Just to supplement all the comments from people who say they dual boot -- I have a monitor/kb/mouse switchbox that is connected to my Windows box (used for gaming and graphics apps) and my Solaris x86 box (firewall, nat, mail, code, etc). It's really nice because I don't have to reboot to use apps in the different OSes and I can fileshare between boxes with samba if both OSes need access to the same files. Also, you can switch between OSes with your keyboard with a lot of switchboxes. This makes it friendly to do little tasks like switching out to write an email on the Solaris box without causing Tribes 2 to barf on Windows. It is also not much more expensive than a single dual boot machine because the *NIX box can just be cobbled together out of old parts.
Good thing Sun preemptively suspected Linus might call the next release 3.0 by renaming Solaris 2.7 as Solaris 7. Now Solaris 8 will be *five* better than Linux 3! Of course, Linus could get rid of the 2 as well and go from 2.4 to '5', which would only be three worse.
On a side note, Java 1.2 -> 2 and Solaris 2.7 -> 7 was one of the most obnoxious things I have ever seen Sun marketing do.
this is one of the few cases where the article itself is more flamebait than the/. article pointing to it. businessweek itself admits that macOSX is more secure than 9 and previous versions, but not until it's about halfway down the article. The article is pretty pointless: 'MacOSX will be more of a target for hackers except it won't really and it's actually more secure than 9 anyway'.
One thing that I wish I saw more press about is the security impacts of default configurations. I think that is one of the biggest places where Windows users get bitten in the ass. The 'I LOVE YOU' thing got spread because outlook defaults to blindly running scripts; my company was spared most of the trouble because the sysadmin had changed that default parameter. If you set up an FTP server on Windows 2000 it defaults to allow anonymous connections. We had a developer testing a piece of code he was writing that used ftp and he discovered a couple weeks later that he had a ton of pirate software under the ftp fileroot because he just turned ftp on and didn't look closely at the default options.
OS manufacturers, including *NIX ones, really need to start thinking about their default configurations. If OSX starts up not running any server daemons (as previous posters have claimed), then it is far more secure than most *NIX distributions, most of which will come up with sendmail, telnet, ftp, finger at the minimum.
My ex-advisor is a chair of the IETFworking group researching automated intrusion detection. Currently they are developing a protocol to pass messages between network devices when a potential breach is detected. It's a really complicated field, both in terms of getting a distributed group of network devices to collaborate to decide whether or not something is a deliberate attack, and in creating a security alert protocol that can't be compromised itself.
hmm. gregoriancalendar does not implement serializable but its parent class calendar does. odd that all the fields in gregoriancalendars kept nulling out when i was transmitting them via RMI, maybe i should have just been casting them as a calendar.
i don't use date becase it is depricated, i wish it wasn't because it's straightforward.
Shooting from the hip without having been able to read the content, it is only theoretically impossible to compress random data assuming uniform distribution of all possible characters. usually there will be enough variance in a finite sized randomly generated file to allow for some compression. the trick is making sure that the addition of the decompression map to the file does not make the file bigger than it was in the first place. (that's why using lossless compression on the same thing twice will frequently give you a bigger file)
my one bitch about GregorianCalendar is that it is not serialized. timestamps are a hugely common thing to transmit over a network, and the only non-depricated class that is designed to handle them can't be transmitted remotely without tweaking. bah.
After reading through these comments, I think that it is noteworthy to mention that there seems to be a lot of difference in 2600 meeting quality based on location. Some people seem to have had good experiences, others (notably the ones from NYC and Toronto) said that people didn't seem to know much or were discussing cellphone stealing techniques. That said, I'll add in my $.02 about my own experiences with Seattle meetings about ten years ago when I was a teenager, just after the bust in that Washington DC area mall.
I only went to three meetings because the vibe was so weird. There were four or five groups of people who met sort of hesitantly together and were exceedingly paranoid about talking about much of anything. The only real lengthy discussion we had was at one meeting when it appeared that there was a guy in a suit watching us from a couple floors above and we were speculating as to whether or not he was FBI or Secret Service or something. I have no idea what the Seattle meetings are like now, but I'm going to have to bet that meeting quality varies quite a bit from place to place. I'm in LA now and haven't felt any great urges to show up at the place, even though a focus of my current work is network security.
i'm real interested about the law against depictions of cruelty to animals that is mentioned in the associated article. sounds like a pretty massive first amendment violation.
sort of ironic there's nothing similar against depictions of violence against humans; the same argument should work for a law for that as well (even fake depictions of violence against humans would encourage others to perform such violence, etc).
this is not really new for this sort of agency. one of the first projects like this that i've been aware is CVS which was built by mitre for the NSA to do collaborative video in the WAN. it is built on top of the open source MOO multiuser engine.
it was available for public download a year or so ago, but i can't find it on their site this moment. here's a white paper: http://www.mitre.org/pubs/edge/june_98/sixth.htm
out of curiosity does anyone know the first instance of monofilament weaponry in science fiction books? i remember one from 'neuromancer', and i believe there was a monofilament climbing tool used in one of 'the stainless steel rat' books.
