Now "What next," that's easily seen as stupid by any semi-literate person who knows English.
Actually it's only viewed as stupid by those who have only taken enough English to be able to criticize. Haven't you ever heard of colloquial writing style?
Ha! I live in Canada where not only is it cruel and unusual punishment but we also pay a provincial income tax thats 40% of our federal tax
plus 15% tax on anything we buy.
Typical Ontario-centric bullshit. Provincial income taxes vary by province you dumbass. And the Federal "Goods and Services Tax" is 7%; the provincial tax again varies by province. Ontario is 8% which gives you your 15%. I believe that it's Alberta has no provincial tax at all.
Get your facts straight; Lord knows with people like you, Canadians look just as stupid as our southern neighbours. What next, back to the imperial system of measurement?
Yes it is. And the inability of Windows to do this is exactly what is causing the problem on Windows. DLL Hell is not having a great many libraries. DLL Hell is having multiple DLLs with the same name but different versions. Windows is unable to differentiate between them. And since in-memory DLLs have priority over on-disk DLLs, breakage depends on the order you've started your applications.
Don't forget that Windows DLLs have a nasty habit of lying about their version so that older software will be happy with a newer version. The first time I heard that this existed I nearly shit a brick; it defies all sense and usefulness to having version numbers in the first place!
Even more interesting, sounds in the real world DO have tons of high-frequency content! 40% of the energy produced by a cymbal crash is above 20 KHz.
This is very important -- you may not hear the 26kHz components but they interact with other frequency components and give you sum and difference components that are well within your hearing range. If you clamp 'em off the low stuff won't be reproduced correctly.
And there's relatively little danger of burning out a tube amp, compared to a mosfet amp or something. Unless, of course, your name is Jimi Hendrix.
Actually I believe that open-circuiting the output stages on tube amps led to a very quick and untimely demise. And a lot of the output stages in decent amps these days has a negative temperature coefficient; they start turning off when they get hot.
Anyone who has worked on control systems knows that most run their own proprietary networks and communications protocols.
Maybe that was true 15 years ago. Today everything is either DeviceNet (US), ProfiBus (Europe), or ModBUS (everywhere) -- there are other protocols but these are the Big Three. And with Industrial Ethernet becoming more and more popular, ModBUS/TCP (ModBus data structure inside a normal TCP packet), it is trivial to fuck up network.
True, you'll likely not know what you're dicking with since you won't have the device configuration files (DeviceNet) unless they were left lying around somewhere but just having the ability to spew trash out to all the industrial devices can cause some pretty massive problems all their own.
A friend of mine has made several flashlights out of LEDs, potentiometers (as dimmers), and short lengths of PVC.
Good idea but I would suggest building a small CMOS oscillator or using a 12C509 PIC and Pulse Width Modulate (PWM) the LED instead of burning up the excess energy in a potentiometer. The batter(y|ies) will last much longer.
Until recently (maybe a year ago) the military introduced timing errors into their signal so that it was impossible to get an exact fix although a ground based unit would increase your percision.
Of course if this story is true it's rediculous.
Why's it rediculous? Selective Availability as it was called buggered with the timing of the satellites. Without a military unit you just couldn't get a perfectly accurate fix on your location. Here is some more info. Basically before you couldn't get a within more than about 100 meters with more than 95% confidence. That's been scrapped as of 1 May 2000; now you have accuracies of about 20m, and even better if you go differential GPS. I know that surveyors were especially happy to have SA dissappear.
Oh well. You could do all that with emacs and have a real editor
Sure, call up the bloatware app of the open source world. Hell even Moz doesn't meet the bloat that emacs has.
And, you could be doing other, real work, instead of farting around pointing and clicking on menus when two or three commands on the keyboard would have the job done.
I am generally a keyboard kind of guy. Like the keyboard, however, mice have their place. In a multi-pane email app, the mouse wins out over the keyboard for quick selection. After that I use the keyboard to scroll up and down, selecting different messages, deleting, etc.
I will never, in all the time I'm on this earth, understand why people obsess with mousing around on menus. It is demonstrably NOT faster than using the keyboard to do the same job.
