The Pentium ran stuff compiled for the 386 rather slowly, and you could get significant performance improvements by using stuff targetting the 586.
The Pentium II did a much, much better job of running 386 code quickly, which is why folks mostly stopped worrying about doing processor-specific builds.
There are certain packages for which the processor type makes a significant difference. Red Hat builds multiple versions of those few -- the kernel, glibc, etc.
Furthermore, as others have pointed out, Red Hat already aligns the code for i686 by default. This is where most of the improvements come from.
If you're really curious, I have tried rebuilding most of Red Hat for a couple versions for the i686, just for the hell of it. One gets no noticeable improvements. It's really a waste of time to spend time building differently. There are a couple programs that use arch-specific assembly (SDL, for instance), but in general, very few software packages are faster when built specifically for a given processor.
You do realize that Fedora is *exactly* the same system that the regular Red Hat releases have been, ever since RHEL came out, and that the RH marketing people are simply trying to play off image ("Oh, you can use this *enterprise*-class Linux distro, or some thing that only techies that like trying out new stuff use"). If you can handle Red Hat 9, you can definitely deal with Fedora Core.
On the other hand unlike the final release of Fedora Core 2, this is a test release, and *is* intended for beta testing. If you don't want to beta test, don't install it.:-)
Note that, whether it's an intended upgrade path or not, I've had luck since Red Hat 5.2 (and back then you were definitely supported to choose "Upgrade" from a CD-ROM) doing upgrades piecemeal by simply upgrading RPMs. Used to cause various breakages, but RH's pretty good at avoiding these these days. You can do the same thing, but much easier, by just making yum or apt point at the new Fedora Core 2 repository, and running an update and installing whatever important stuff has been added. Very cool, and a good way to stress-test dependencies that RH has added.;-)
Why would 60Hz signals accelerate tumor growth without causing them? It just doesn't make sense...
Re:This Kind of Thing Keep Happening...
on
Buddylinks Stinks
·
· Score: 1
I'm dubious as to whether this sort of approach would block the buddylinks worm -- the usage is not that excessive or unusual.
If you have application-level analysis at your firewall, you're talking about more money and maintenance and issues that will crop up. And you want to maintain fairly loose bounds on "legitimate" activity, or else you get false positives for evil activity.
Advances in FW tech have surely made such a thing possible. I doubt that the average home user needs much more than HTTP and SMTP, and certainly doesn't need inbound access!
Ugh. I think we're going to have to remain at an impasse over this. I strongly oppose this point of view. I see people that put in port-blocking firewalls "just because" as being a significant issue in blocking the development of new software on the Internet. (The same goes for NAT and UDP issues -- there's a lot of software that works fine in the absence of NAT that NAT has shredded.
There is only trademark infringement if the usage is likely to cause confusion. Generally, the usage needs to be within the same industry or product category. It is unlikely that people will confuse SiteFinder with your site. Simply having a database that takes in a trademarked keyword and returns results related to that keyword is legitimate -- even if some of the result refer to competitors.
That was brilliantly written -- the prose and the logic were good.
Of course, since I'm responding, I don't think that the logic is perfect.:-) Not all politicians are purely greed-motivated. You just need to get someone in place that is willing to do something that may make a bit less money for them but that is a Good Thing. Sure, maybe just about every politician makes a money-driven decision every now and then, but I suspect that for most politicians, not every decision is money-driven.
This is how a lot of good things happen. You just need to make the most of those non-money-influenced decisions.
I'd like to see a couple of governmental technology advisory boards that consist *entirely* of PhDs from universities -- people that are *not* ex-CEOs and are less likely to have old business buddies that they're willing to do favors for.
Almost all of the non-ISO country-code TLDs are country-specific. They are simply legacy from when the United States was the only player involved..com is US commercial,.mil US military,.gov US government, and.net US network provider. (AFAIK, the only non-US non-ISO TLDs are the recent additions, like.int and.info). The TLDs have been abused (thanks in no small part to money-hungry registrars like Verisign, which encouraged you to abuse the TLD system -- "add foo.org and foo.net to your foo.com registration with one click!"
Your proposed system is a good idea -- it's already present.:-)
# ARIN WHOIS database, last updated 2004-02-11 19:15 # Enter ? for additional hints on searching ARIN's WHOIS database.
Googling for clickspring llc turns up a number of hits. Apparently, ClickSpring has been in the business of writing advertising worms and trojans commercially for some time now. They are responsible for PurityScan as well as some other nasties out there.