I heard a claim from a 'famous scientist' that major inventions/discoveries take about 60 years to come into common use after they're invented. Anyone know where that one came from? The source was "Cosmic Trigger" by Robert Anton Wilson, but the scientist he was referring to was someone else. Anyways, electricity took about 60 years, wire/telephony took about that long, etc.
To answer the other bit, technologies, in almost all cases, can only be classified by the US government if they were developed with government funding. There may very well be export restrictions on the stuff though.
The problem with this is that the default should be secure, not insecure. Why should a flag be set so that every outlook/exchange admin who is in the know has to go change it on every install that they'd ever do?
Maybe this is one of those things they teach all the MSCE's so that they can prove their credentials are worth something.
I work for an e-consulting firm, and there's a lot of H1's working there. My experience is that the ones that we hire directly tend to be excellent, while the ones that we get through staffing firms tend to not know what they're doing, and are usually paid about 25% or less than what their firm bills us. This makes me think that the big problem is the staffing firms that try to push H1's on the market, rather than H1's in general.
I do think there is a shortage of good IT workers here - I have seen what my management goes through, and there's tons of people who fill their resume with buzzwords that look great, but once you get them in for an interview it becomes immediately clear that they don't know what they are talking about.
A portable manufacturer was doing this with direct writing to the laptop screen a while ago. You just had a laptop sized device, and wrote on it. A look at their website www.grid.com shows me they are still doing it, but catering to a different market. (See the "pen based products") I remember it was being advertised in Wired a ways back (2+ years ago) marketed for regular consumers, and it must not have gotten very far.
Another variant of this is from Wacom. They have a line of their graphics tablets that have a LCD screens behind them so an artist can draw right on the display. I don't think they do handwriting recognition, though.
Multiplexing data over a single fiber is already done with DWDM (Dense Wave Dimension Multiplexing). I have heard of up to 32 colors of OC-192 over a fiber pair in the real world, and over 200 in the lab. It sounds like this new material should be able to push that number even higher. Fiber pairs take up a lot less room than coax though, I wonder what this means in terms of bundling a bunch of them. Another thing to wonder about is if this technology will reduce the number of repeaters needed in a WAN. That's especially important when you talk about areas which are hard to get to, like transatlantic cables.
He has some interesting ideas, but they seem pretty bizarre when one would actually try to put them into practice. I think his big beef should be more with the user interface and less with the computer filesystem. He also has some notions about distributed computing that have been tossed around for at least a decade and are more or less unworkable in reality.
On the latter, he mentions that it would be great that if instead of amazon.com being run off of big servers, it would all run on the client machines. Now this is all well and good for easily distributed processes that are not time sensitive such as Seti@Home where you have a lot of users who are volunteering processing cycles, but that is a totally bizarre idea for a realtime application such as amazon.com. Websites already make use of client processor cycles with client side javascript, Java applets, and flash movies, to name a couple. He seems to be talking about actually distributing amazon.com though, which would imply that clients would communicate with each other. This is totally bizarre for a lot of reasons - the product database would have to be kept in one place, which would mean that amazon would still have to maintain big servers, they would just be one tier back. This is not to mention the issue of security - can one really distribute the checkout and credit card checking process without just attaching back to a single server?
His filesystem ideas seem more or less to just use a relational database model rather than a heirarchical model. That's interesting, but I wonder if he should just look at relational user interfaces rather than a relational filesystem. It is damned easy to delete the wrong thing with an rm -r with a symbolic link in the wrong place, I would hate to find out what the implications of using a relational filesystems would be for operations like that. In my mind, only having one way to get to each file is a good thing.
Some of the ideas involving "lifestreams" are very intriguing to me - I like the notion of users having a presence that can be interacted with directly rather than their computers having a presence. That would certainly help solve some of my email problems.
I think I will stop now as this is getting long.:-)
I saw a fairly slick way to lock a user into a site on teckchek (www.teckcheck.com), which gives technical tests over the web. When you go into a test, it uses javascript to pop up another browser window with no buttons that fills up your entire screen. It rebinds the keys people can use to swap out of it, so until you finish or exit out of the test you are unable to close the window.