For most circumstances I would agree with you on the speed issue. Try calming a crying infant in one arm and read the latest CBC news with just a keyboard. The mouse is demonstratably faster in situations such as those. And it has also been demonstratably proven that the WIMP interface is more intuitave than [esc]:wq (Yes I use vi).
Like a lot of other people, I've used multiple clients and I have no doubt that the text-based clients are the fastest and easiest to use. You can't do anything with fatal OE that I can't do as well or better in emacs -- well, except produce HTML mail.
For me, OE is the best. I personally don't care what you use unless I haven't heard of it before and thus perhaps persuade me to try it. As far as emacs goes -- well I'm not going to open up that can of worms. I don't like it and that's all there is to say on that particular subject. It works for you and that's good; I'm happy you're happy. It won't do it for me, though.
And since when did this become a pissing match as to what the OS-that-thinks-its-an-editor and an app which has a specific defined purpose can do? I stated what I liked about OE and why because someone (possibly you) had said that OE was a bloated piece of shite. I feel I've proven my point.
Outlook Express & Communicator are good & full-featured???
I prefer Outlook Express to all other email clients I've tried, including Communicator (Netscape or Moz), KMail, Pine, Mutt, Eudora, Mulberry, Sylpheed, Kiltdown and probably another half dozen which I've forgotten the names of.
Why do I prefer Outlook Express? It's ONE app for mail and news. It's straightforward, has pretty damn good filtering (No I don't need regexps, thanks for asking), multiple POP, IMAP and NNTP accounts, works with LDAP, doesn't barf on attachments or HTML mail (ewww...), I've never had it crash out on me and it is pretty damn fast unless there's a 10M attachment. NO OTHER EMAIL APP WORKS AS WELL FOR ME. Get it? Got it? Good.
Believe you me, I want an OE clone for Linux. I run Linux on my laptop but run Win4Lin so I can get OE, IE, Office and my Win32 dev tools. I don't use IE all that much (Opera and Konq rock my world) but there has been nothing which works as well for me as Outlook Express for email and news. And that's sad because I could do a lot of my work without booting Win4Lin if I could only get a decent email client.
Security? Yeah OE blows for security. I run qmail on my mail servers and the HTML-trap procmail script cleanses all my incoming messages. I've never seen an ILOVEYOU, autorun.vbs or Word macro virus. If I were running OE in a "wild" environment I'd be crazy but I have a nice firewall at work and a decent firewall at home. I'm not running "in the wild." OE works nice for me.
They are bloatware and foist HTML
and MS DOC format into emails.
I've never had MS Word or.DOC emails come out of OE. And it certainly doesn't foist HTML email on me. One configuration switch and it's all plain text, baby. As far as bloatware goes, OE is actually pretty nice in that respect.
So I guess "leech" must be a technical term meaning "person who actually attempts to use the unlimited connect time we advertised to them", eh?
Nope; we define the actual service as "unlimited interactive" -- meaning you must be in front of the computer, using it most of the time, not sitting idle.
You won't get booted for downloading all the linux kernel releases (several days' worth of traffic)... you do that every week though and your hours will rack up and you will be booted. Fairly simple.
Hmm, would be nice if all modules worked that way. I had to add some modules for ip_sec to my firewall this weekend and tried doing it without recompiling everything. Sure enough had to patch the kernel also and recompile and stuff.
Kinda sucks on dual 50Mhz Sparc procs.
Sounds like development code to me. If you're gonna run dev code, get a development machine. My firewall is an 80386DX/33 but all my compilation takes place on a dual Cel433 with 256M of memory. It takes the sting out, let me tell you.:-)
Now I've had an ATI Radeon for a month or so, and have yet to experience a single crash. And the Radeon DRI drivers are still alpha code... Go ATI, I say!
Until you realize that they too are holding out... TV-Out, hardware MPEG2 decode... these things aren't available on your beloved ATI cards when used under Linux. Go Trident, I say!:-)
AT and ATX power supplies give you +/-5V and +/-12V. That's where the similarity ends.
