Please tell me this message was supposed to be posted next saterday.
It would indeed be an excellent April Fool's day joke, given how many people seriously reply to this. Of course it can not be serious. It's surprising however that no-one realizes that this must be joke. If the poster had given the similar silly argument "we should abandon all airplane flights -- after all, who ever needs to ever visit people in other states, or -heaven forbid- other countries", then I think the joke would be too obvious.
Actually, given the many serious replies, it makes makes me realize that a lot of people apparently still don't grasp what the Internet (or thinking globally) is about. If you really grok it, then the only conclussion can be that this article is a prank.
CNIT in Italy has reached up to 2.5 Tb/s; I do not know the details, but I once witnessed a presentation by one of their scientists, Gianluca Meloni. He seem to have a paper published in proceedings of ECOC 2005, called "10GHz to 2.5THz Optical Frequency Multiplication". Surely that contains more information.
By the way -- 0.5s * 1Tb/s = 500 Gbit = 64 GByte = 58 GiByte. Pretty long movie, I'd say:-)
I fear you missed the boat too, when it comes to the history:
Indeed, you are right that the patent is NOT AT ALL IPv6 related. So both PUBPAT, ZDNet and Slashdot are wrong here.
It is related to IPv4 link-local addresses. Indeed, the 169.254/16. address you mention. You're correct there. Now, please read RFC 3927 for all details.
As you can see in Appendix A of RFC 3927, these "IPv4 link-local addresses" where implement in Windows 98 and Mac OS 8. You missed the boat completely here.
This technology is NOT "owned" by MS as you claim.
However, both Microsoft AND Apple too(!) do claim IPR over this technology.
It is VERY interesting to note that both Apple's patents were submitted March 21, 1985, and have thus expired last week (!).
Microsofts patent is so new (filed 1998) it's laughable. There is VERY, VERY mucht prior art. Take Mac OS 8 (released 1997).
Please, read before responding:
RFC 3927, Dynamic Configuration of IPv4 Link-Local Addresses.
Both Apple's and MicroSoft patent claims are also old news:
Interestingly enough, Apple also claims IPR on IPv4 Link-Local addresses (which, _finally_ have been standardized this month in RFC 3927). However, those patents have been filed in March 1985, meaning that they just expired (If I understand correctly that patents expire 20 years after their filing).
He didn't actually sign it. He problaby did get to show a few (licence + NDA) agreements and just clicked OK, before sharing it with a couple of friends.
Now, do you read every licence you get with your software. I mean EACH AND EVERY ONE? I did read it about five times. And about 500 times, I just click "Accept" without reading it. Can you honestly say you never violated one ever? For example, when you get a new computer, do you go through each and every licence of every program to check if you can transfer it to you new computer (you can if the licence is 'only one user', but not if it says 'only on one computer').
I agree the kid was wrong in retrospect, and he admits that. However, I'm question you if you live up to the letter of each and every agreement you get, especially when it comes to software!
Zeroconf consists of 3 things: (1) IPv4 link local configuration; (2) translate between names and IP without DNS and (3) service discovery. (let me ignore allocating IP multicast addresses)
ZeroConf, at least, the IETF workgroup of that name, is now about to close, and only tackled the first item. Item (2) is tackled both by multicast DNS (by Apple in Rendezvous) as well as LLMNR (in the less-used DNSEXT IETF workgroup). Item 3 is tackled either with DNS-SD (by Apple in Rendezvous), as well as by uglier protocols like UPnP.
Though the problem is the same, and is called "ZeroConf", the solution differ. Clearly Apple has chosen LL + mDNS + DNS-SD
But admittedly, I'm sure Taco didn't know this.:-) For most common talk: zeroconf = rendezvous.
I would say that ZeroConf and multicast DNS are great for small (home) networks (which don't have a DNS server), and other technologies, like SRV records in DNS, are great for a medium or large network.