I was up at Sun in Feb. for the platinum beta test for Solaris. At that point, here is what they said about the source code:
It is not "Open Source", they are just trying to make as much source available as possible as a service to Solaris users to aid in determining bugs.
The reason they are not immediately opening all of the source is that they have had 200+ subcontractors work on various pieces of the code. Some of these subcontractors are no longer in business so the legal issues pertaining to code release are hazy. Some of these subcontractors may not want source they developed published, and Sun is negoitating with those.
this is a redundant comment by now, but why the hell is slashdot posting an article with the title 'the end of the internet' when all it is about is schools using traffic shaping? traffic shaping and other QoS mechanisms have been around for YEARS. not only is the article just about traffic shaping, but it is about doing it in the LAN and in the uplink to the WAN, it is not like it is about ISPs doing major traffic shaping to throttle napster across the backbones or naything. i wish that/. articles themselves could be moderated down so i didn't have to go read the link it was pointing to and find out it has nothing to do with the slashdot summary of the thing.
I think spectecjr was trying to say that the person who had said that C# is a Java ripoff was wrong, rather than saying that either was better than the other.
Your point about wrapper classes is a good thing to mention, but the rest of the points that were made are irrevelant given the topic of whether or not C# is a Java ripoff.
Slashdot Mirror
Just to supplement all the comments from people who say they dual boot -- I have a monitor/kb/mouse switchbox that is connected to my Windows box (used for gaming and graphics apps) and my Solaris x86 box (firewall, nat, mail, code, etc). It's really nice because I don't have to reboot to use apps in the different OSes and I can fileshare between boxes with samba if both OSes need access to the same files. Also, you can switch between OSes with your keyboard with a lot of switchboxes. This makes it friendly to do little tasks like switching out to write an email on the Solaris box without causing Tribes 2 to barf on Windows. It is also not much more expensive than a single dual boot machine because the *NIX box can just be cobbled together out of old parts.
Good thing Sun preemptively suspected Linus might call the next release 3.0 by renaming Solaris 2.7 as Solaris 7. Now Solaris 8 will be *five* better than Linux 3! Of course, Linus could get rid of the 2 as well and go from 2.4 to '5', which would only be three worse.
On a side note, Java 1.2 -> 2 and Solaris 2.7 -> 7 was one of the most obnoxious things I have ever seen Sun marketing do.
One thing that I wish I saw more press about is the security impacts of default configurations. I think that is one of the biggest places where Windows users get bitten in the ass. The 'I LOVE YOU' thing got spread because outlook defaults to blindly running scripts; my company was spared most of the trouble because the sysadmin had changed that default parameter. If you set up an FTP server on Windows 2000 it defaults to allow anonymous connections. We had a developer testing a piece of code he was writing that used ftp and he discovered a couple weeks later that he had a ton of pirate software under the ftp fileroot because he just turned ftp on and didn't look closely at the default options.
OS manufacturers, including *NIX ones, really need to start thinking about their default configurations. If OSX starts up not running any server daemons (as previous posters have claimed), then it is far more secure than most *NIX distributions, most of which will come up with sendmail, telnet, ftp, finger at the minimum.
My ex-advisor is a chair of the IETF working group researching automated intrusion detection. Currently they are developing a protocol to pass messages between network devices when a potential breach is detected. It's a really complicated field, both in terms of getting a distributed group of network devices to collaborate to decide whether or not something is a deliberate attack, and in creating a security alert protocol that can't be compromised itself.
i don't use date becase it is depricated, i wish it wasn't because it's straightforward.
Shooting from the hip without having been able to read the content, it is only theoretically impossible to compress random data assuming uniform distribution of all possible characters. usually there will be enough variance in a finite sized randomly generated file to allow for some compression. the trick is making sure that the addition of the decompression map to the file does not make the file bigger than it was in the first place. (that's why using lossless compression on the same thing twice will frequently give you a bigger file)
my one bitch about GregorianCalendar is that it is not serialized. timestamps are a hugely common thing to transmit over a network, and the only non-depricated class that is designed to handle them can't be transmitted remotely without tweaking. bah.
I only went to three meetings because the vibe was so weird. There were four or five groups of people who met sort of hesitantly together and were exceedingly paranoid about talking about much of anything. The only real lengthy discussion we had was at one meeting when it appeared that there was a guy in a suit watching us from a couple floors above and we were speculating as to whether or not he was FBI or Secret Service or something. I have no idea what the Seattle meetings are like now, but I'm going to have to bet that meeting quality varies quite a bit from place to place. I'm in LA now and haven't felt any great urges to show up at the place, even though a focus of my current work is network security.
sounds just like the marijuana laws in amsterdam. you can't buy it in bulk or produce it, but you can sell it and buy it in small quantities.