ATX power supplies also give a +3.3V an always-on 5V (I think) and have the soft power switch (ground the green wire to toggle on and off). While it's easy to rewire the power switch, that 3.3V is gonna be tricky unless you know what you're doing. We're not talking about throwing on a cheap-ass LM2983 and a couple caps. A 250W ATX supply has a 14A rated 3.3V supply.
It's not just dead time. I know that at the very least my ISP explicitly states that they will disconnect based on connection time, not on usage level, when they feel the need to disconnect people. I'd strongly recommend looking at your TOS to find out.
Yup; the ISP I help admin keeps track of average monthly usage. If we run into an "all lines full" situation for more than 15 minutes we drop the top 2 people currently online. Repeat every 5 minutes ad nauseam. It's plainly stated in our ToS and the people who get on to just check their mail aren't ever in danger of getting dropped during their daily 5 minutes of usage.
It seems to work out quite well for everyone because the leeches usually get pissed off and leave for another ISP, which helps keep our user/line ratio in order and helps us keep costs down (we don't have to subsidize leeches by raising rates). Hell we even have a number of always-on people who don't seem to mind our disconnections because a) they know they are using it as always on and accept the occassional drop and b) we have the most competitive price in the area.
Hmm... it could be because I'm using reiserfs - I'll try a build on my ext2 partition (I've got 5 GB free on there, so it's not prob).
The Dual466 system runs reiserfs on its data drive (actually 5 6.4G IDE drives in software RAID0 under LVM) -- I know that tail packing slows down reiserfs like winter does molasses; I have that (and atime updates) turned off.
n hour? That's nothing - KDE2.2alpha2 takes about 16 hours to build! (no kidding - I started it building at 12 hours ago, and it's about 3/4 done)
What the hell are you doing? KDE2.2 compiles for me in about 2 hours. That includes manually un-tarring,./configuring, and make/make installing. QT is by far the longest compile of the lot for KDE.
This is on a Cel300 with 256M of memory and running X. On my dual 450 it takes a little more than an hour.
Re:Games: XFree86 with DRI, or Linux FBDev?
on
XFree 4.1.0 Out
·
· Score: 2
don't think there's anything wrong with including a video driver in the kernel. Tying the kernel to a desktop is another story. The Linux framebuffer console is how video should have been all along. No need to be root in order to access the display. This is probably good for security as well as stability.
I'm mixed on the issue. Drivers are drivers, this much is true. Just because they are video drivers doesn't mean they don't belong in the kernel. (Hell my nVidia card has to have kernel drivers.)
At the same time, however, I think it hampers the rate of development for newer, faster and better drivers (video or otherwise). Video drivers especially change quickly as newer cards come out. It's not like SCSI or IDE or sound or network interfaces... they all seem fairly "standard" -- Video still seems to be a fast moving target.
Re:Gods you fiends! Here's your changelog...
on
XFree 4.1.0 Out
·
· Score: 1
EEK! Mike Harris is the premier of Ontario, Canada. And also I guy I hate. Now he's working on Free Software?!?! What is the world coming to....
If memory serves there used to be a Mike Harris hanging around either FidoNet or DragoNet in the 519 area code. His.sig was always "No I am NOT the premier of Ontario".
FidoNet... Christ... 1:221/[something].77 was my point.. I forget how the addressing works now. That was YEARS ago.
XFree86 4.0.3 does not include Mach64 render code, so there is no anti-aliasing or anything.
While XFree86 4.0.3 does not have RENDER for Mach64, CVS does. I am running right now on this laptop. Used the antialias howto on dot.kde.org -- everything under 8 and above 12 is antialiased, it is sweet. And since I'm on an LCD subpixel antialiasing makes small text MUCH clearer than standard greylevel antialias.
I would love to know how to turn on antialiasing for any italics regardless of size; I tried a few variations of the configuration outlined on dot.kde.org but nothing seems to work. If anyone has any success here, please let me know.
Maybe you do this, but most of the internet does despite efforts to the contrary -- MFS once threatened to disconnect people who weren't blocking spoofed traffic entering and leaving their network.