One thing I have to applaud Apple for is the awesome integration of Java and Objective C. For example, you can easily write one class in Objective C and the other in Java. Cocoa typically handles things like the translation of a java.lang.String to a (Cocoa) NSString and visa versa. Just awesome.
XCode also happens to be very modular, so adding additional "support" to XCode for additional languages should be something third parties are perfectly capable of.
Yes, this is true. However, it is far from trivial. Beside the documentation, there will be issues on integration with the Cocoa API, and with integration with the XCode and Interface Builder (IB) tools. All of these three will be hard, IMHO.
For example, as a former Pascal programmer, I am interested in attempts to add support for gpc to Xcode. However, this hasn't been picked up yet. Issues that I am vaguely aware of are:
Language integration. Objects are implemented differently in Pascal and Java, C++ and Objective C. In fact, Metrowerks Pascal, Delphi Pascal and the Object Pascal standard do it differently. Please correct me if I'm wrong (I'm not an expert), but I understand that some languages only create a new pointer if you assign, and other languages create the whole object structure. This will pose problems with integration. Beside the obvious object support, a language probably needs to support features like try-catch to make a good integration with the Cocoa API.
Keeping up with the Cocoa API. When Apple supported Pascal, they had a up-to-date Carbon API available for Pascal (Obviously, Carbon was originally written in Pascal, later parts in C or C++). The last few years, it is obvious that the API is released in C++ and automatically translated to Pascal. However, this means that there is a continuous struggle to keep it up-to-date. There are many flaws were this translation goes astray, for example because of the lack of "var" parameters in C (C uses pointers for that), so the API ends up with "var int {CONST} param" uglyness. And this is for a now pretty static API, Carbon, which had relative good Pascal support. I'm sure that translating the Cocoa API will face these problems as well.
Adding support to XCode and Interface Builder. For example, with IB, you can now only specify actions and targets (the interfaces to your classes) for either Java or Objective C. Though I'm not sure if there is need to alter IB, even if not, this WILL make it harder for novice programmers to use this, if only because they can't automatically generate the class files.
At the recent Supercomputing 2003 conference, quite a few exhibitors demonstrated hughe images. Though this mayb be the first (or one of the first) non-scientific usage, it is not unique.
I would recommend anyone to view these type of images on a tile display. At SC 2003, at least EVL and SGI did show dome impressive demo's (in particular, SGI did show some interesting geographic imagery software).
An outcome of this could be that ICANN is replaced with a governement-led organisation. ISOC, the Internet Society, is worried about this, and has issued a statement with comments on this matter.
In fact, they did very much push to get this topic on the agenda, and even sent a mail last week asking all memebers to fill in a survey on this matter.
[...] during preparations for the World Summit for Information Society (WSIS), the top-level, United Nations-sponsored conference to be held in December, several government delegations have suggested replacing ICANN with an inter-governmental body. (Details at
www.wsis.org.
A new virus released by spammers on Saturday 1st November is infecting computers worldwide, and this time the purpose of the virus is to attack www.Spamhaus.org. [...]
Reading the reports, I'd say Apple should fix #1 and #3. I hope they will. But this is not quite anything urgent: the first is an apparent bug if you turn on core files. Well, they are off by default. The second bugtraq has two parts: the first half should be fixed by Apple (changing permissions of dirs when copying between disk images); the second half is not Apple's fault. The argv[] buffer overflow is a stability issue, not a security issue IMHO, And since it so extremely rare, I don't consider it a bug deal.
I think the permission thing is the biggest deal. If you are concerned, you can run these commands: % find/Applications -type d -perm -002 -print (the article mentions an autofix with: find/Applications -type d -exec chmod o-w {} \; but this may break apps) %/usr/sbin/diskutil verifyPermissions $diskname (where diskname might be / ) This command (available in MacOS x.2 and higher), will compare the permissions to that of the original installer (as stored in/Library/Receipts). To repair, type: %/usr/sbin/diskutil repairPermissions/
I got one of those "air-space" developed pens (you can buy them as a gift). The trick is that the ink is pressurized. Well, it's soo-soo. It did work on wet bathroom tiles, but just any pen it stoped when it was half-full.