Normally I wouldn't care -- another Windows virus -- but now I'm getting masses of useless messages from infected friends.
Obviously, nobody has bothered to charge ClickSpring with computer crime charges, which is quite frusterating.
Unlikely. The whois information for wgutv.com refers to a register.com administrator...unless this was intended to be a joke and just went over my head.:-(
Also, I went through a ton of shells back in the day to find one to settle on -- doesn't rc lack job control? I consider that a pretty mandatory feature...
Speeding is prosecuted. I should know, I just got back from traffic school. Why should this be any different? Sure, nobody is going to get killed pirating MP3s, but if that's your justification then do we get rid of all non-violent crime?
I'm not advocating eliminating either speeding or copyright infringement enforcement. What I would find disturbing is this particular approach to enforcement -- setting up shell companies that simply log all data and potential copyright infringement users engage in. This *would* be nearly equivalent (from an infringement enforcement standpoint) to installing a keylogger/screen dumper. It would be much like hiding (without telling the consumer) speed-tracking devices in cars. The reason I drew the parallel to speeding is that most people speed at times and don't want to get caught for it, and would find such an enforcement tactic undesireable -- even though speeding can cause far nastier problems (and kills many people each year). I was guessing that the original poster probably occasionally speeds, and wanted him to consider the position of the person who might get caught by this.
We don't need to play brain-dead political games with these losers. It's our internet, not theirs. We have the right to totally ignore any and all of ICANN's setup and use our own DNS servers without notice and without asking for their permission.
While a shift is not as trivial as you make it, I do agree with on major point.
Paul Vixie has been running around trying to ensure that nobody acts "immaturely" or engages in name-calling with Verisign. He's desperate to be taken seriously.
That's ridiculous. Verisign, not the engineers criticising them, is the side lacking respectability. The engineers run and design the networks and control the systems that Verisign uses. Verisign is a comparatively tiny collection of a few people who have buddies in politics, scientists, and engineers.
Nobody should feel constrained in their online conversation for fear of "sounding respectable". The engineers who run the networks need prove nothing. They are running things. The only organization that has to worry about image at all is Verisign, which must seem at least impartial and benevolent enough to keep ICANN from axing their monopoly, which could be done.
Verisign was granted a special, unique opportunity to get money for doing almost no work (some bandwidth and adding an entry to a database). Yes, they *can* be expected not to play hardball, as would be accepted in a general business arena, as they are not operating as a regular business. They have a monopoly that was granted to them that they do very well off of. If they want to continuously test their limits and see how much additional money they can soak people for, ICANN and other engineers are under no requirement to keep granting Verisign the right to continue making vast amounts of money for almost no effort.
Verisign has clearly indicated that it is not currently willing to operate a public trust in good faith. They have continued to spout what most engineers consider to be bullshit, and have ignored frusterated feedback. Unfortunately, we have only one remedy, aside from formal complaints from ICANN (which have already been tried), and that is threats against and ultimately termination of Verisign's special privileges. Doing so will mean work for a lot of systems around the world, temporary service interruptions, bad blood at Verisign (and with political buddies of Verisign) and the risk that nobody else will be willing to step up after Verisign (given that their role might be terminated). Verisign is gambling that the Internet's collection of network engineers do not have the balls to actually terminate their role with a certain amount of bad behavior on their part. I am increasingly wanting to see Verisign's gamble proven wrong.
Shifting to OpenNIC or similar has its own set of problems -- can the same level of service be provided? What happens when an name schisms start appearing?
However, it may be better to be safe than sorry. Every day, Verisign makes it harder and harder to extricate them from a position where they can feed on vast amounts of technology money. This is acceptable, as long as they operate in good faith, which they have not done. Verisign's management has tried deceptive renewal forms sent to Verisign competitors. They have tried mucking about with fundamental components of the Internet. They may not be at a point where they must immediately be replaced, but I think that they are at a point where they must be made to modify their behavor or be terminated.
Gore did do a big deal to publicize the Internet. He pushed federal funds to ensure that as many people had Internet access as soon as possible. He didn't create the Internet, and I don't think that he was involved in any early funding, but he can take a reasonable share of credit for how quickly the Internet took off in the home and school. This was a big success for him, and since it was a significant point that he campaigned on, he wanted to crow about it. He did a poor job when he credited himself, but still, he certainly did do good things for the Internet.
Flash has been the cause of numerous security issues over the years. If security is a concern for you, enough to switch browsers, you should seriously consider not leaving Flash enabled.