I tend to think that far less of the edge routers are doing this than you may think. As you said, filtering becomes expensive as the routing tables become larger and companies are not wanting to spend the money on new equipment. That's why I'm suggesting that only the edges perform this type of filtering. You have a limited number of interfaces and the IPs flowing between them are far smaller than the number of acceptable addresses flying through a core router's interfaces.
Also, I'm not talking about the dialup providers. The amount of bandwidth a single (or even 1000) dialup users can provide doesn't even register on the map of these DDoS attacks. But every DSL and cable provider should be filtering their incoming traffic.
These attacks are verydifficult to defend against, because they simulate actual valid traffic. And you can't block the source address, because it's invalid.
If the address is invalid it should have been blocked by the outgoing routers or the router at your edge. i.e. my firewall won't pass any of the private IP blocks and the routers at my ISP drop packets from IPs in those same private blocks.
Now spoofing routable address is a different story but again if the edge routers simply dropped traffic from IPs it shouldn't see on that particular interface you'd see a hell of a lot less DDoS than you do now.
ou gotta read how he complains about the standart socket implementation of Win2k / XP.
First people complain about the lack of, now because its there, damn give MS a break.
I don't know about you, but my routers simply drop traffic which doesn't come from an IP that isn't from the interface it's coming in on. I don't care what they users run because the edge routers won't allow spoofed traffic... well not unless they spoof another IP within the block(s) on the interface.:-)
Slashdot Mirror
Now "What next," that's easily seen as stupid by any semi-literate person who knows English.
Actually it's only viewed as stupid by those who have only taken enough English to be able to criticize. Haven't you ever heard of colloquial writing style?
Ha! I live in Canada where not only is it cruel and unusual punishment but we also pay a provincial income tax thats 40% of our federal tax
plus 15% tax on anything we buy.
Typical Ontario-centric bullshit. Provincial income taxes vary by province you dumbass. And the Federal "Goods and Services Tax" is 7%; the provincial tax again varies by province. Ontario is 8% which gives you your 15%. I believe that it's Alberta has no provincial tax at all.
Get your facts straight; Lord knows with people like you, Canadians look just as stupid as our southern neighbours. What next, back to the imperial system of measurement?
Yes it is. And the inability of Windows to do this is exactly what is causing the problem on Windows. DLL Hell is not having a great many libraries. DLL Hell is having multiple DLLs with the same name but different versions. Windows is unable to differentiate between them. And since in-memory DLLs have priority over on-disk DLLs, breakage depends on the order you've started your applications.
Don't forget that Windows DLLs have a nasty habit of lying about their version so that older software will be happy with a newer version. The first time I heard that this existed I nearly shit a brick; it defies all sense and usefulness to having version numbers in the first place!
Even more interesting, sounds in the real world DO have tons of high-frequency content! 40% of the energy produced by a cymbal crash is above 20 KHz.
This is very important -- you may not hear the 26kHz components but they interact with other frequency components and give you sum and difference components that are well within your hearing range. If you clamp 'em off the low stuff won't be reproduced correctly.
And there's relatively little danger of burning out a tube amp, compared to a mosfet amp or something. Unless, of course, your name is Jimi Hendrix.
Actually I believe that open-circuiting the output stages on tube amps led to a very quick and untimely demise. And a lot of the output stages in decent amps these days has a negative temperature coefficient; they start turning off when they get hot.
Anyone who has worked on control systems knows that most run their own proprietary networks and communications protocols.
Maybe that was true 15 years ago. Today everything is either DeviceNet (US), ProfiBus (Europe), or ModBUS (everywhere) -- there are other protocols but these are the Big Three. And with Industrial Ethernet becoming more and more popular, ModBUS/TCP (ModBus data structure inside a normal TCP packet), it is trivial to fuck up network.
True, you'll likely not know what you're dicking with since you won't have the device configuration files (DeviceNet) unless they were left lying around somewhere but just having the ability to spew trash out to all the industrial devices can cause some pretty massive problems all their own.