No, I just prefer the solution the Russians are said to use: no fancy pressured pens. They just used plain crayons...those work great.
Indeed, there does not seem to be a Greec education store; in fact, there does not seem to be an Greec Apple Store at all (See list of International stores.
For US-based addresses it's either MapQuest or Yahoo maps. I think they are very simular.
For Dutch maps (where I live), I use locatienet or Andes. The first one being slightly better.
There are way too many options nowadays. See Oddens for a collection of links, including to historic maps (not useful if you just want to find an address, only for the curious of heart).
Well, I certainly hope these MO drives are more reliable then the older ones. About five years ago, I believed the claim that MO was more reliable then plain hard disks. This was in the time that a 700 MB harddisk did cost me $350. So I bought a Maxtor MO driver for over $1000. It did work great for a year, and I expected my data was more safe then on a HD, because these things where in caddies and reading was done optical, instead of magentical. I was wrong, after about a year, the disks got bad sectors without warning. About the same happened half a year to my dad who also got a MO drive. It was plainly the worst hardware I ever bought. Period.
Nowadays, I use a hard disk for daily backups (just a second 120 GB disk), and DVD's for monthly backup, which I just put in a box in an other building.
So far, I never had problems. When I expect one of my drives is starting to fail, I just replace it. A 120 GB disk is so cheap nowadays. And Disks are so lovely fast (and it can be automated, since there is no tape to be mounted manually). For corporate backups, I would probably just use 2 DVD's, and store them in seperate locations in well conditioned rooms.
It might be bad faith, but nowadays I just wouldn't trust an MO anymore. Sorry if that does wrong on the improved reliability.
I was reasoning that making a back-to-back connection between computers would never give you the LSR award: you would loose on distance.
Of course, this does not apply to throwing a book: That would be the correct unit (bit-meter/second). However, you'll have to throw REALLY fast. Take the example of frisbeeing a DVD across the room. Let's a asume the DVD takes 5 Gigabyte (about 40 Gigabit for sake of simplicity), and it is thrown at 10 meter/second (dunno, I estimate it's between 10-20 m/s). Then your data movement speed would be 400,000 megabit-meter/second.
That would put you off by a factor 100,000 of the current record (7000 km * 5.5 Gbit/s = 38,500,000,000 megabit-meter/second, provided I didn't make any errors in here).
Either you need to throw a lot faster, or throw more books/DVDs. That would be the multi-stream variant of the awared;-)
To win a Landspeed Record (that is what we're talking about), you do not to get just a high bandwidth, but you got to have a high bandwidth across a big distance.
That's why the parameter you need to blow is "megabit-meters/second", not just megabit/second. At least that's what it says on the award I'm just holding in my hand (one of the older awards which was already shattered to pieces previous year;-) ).
First, this "innovation" is not unique: though the fuss is much bigger with the.com and.net domains, VeriSign correctly claims they are not the only one to deplay top-level zone wildcards.:.cc.cx.mp.museum.nu.ph.pw.td.tk and.ws do so as well.
The slashdot article suggest that a standard was broken. It is not, and the editorial does not mention it. For the record, the Internet Architecture Board wrote:
"We must emphasize that, technically, this was a legitimate use of wildcard records that did not in any way violate the DNS specifications themselves. One of our main points here is that simply complying with the letter of the protocol specification is not sufficient to ensure the operational stability of the applications which depend on the DNS: there are protocol features which simply are not safe to use in some circumstances." --
IAB Commentary: Architectural Concerns on the use of DNS Wildcards
Last, contrary to what VeriSign claims, it DOES break (parts) of the Internet. See all examples mentioned in the IAB advisory above. For example, consider what happens if you type "myprinter.mydoman.com" instead of "myprinter.mydomain.com" in your CUPS configuration. Without the wildcard, you get a simple message that you made a typo. With the wildcard, you have a hell of a job finding out why you can't connect to your printer.