The worst thing about having to use IE occasionally for me is the long-term bug that when the thing is rendering, it completely screws up window dragging. If you're waiting for a page to finish downloading, and are dragging a window while data is coming down (which I frequently do), your drag will be "cancelled". The window will simply appear in its original location, and the little gray rectangle that you've been dragging will be gone.
My boss complained bitterly about this when he was first moved to a version of IE that ran like this.
Frankly, *I* attribute text overlapping to the people who ever decided that it would be a good idea to try and make HTML a halfassed layout language instead of a markup language. I've been seeing more and more overlapping text and images since CSS was introduced, because browsers *let webpages do so* now.
...if you're a clever perlmaster, and can come up with a short way to synthesize usernames other than just using the entries in the system wordlist, feel free to post it.
Chew on their database
on
Buddylinks Stinks
·
· Score: 2, Interesting
You can stress-test their system by running the following script:
This will start removing everyone in their database, and will also eat cycles on their system.
Re:This Kind of Thing Keep Happening...
on
Buddylinks Stinks
·
· Score: 2, Interesting
This has nothing to do with firewalls. All traffic is going through legitimate programs -- AIM/IE. As a matter of fact, firewalls can make these problems worse, since legitimate people try to tunnel more crap through things like IE requests to avoid having their program set of alarms, etc.
Personal firewalls are, frankly, the worst thing to hit the Net sinc AOL.
It *would* be interesting to sandbox programs that can use the Internet to some degree. This cannot be done on Windows anytime soon (thanks, IE), but could be considered on other platforms.
Heh -- you'd have a browser linked against KDE (which probably means QT as well) libraries, and against the GTK2 suite. *That* would use some memory and have some startup time.:-)
Slashdot Mirror
Oh, the Red Hat engineers aren't stupid.
The Pentium ran stuff compiled for the 386 rather slowly, and you could get significant performance improvements by using stuff targetting the 586.
The Pentium II did a much, much better job of running 386 code quickly, which is why folks mostly stopped worrying about doing processor-specific builds.
There are certain packages for which the processor type makes a significant difference. Red Hat builds multiple versions of those few -- the kernel, glibc, etc.
Furthermore, as others have pointed out, Red Hat already aligns the code for i686 by default. This is where most of the improvements come from.
If you're really curious, I have tried rebuilding most of Red Hat for a couple versions for the i686, just for the hell of it. One gets no noticeable improvements. It's really a waste of time to spend time building differently. There are a couple programs that use arch-specific assembly (SDL, for instance), but in general, very few software packages are faster when built specifically for a given processor.
Gack -- the general Linux X10 driver -- wish -- *only* works on 2.4.
Sigh.
You do realize that Fedora is *exactly* the same system that the regular Red Hat releases have been, ever since RHEL came out, and that the RH marketing people are simply trying to play off image ("Oh, you can use this *enterprise*-class Linux distro, or some thing that only techies that like trying out new stuff use"). If you can handle Red Hat 9, you can definitely deal with Fedora Core.
:-)
On the other hand unlike the final release of Fedora Core 2, this is a test release, and *is* intended for beta testing. If you don't want to beta test, don't install it.
Note that, whether it's an intended upgrade path or not, I've had luck since Red Hat 5.2 (and back then you were definitely supported to choose "Upgrade" from a CD-ROM) doing upgrades piecemeal by simply upgrading RPMs. Used to cause various breakages, but RH's pretty good at avoiding these these days. You can do the same thing, but much easier, by just making yum or apt point at the new Fedora Core 2 repository, and running an update and installing whatever important stuff has been added. Very cool, and a good way to stress-test dependencies that RH has added. ;-)
Why would 60Hz signals accelerate tumor growth without causing them? It just doesn't make sense...
I'm dubious as to whether this sort of approach would block the buddylinks worm -- the usage is not that excessive or unusual.
If you have application-level analysis at your firewall, you're talking about more money and maintenance and issues that will crop up. And you want to maintain fairly loose bounds on "legitimate" activity, or else you get false positives for evil activity.
Advances in FW tech have surely made such a thing possible. I doubt that the average home user needs much more than HTTP and SMTP, and certainly doesn't need inbound access!
Ugh. I think we're going to have to remain at an impasse over this. I strongly oppose this point of view. I see people that put in port-blocking firewalls "just because" as being a significant issue in blocking the development of new software on the Internet. (The same goes for NAT and UDP issues -- there's a lot of software that works fine in the absence of NAT that NAT has shredded.
