Just to play devil's advocate for a moment...
Both the Mythbusters and the MIT team devoted a few days or perhaps a week or two to the task. Archimedes presumably had years, maybe decades, to work on this thing. Sure, it takes a long time to manufacture and polish thousands of mirrors -- but they had time and they had plenty of manual labor available. And yes it's very tricky to align all the mirrors correctly, but you can use a rigid structure, and once you get the alignment worked out you can lock the positions down so they can't move.
Granted, the technology and readily available parts of the modern age means that a week of MIT students' time is probably equivalent to 6 months of Greek time.
I love the fact that this was moderated Informative.
If I recall correctly it took them four setups to test the chicken myth:
They kept one of the glass targets that was shot with a frozen chicken, because it had a very nice round hole in the center all the way through. It's part of their "backdrop of crazy crap" and you can see it when they are sitting around their table with the blueprints.
The whole "it's based on science!" angle in the article really reminds me of the old claims of phrenology and other quakery. "We can map your brain!" Yeah, whatever.
Ahhhh, mid/late 90s web design. I remember it well, and don't miss it one bit.
/flashbacks
The only thing missing was the animated GIF of the letter going into the mailbox.
Oh and the rainbow gradient horizontal divider line.
Uhhh, well for one thing it makes it easy and simple for one user to have more than one Firefox profile. I would be highly annoyed if I had to create a whole seperate user account (and log off and on again) just to have another firefox profile. Imposing a 1:1 mapping of user accounts and Firefox profiles would suck.
The random directory name is done to mitigate the damage should there be some kind of exploit that would allow an attacker to read or write files on disk though Javascript (or whatever). If the profile directory is the same stock name for every installation, then the attacker can just hardcode that path in the exploit and for 99.9% of users he will be able to successfully read their cookies / stored logins / etc. If the directory name is random then it makes it somewhat harder. But it's a double edged sword, since as others have already pointed out, it makes the life of the sysadmin who wishes to do something in the profile via a script harder.
It would have to be the modified version, or whatever is "live" on the server at the time. If not, it would be directly analogous to saying "you can take a GPL program, distribute modified binaries, and ship the unmodified upstream source tarball to satisfy the GPL." That would never fly, and so you would have to assume that the same thing would apply to the source-download-link.
Well some people are trying to do that. The problem is that all the large gaming download sites just love the standard ftp mirror system -- it creates long lines when things are released which gives them the perfect opportunity to sell their "premium" / "gold" level access. If they poured half the resources and bandwidth into seeders on torrents as they do ftp sites, you would see a vast improvement.
Oh dear god, that "-exec rm {}" stuff is still horrible. What happens if you have a file with spaces in the filename? What if you had a file named "-r something"? And what if you're trying to delete 10,000 files, you really do NOT want to invoke rm 10,000 times. Better to use xargs
find "${DATADIR}" -type f -print0 | xargs -0 rm --
This will take care of any arbitrary number of files efficiently, and it does not break when filenames contain spaces or other weird characters.