Though I applaud ICANN this time for forcing VeriSign to it's knees, I'm surprised that they do not tacle all top-level wildcards:
Though the.com and.net wildcards have much more impact, there were and still are wildcards at:.cc,.cx,.museum,.nu,.ph,.td,.tk, and.ws.
Why are these not removed? This seems like unfair ruling by ICANN.
According to the Internet Architecture Board (IAB):
"We must emphasize that, technically, this was a legitimate use of wildcard records that did not in any way violate the DNS specifications themselves. One of our main points here is that simply complying with the letter of the protocol specification is not sufficient to ensure the operational stability of the applications which depend on the DNS: there are protocol features which simply are not safe to use in some circumstances."
I don't expect that a single lawsuit will change EMI's policy.
However, if a CD is judged "defective" if it does not adhere to standards, I wonder if the same logic can be applied to software. I've got plenty of software that works for 90%, but does not completely follow the standard. Can I go to court and get a refund?
Slashdot Mirror
Please tell me this message was supposed to be posted next saterday.
It would indeed be an excellent April Fool's day joke, given how many people seriously reply to this. Of course it can not be serious. It's surprising however that no-one realizes that this must be joke. If the poster had given the similar silly argument "we should abandon all airplane flights -- after all, who ever needs to ever visit people in other states, or -heaven forbid- other countries", then I think the joke would be too obvious.
Actually, given the many serious replies, it makes makes me realize that a lot of people apparently still don't grasp what the Internet (or thinking globally) is about. If you really grok it, then the only conclussion can be that this article is a prank.
CNIT in Italy has reached up to 2.5 Tb/s; I do not know the details, but I once witnessed a presentation by one of their scientists, Gianluca Meloni. He seem to have a paper published in proceedings of ECOC 2005, called "10GHz to 2.5THz Optical Frequency Multiplication". Surely that contains more information.
:-)
By the way -- 0.5s * 1Tb/s = 500 Gbit = 64 GByte = 58 GiByte. Pretty long movie, I'd say
- Indeed, you are right that the patent is NOT AT ALL IPv6 related. So both PUBPAT, ZDNet and Slashdot are wrong here.
- It is related to IPv4 link-local addresses. Indeed, the 169.254/16. address you mention. You're correct there. Now, please read RFC 3927 for all details.
- As you can see in Appendix A of RFC 3927, these "IPv4 link-local addresses" where implement in Windows 98 and Mac OS 8. You missed the boat completely here.
- This technology is NOT "owned" by MS as you claim.
- However, both Microsoft AND Apple too(!) do claim IPR over this technology.
- It is VERY interesting to note that both Apple's patents were submitted March 21, 1985, and have thus expired last week (!).
- Microsofts patent is so new (filed 1998) it's laughable. There is VERY, VERY mucht prior art. Take Mac OS 8 (released 1997).
Please, read before responding:- RFC 3927, Dynamic Configuration of IPv4 Link-Local Addresses.
Both Apple's and MicroSoft patent claims are also old news:Interestingly enough, Apple also claims IPR on IPv4 Link-Local addresses (which, _finally_ have been standardized this month in RFC 3927). However, those patents have been filed in March 1985, meaning that they just expired (If I understand correctly that patents expire 20 years after their filing).
- zeroconf-ipv4-linklocal.txt
http://www.ietf.org/ietf/IPR/apple-ipr-draft-ietf
He didn't actually sign it. He problaby did get to show a few (licence + NDA) agreements and just clicked OK, before sharing it with a couple of friends.
Now, do you read every licence you get with your software. I mean EACH AND EVERY ONE? I did read it about five times. And about 500 times, I just click "Accept" without reading it. Can you honestly say you never violated one ever? For example, when you get a new computer, do you go through each and every licence of every program to check if you can transfer it to you new computer (you can if the licence is 'only one user', but not if it says 'only on one computer').