Oh, well.
There is only trademark infringement if the usage is likely to cause confusion. Generally, the usage needs to be within the same industry or product category. It is unlikely that people will confuse SiteFinder with your site. Simply having a database that takes in a trademarked keyword and returns results related to that keyword is legitimate -- even if some of the result refer to competitors.
That was brilliantly written -- the prose and the logic were good.
:-) Not all politicians are purely greed-motivated. You just need to get someone in place that is willing to do something that may make a bit less money for them but that is a Good Thing. Sure, maybe just about every politician makes a money-driven decision every now and then, but I suspect that for most politicians, not every decision is money-driven.
Of course, since I'm responding, I don't think that the logic is perfect.
This is how a lot of good things happen. You just need to make the most of those non-money-influenced decisions.
I'd like to see a couple of governmental technology advisory boards that consist *entirely* of PhDs from universities -- people that are *not* ex-CEOs and are less likely to have old business buddies that they're willing to do favors for.
Almost all of the non-ISO country-code TLDs are country-specific. They are simply legacy from when the United States was the only player involved. .com is US commercial, .mil US military, .gov US government, and .net US network provider. (AFAIK, the only non-US non-ISO TLDs are the recent additions, like .int and .info). The TLDs have been abused (thanks in no small part to money-hungry registrars like Verisign, which encouraged you to abuse the TLD system -- "add foo.org and foo.net to your foo.com registration with one click!"
:-)
Your proposed system is a good idea -- it's already present.
Let's take a brief look at these folks:
$ host buddylinks.net
buddylinks.net has address 63.251.131.235
$ whois 63.251.131.235
[Querying whois.arin.net]
[whois.arin.net]
Internap Network Services NETBLK-PNAP-11-99 (NET-63-251-0-0-1)
63.251.0.0 - 63.251.255.255
ClickSpring LLC INAP-BSN-CLICKSPRING-0041 (NET-63-251-131-232-1)
63.251.131.232 - 63.251.131.239
# ARIN WHOIS database, last updated 2004-02-11 19:15
# Enter ? for additional hints on searching ARIN's WHOIS database.
Googling for clickspring llc turns up a number of hits. Apparently, ClickSpring has been in the business of writing advertising worms and trojans commercially for some time now. They are responsible for PurityScan as well as some other nasties out there.
Normally I wouldn't care -- another Windows virus -- but now I'm getting masses of useless messages from infected friends.
Obviously, nobody has bothered to charge ClickSpring with computer crime charges, which is quite frusterating.
Unlikely. The whois information for wgutv.com refers to a register.com administrator...unless this was intended to be a joke and just went over my head. :-(
Also, I went through a ton of shells back in the day to find one to settle on -- doesn't rc lack job control? I consider that a pretty mandatory feature...
Because bash/zsh are everywhere, and rc is as scarce as hen's teeth -- and I have to use other systems than my own.
/usr/share/dict/words`; do curl http://www.buddylinks.net/support.php?sn=$w >/dev/null; done
It's just as easy in sh:
for w in `cat
Speeding is prosecuted. I should know, I just got back from traffic school. Why should this be any different? Sure, nobody is going to get killed pirating MP3s, but if that's your justification then do we get rid of all non-violent crime?
I'm not advocating eliminating either speeding or copyright infringement enforcement. What I would find disturbing is this particular approach to enforcement -- setting up shell companies that simply log all data and potential copyright infringement users engage in. This *would* be nearly equivalent (from an infringement enforcement standpoint) to installing a keylogger/screen dumper. It would be much like hiding (without telling the consumer) speed-tracking devices in cars. The reason I drew the parallel to speeding is that most people speed at times and don't want to get caught for it, and would find such an enforcement tactic undesireable -- even though speeding can cause far nastier problems (and kills many people each year). I was guessing that the original poster probably occasionally speeds, and wanted him to consider the position of the person who might get caught by this.
I'd also be interested in the opinion of NANOG and of certain USENET groups. They'd probably have at *least* as much weight as Slashdot.
We don't need to play brain-dead political games with these losers. It's our internet, not theirs. We have the right to totally ignore any and all of ICANN's setup and use our own DNS servers without notice and without asking for their permission.
While a shift is not as trivial as you make it, I do agree with on major point.
Paul Vixie has been running around trying to ensure that nobody acts "immaturely" or engages in name-calling with Verisign. He's desperate to be taken seriously.