A friend of mine has made several flashlights out of LEDs, potentiometers (as dimmers), and short lengths of PVC.
Good idea but I would suggest building a small CMOS oscillator or using a 12C509 PIC and Pulse Width Modulate (PWM) the LED instead of burning up the excess energy in a potentiometer. The batter(y|ies) will last much longer.
Until recently (maybe a year ago) the military introduced timing errors into their signal so that it was impossible to get an exact fix although a ground based unit would increase your percision. Of course if this story is true it's rediculous.
Why's it rediculous? Selective Availability as it was called buggered with the timing of the satellites. Without a military unit you just couldn't get a perfectly accurate fix on your location. Here is some more info. Basically before you couldn't get a within more than about 100 meters with more than 95% confidence. That's been scrapped as of 1 May 2000; now you have accuracies of about 20m, and even better if you go differential GPS. I know that surveyors were especially happy to have SA dissappear.
or when an actual satellite is downstreaming over my potatoes' garden?
Your potatoes have a garden! Now those are some advanced spuds!
Oh well. You could do all that with emacs and have a real editor
Sure, call up the bloatware app of the open source world. Hell even Moz doesn't meet the bloat that emacs has.
And, you could be doing other, real work, instead of farting around pointing and clicking on menus when two or three commands on the keyboard would have the job done.
I am generally a keyboard kind of guy. Like the keyboard, however, mice have their place. In a multi-pane email app, the mouse wins out over the keyboard for quick selection. After that I use the keyboard to scroll up and down, selecting different messages, deleting, etc.
I will never, in all the time I'm on this earth, understand why people obsess with mousing around on menus. It is demonstrably NOT faster than using the keyboard to do the same job.
For most circumstances I would agree with you on the speed issue. Try calming a crying infant in one arm and read the latest CBC news with just a keyboard. The mouse is demonstratably faster in situations such as those. And it has also been demonstratably proven that the WIMP interface is more intuitave than [esc]:wq (Yes I use vi).
Like a lot of other people, I've used multiple clients and I have no doubt that the text-based clients are the fastest and easiest to use. You can't do anything with fatal OE that I can't do as well or better in emacs -- well, except produce HTML mail.
For me, OE is the best. I personally don't care what you use unless I haven't heard of it before and thus perhaps persuade me to try it. As far as emacs goes -- well I'm not going to open up that can of worms. I don't like it and that's all there is to say on that particular subject. It works for you and that's good; I'm happy you're happy. It won't do it for me, though.
And since when did this become a pissing match as to what the OS-that-thinks-its-an-editor and an app which has a specific defined purpose can do? I stated what I liked about OE and why because someone (possibly you) had said that OE was a bloated piece of shite. I feel I've proven my point.
Outlook Express & Communicator are good & full-featured???
I prefer Outlook Express to all other email clients I've tried, including Communicator (Netscape or Moz), KMail, Pine, Mutt, Eudora, Mulberry, Sylpheed, Kiltdown and probably another half dozen which I've forgotten the names of.
Why do I prefer Outlook Express? It's ONE app for mail and news. It's straightforward, has pretty damn good filtering (No I don't need regexps, thanks for asking), multiple POP, IMAP and NNTP accounts, works with LDAP, doesn't barf on attachments or HTML mail (ewww...), I've never had it crash out on me and it is pretty damn fast unless there's a 10M attachment. NO OTHER EMAIL APP WORKS AS WELL FOR ME. Get it? Got it? Good.
Believe you me, I want an OE clone for Linux. I run Linux on my laptop but run Win4Lin so I can get OE, IE, Office and my Win32 dev tools. I don't use IE all that much (Opera and Konq rock my world) but there has been nothing which works as well for me as Outlook Express for email and news. And that's sad because I could do a lot of my work without booting Win4Lin if I could only get a decent email client.
Security? Yeah OE blows for security. I run qmail on my mail servers and the HTML-trap procmail script cleanses all my incoming messages. I've never seen an ILOVEYOU, autorun .vbs or Word macro virus. If I were running OE in a "wild" environment I'd be crazy but I have a nice firewall at work and a decent firewall at home. I'm not running "in the wild." OE works nice for me.