I agree the kid was wrong in retrospect, and he admits that. However, I'm question you if you live up to the letter of each and every agreement you get, especially when it comes to software!
I usually try to fill in the address of the company, so that in case they start sending junk mail, they just get it back.
And to prevent them from gathering statistical data, at parties I often if I can change cards with other people. Always fun!
Zeroconf consists of 3 things: (1) IPv4 link local configuration; (2) translate between names and IP without DNS and (3) service discovery. (let me ignore allocating IP multicast addresses)
:-)
ZeroConf, at least, the IETF workgroup of that name, is now about to close, and only tackled the first item. Item (2) is tackled both by multicast DNS (by Apple in Rendezvous) as well as LLMNR (in the less-used DNSEXT IETF workgroup). Item 3 is tackled either with DNS-SD (by Apple in Rendezvous), as well as by uglier protocols like UPnP.
Though the problem is the same, and is called "ZeroConf", the solution differ. Clearly Apple has chosen LL + mDNS + DNS-SD
But admittedly, I'm sure Taco didn't know this.
For most common talk: zeroconf = rendezvous.
I would say that ZeroConf and multicast DNS are great for small (home) networks (which don't have a DNS server), and other technologies, like SRV records in DNS, are great for a medium or large network.
One thing I have to applaud Apple for is the awesome integration of Java and Objective C. For example, you can easily write one class in Objective C and the other in Java. Cocoa typically handles things like the translation of a java.lang.String to a (Cocoa) NSString and visa versa. Just awesome.
XCode also happens to be very modular, so adding additional "support" to XCode for additional languages should be something third parties are perfectly capable of.
Yes, this is true. However, it is far from trivial. Beside the documentation, there will be issues on integration with the Cocoa API, and with integration with the XCode and Interface Builder (IB) tools. All of these three will be hard, IMHO.
For example, as a former Pascal programmer, I am interested in attempts to add support for gpc to Xcode. However, this hasn't been picked up yet. Issues that I am vaguely aware of are:
I have to say, I'm impressed they found the cause. That's a very neat piece of investigation, I'd say. Great job!
I would recommend anyone to view these type of images on a tile display. At SC 2003, at least EVL and SGI did show dome impressive demo's (in particular, SGI did show some interesting geographic imagery software).
In fact, they did very much push to get this topic on the agenda, and even sent a mail last week asking all memebers to fill in a survey on this matter.
From the newsletter of ISOC:
Web-confidential might also be considered prior art (I'm not a lawyer though). It originates from 1998.
Reading the reports, I'd say Apple should fix #1 and #3. I hope they will. But this is not quite anything urgent: the first is an apparent bug if you turn on core files. Well, they are off by default. The second bugtraq has two parts: the first half should be fixed by Apple (changing permissions of dirs when copying between disk images); the second half is not Apple's fault. The argv[] buffer overflow is a stability issue, not a security issue IMHO, And since it so extremely rare, I don't consider it a bug deal.
/Applications -type d -perm -002 -print /Applications -type d -exec chmod o-w {} \; but this may break apps) /usr/sbin/diskutil verifyPermissions $diskname /Library/Receipts). To repair, type: /usr/sbin/diskutil repairPermissions /
I think the permission thing is the biggest deal. If you are concerned, you can run these commands:
% find
(the article mentions an autofix with: find
%
(where diskname might be / )
This command (available in MacOS x.2 and higher), will compare the permissions to that of the original installer (as stored in
%
I got one of those "air-space" developed pens (you can buy them as a gift). The trick is that the ink is pressurized. Well, it's soo-soo. It did work on wet bathroom tiles, but just any pen it stoped when it was half-full.
No, I just prefer the solution the Russians are said to use: no fancy pressured pens. They just used plain crayons...those work great.