That's ridiculous. Verisign, not the engineers criticising them, is the side lacking respectability. The engineers run and design the networks and control the systems that Verisign uses. Verisign is a comparatively tiny collection of a few people who have buddies in politics, scientists, and engineers.
Nobody should feel constrained in their online conversation for fear of "sounding respectable". The engineers who run the networks need prove nothing. They are running things. The only organization that has to worry about image at all is Verisign, which must seem at least impartial and benevolent enough to keep ICANN from axing their monopoly, which could be done.
Verisign was granted a special, unique opportunity to get money for doing almost no work (some bandwidth and adding an entry to a database). Yes, they *can* be expected not to play hardball, as would be accepted in a general business arena, as they are not operating as a regular business. They have a monopoly that was granted to them that they do very well off of. If they want to continuously test their limits and see how much additional money they can soak people for, ICANN and other engineers are under no requirement to keep granting Verisign the right to continue making vast amounts of money for almost no effort.
Verisign has clearly indicated that it is not currently willing to operate a public trust in good faith. They have continued to spout what most engineers consider to be bullshit, and have ignored frusterated feedback. Unfortunately, we have only one remedy, aside from formal complaints from ICANN (which have already been tried), and that is threats against and ultimately termination of Verisign's special privileges. Doing so will mean work for a lot of systems around the world, temporary service interruptions, bad blood at Verisign (and with political buddies of Verisign) and the risk that nobody else will be willing to step up after Verisign (given that their role might be terminated). Verisign is gambling that the Internet's collection of network engineers do not have the balls to actually terminate their role with a certain amount of bad behavior on their part. I am increasingly wanting to see Verisign's gamble proven wrong.
Shifting to OpenNIC or similar has its own set of problems -- can the same level of service be provided? What happens when an name schisms start appearing?
However, it may be better to be safe than sorry. Every day, Verisign makes it harder and harder to extricate them from a position where they can feed on vast amounts of technology money. This is acceptable, as long as they operate in good faith, which they have not done. Verisign's management has tried deceptive renewal forms sent to Verisign competitors. They have tried mucking about with fundamental components of the Internet. They may not be at a point where they must immediately be replaced, but I think that they are at a point where they must be made to modify their behavor or be terminated.
Gore did do a big deal to publicize the Internet. He pushed federal funds to ensure that as many people had Internet access as soon as possible. He didn't create the Internet, and I don't think that he was involved in any early funding, but he can take a reasonable share of credit for how quickly the Internet took off in the home and school. This was a big success for him, and since it was a significant point that he campaigned on, he wanted to crow about it. He did a poor job when he credited himself, but still, he certainly did do good things for the Internet.
Flash has been the cause of numerous security issues over the years. If security is a concern for you, enough to switch browsers, you should seriously consider not leaving Flash enabled.
The worst thing about having to use IE occasionally for me is the long-term bug that when the thing is rendering, it completely screws up window dragging. If you're waiting for a page to finish downloading, and are dragging a window while data is coming down (which I frequently do), your drag will be "cancelled". The window will simply appear in its original location, and the little gray rectangle that you've been dragging will be gone.
My boss complained bitterly about this when he was first moved to a version of IE that ran like this.
Frankly, *I* attribute text overlapping to the people who ever decided that it would be a good idea to try and make HTML a halfassed layout language instead of a markup language. I've been seeing more and more overlapping text and images since CSS was introduced, because browsers *let webpages do so* now.
Good *god*, the first site renders incredibly slowly on Firebird 0.7.
...if you're a clever perlmaster, and can come up with a short way to synthesize usernames other than just using the entries in the system wordlist, feel free to post it.
You can stress-test their system by running the following script:
/usr/share/dict/words| perl -pe 'system("curl http://www.buddylinks.net/support.php?sn=$_");' >/dev/null
cat
This will start removing everyone in their database, and will also eat cycles on their system.
This has nothing to do with firewalls. All traffic is going through legitimate programs -- AIM/IE. As a matter of fact, firewalls can make these problems worse, since legitimate people try to tunnel more crap through things like IE requests to avoid having their program set of alarms, etc.
Personal firewalls are, frankly, the worst thing to hit the Net sinc AOL.
It *would* be interesting to sandbox programs that can use the Internet to some degree. This cannot be done on Windows anytime soon (thanks, IE), but could be considered on other platforms.
Heh -- you'd have a browser linked against KDE (which probably means QT as well) libraries, and against the GTK2 suite. *That* would use some memory and have some startup time. :-)