Read the site before you comment. On the HOMEPAGE of www.tradewars.com it says they bought the rights. Sheesh.
i'm real interested about the law against depictions of cruelty to animals that is mentioned in the associated article. sounds like a pretty massive first amendment violation.
sort of ironic there's nothing similar against depictions of violence against humans; the same argument should work for a law for that as well (even fake depictions of violence against humans would encourage others to perform such violence, etc).
--grue
this is not really new for this sort of agency. one of the first projects like this that i've been aware is CVS which was built by mitre for the NSA to do collaborative video in the WAN. it is built on top of the open source MOO multiuser engine.
it was available for public download a year or so ago, but i can't find it on their site this moment. here's a white paper: http://www.mitre.org/pubs/edge/june_98/sixth.htm
out of curiosity does anyone know the first instance of monofilament weaponry in science fiction books? i remember one from 'neuromancer', and i believe there was a monofilament climbing tool used in one of 'the stainless steel rat' books.
To answer the other bit, technologies, in almost all cases, can only be classified by the US government if they were developed with government funding. There may very well be export restrictions on the stuff though.
Maybe this is one of those things they teach all the MSCE's so that they can prove their credentials are worth something.
What were you implying when you said in your second debate with Gore that the Columbine killers' "hearts were blackened by the Internet"?
I do think there is a shortage of good IT workers here - I have seen what my management goes through, and there's tons of people who fill their resume with buzzwords that look great, but once you get them in for an interview it becomes immediately clear that they don't know what they are talking about.
Another variant of this is from Wacom. They have a line of their graphics tablets that have a LCD screens behind them so an artist can draw right on the display. I don't think they do handwriting recognition, though.
Maybe Napster should be ordered to pay RIAA all the money it would have lost during the injunction.
Multiplexing data over a single fiber is already done with DWDM (Dense Wave Dimension Multiplexing). I have heard of up to 32 colors of OC-192 over a fiber pair in the real world, and over 200 in the lab. It sounds like this new material should be able to push that number even higher. Fiber pairs take up a lot less room than coax though, I wonder what this means in terms of bundling a bunch of them. Another thing to wonder about is if this technology will reduce the number of repeaters needed in a WAN. That's especially important when you talk about areas which are hard to get to, like transatlantic cables.
On the latter, he mentions that it would be great that if instead of amazon.com being run off of big servers, it would all run on the client machines. Now this is all well and good for easily distributed processes that are not time sensitive such as Seti@Home where you have a lot of users who are volunteering processing cycles, but that is a totally bizarre idea for a realtime application such as amazon.com. Websites already make use of client processor cycles with client side javascript, Java applets, and flash movies, to name a couple. He seems to be talking about actually distributing amazon.com though, which would imply that clients would communicate with each other. This is totally bizarre for a lot of reasons - the product database would have to be kept in one place, which would mean that amazon would still have to maintain big servers, they would just be one tier back. This is not to mention the issue of security - can one really distribute the checkout and credit card checking process without just attaching back to a single server?
His filesystem ideas seem more or less to just use a relational database model rather than a heirarchical model. That's interesting, but I wonder if he should just look at relational user interfaces rather than a relational filesystem. It is damned easy to delete the wrong thing with an rm -r with a symbolic link in the wrong place, I would hate to find out what the implications of using a relational filesystems would be for operations like that. In my mind, only having one way to get to each file is a good thing.
Some of the ideas involving "lifestreams" are very intriguing to me - I like the notion of users having a presence that can be interacted with directly rather than their computers having a presence. That would certainly help solve some of my email problems.
I think I will stop now as this is getting long. :-)
I saw a fairly slick way to lock a user into a site on teckchek (www.teckcheck.com), which gives technical tests over the web. When you go into a test, it uses javascript to pop up another browser window with no buttons that fills up your entire screen. It rebinds the keys people can use to swap out of it, so until you finish or exit out of the test you are unable to close the window.
this is a redundant comment by now, but why the hell is slashdot posting an article with the title 'the end of the internet' when all it is about is schools using traffic shaping? traffic shaping and other QoS mechanisms have been around for YEARS. not only is the article just about traffic shaping, but it is about doing it in the LAN and in the uplink to the WAN, it is not like it is about ISPs doing major traffic shaping to throttle napster across the backbones or naything. i wish that /. articles themselves could be moderated down so i didn't have to go read the link it was pointing to and find out it has nothing to do with the slashdot summary of the thing.
Your point about wrapper classes is a good thing to mention, but the rest of the points that were made are irrevelant given the topic of whether or not C# is a Java ripoff.