It seems that Apples offers student discounts in most european countries.
For example, there are Dutch, German and Swedish educational stores.
Indeed, there does not seem to be a Greec education store; in fact, there does not seem to be an Greec Apple Store at all (See list of International stores.
For US-based addresses it's either MapQuest or Yahoo maps. I think they are very simular.
For Dutch maps (where I live), I use locatienet or Andes. The first one being slightly better.
There are way too many options nowadays. See Oddens for a collection of links, including to historic maps (not useful if you just want to find an address, only for the curious of heart).
Well, I certainly hope these MO drives are more reliable then the older ones. About five years ago, I believed the claim that MO was more reliable then plain hard disks. This was in the time that a 700 MB harddisk did cost me $350. So I bought a Maxtor MO driver for over $1000. It did work great for a year, and I expected my data was more safe then on a HD, because these things where in caddies and reading was done optical, instead of magentical. I was wrong, after about a year, the disks got bad sectors without warning. About the same happened half a year to my dad who also got a MO drive. It was plainly the worst hardware I ever bought. Period.
Nowadays, I use a hard disk for daily backups (just a second 120 GB disk), and DVD's for monthly backup, which I just put in a box in an other building.
So far, I never had problems. When I expect one of my drives is starting to fail, I just replace it. A 120 GB disk is so cheap nowadays. And Disks are so lovely fast (and it can be automated, since there is no tape to be mounted manually). For corporate backups, I would probably just use 2 DVD's, and store them in seperate locations in well conditioned rooms.
It might be bad faith, but nowadays I just wouldn't trust an MO anymore. Sorry if that does wrong on the improved reliability.
I stand corrected.
;-)
I was reasoning that making a back-to-back connection between computers would never give you the LSR award: you would loose on distance.
Of course, this does not apply to throwing a book: That would be the correct unit (bit-meter/second). However, you'll have to throw REALLY fast. Take the example of frisbeeing a DVD across the room. Let's a asume the DVD takes 5 Gigabyte (about 40 Gigabit for sake of simplicity), and it is thrown at 10 meter/second (dunno, I estimate it's between 10-20 m/s). Then your data movement speed would be 400,000 megabit-meter/second.
That would put you off by a factor 100,000 of the current record (7000 km * 5.5 Gbit/s = 38,500,000,000 megabit-meter/second, provided I didn't make any errors in here).
Either you need to throw a lot faster, or throw more books/DVDs. That would be the multi-stream variant of the awared
To win a Landspeed Record (that is what we're talking about), you do not to get just a high bandwidth, but you got to have a high bandwidth across a big distance.
;-) ).
That's why the parameter you need to blow is "megabit-meters/second", not just megabit/second. At least that's what it says on the award I'm just holding in my hand (one of the older awards which was already shattered to pieces previous year
The slashdot article suggest that a standard was broken. It is not, and the editorial does not mention it. For the record, the Internet Architecture Board wrote:
Last, contrary to what VeriSign claims, it DOES break (parts) of the Internet. See all examples mentioned in the IAB advisory above. For example, consider what happens if you type "myprinter.mydoman.com" instead of "myprinter.mydomain.com" in your CUPS configuration. Without the wildcard, you get a simple message that you made a typo. With the wildcard, you have a hell of a job finding out why you can't connect to your printer.
Though I applaud ICANN this time for forcing VeriSign to it's knees, I'm surprised that they do not tacle all top-level wildcards: Though the .com and .net wildcards have much more impact, there were and still are wildcards at: .cc, .cx, .museum, .nu, .ph, .td, .tk, and .ws.
Why are these not removed? This seems like unfair ruling by ICANN.
I don't expect that a single lawsuit will change EMI's policy.
However, if a CD is judged "defective" if it does not adhere to standards, I wonder if the same logic can be applied to software. I've got plenty of software that works for 90%, but does not completely follow the standard. Can I go to court and get a refund?