They are bloatware and foist HTML and MS DOC format into emails.
I've never had MS Word or .DOC emails come out of OE. And it certainly doesn't foist HTML email on me. One configuration switch and it's all plain text, baby. As far as bloatware goes, OE is actually pretty nice in that respect.
So I guess "leech" must be a technical term meaning "person who actually attempts to use the unlimited connect time we advertised to them", eh?
Nope; we define the actual service as "unlimited interactive" -- meaning you must be in front of the computer, using it most of the time, not sitting idle.
You won't get booted for downloading all the linux kernel releases (several days' worth of traffic)... you do that every week though and your hours will rack up and you will be booted. Fairly simple.
Hmm, would be nice if all modules worked that way. I had to add some modules for ip_sec to my firewall this weekend and tried doing it without recompiling everything. Sure enough had to patch the kernel also and recompile and stuff. Kinda sucks on dual 50Mhz Sparc procs.
Sounds like development code to me. If you're gonna run dev code, get a development machine. My firewall is an 80386DX/33 but all my compilation takes place on a dual Cel433 with 256M of memory. It takes the sting out, let me tell you. :-)
Now I've had an ATI Radeon for a month or so, and have yet to experience a single crash. And the Radeon DRI drivers are still alpha code... Go ATI, I say!
Until you realize that they too are holding out... TV-Out, hardware MPEG2 decode... these things aren't available on your beloved ATI cards when used under Linux. Go Trident, I say! :-)
AT and ATX power supplies give you +/-5V and +/-12V. That's where the similarity ends.
ATX power supplies also give a +3.3V an always-on 5V (I think) and have the soft power switch (ground the green wire to toggle on and off). While it's easy to rewire the power switch, that 3.3V is gonna be tricky unless you know what you're doing. We're not talking about throwing on a cheap-ass LM2983 and a couple caps. A 250W ATX supply has a 14A rated 3.3V supply.
Nice idea but not really worth trying.
It's not just dead time. I know that at the very least my ISP explicitly states that they will disconnect based on connection time, not on usage level, when they feel the need to disconnect people. I'd strongly recommend looking at your TOS to find out.
Yup; the ISP I help admin keeps track of average monthly usage. If we run into an "all lines full" situation for more than 15 minutes we drop the top 2 people currently online. Repeat every 5 minutes ad nauseam. It's plainly stated in our ToS and the people who get on to just check their mail aren't ever in danger of getting dropped during their daily 5 minutes of usage.
It seems to work out quite well for everyone because the leeches usually get pissed off and leave for another ISP, which helps keep our user/line ratio in order and helps us keep costs down (we don't have to subsidize leeches by raising rates). Hell we even have a number of always-on people who don't seem to mind our disconnections because a) they know they are using it as always on and accept the occassional drop and b) we have the most competitive price in the area.
Hmm... it could be because I'm using reiserfs - I'll try a build on my ext2 partition (I've got 5 GB free on there, so it's not prob).
The Dual466 system runs reiserfs on its data drive (actually 5 6.4G IDE drives in software RAID0 under LVM) -- I know that tail packing slows down reiserfs like winter does molasses; I have that (and atime updates) turned off.
I'm talking about ALL of KDE2.2, but just libs and base.
I understand that. How about this list?
(yes this is the 2.1 tree, and koffice wasn't included, that is a pig of a package)
n hour? That's nothing - KDE2.2alpha2 takes about 16 hours to build! (no kidding - I started it building at 12 hours ago, and it's about 3/4 done)
What the hell are you doing? KDE2.2 compiles for me in about 2 hours. That includes manually un-tarring, ./configuring, and make/make installing. QT is by far the longest compile of the lot for KDE.
This is on a Cel300 with 256M of memory and running X. On my dual 450 it takes a little more than an hour.
don't think there's anything wrong with including a video driver in the kernel. Tying the kernel to a desktop is another story. The Linux framebuffer console is how video should have been all along. No need to be root in order to access the display. This is probably good for security as well as stability.
I'm mixed on the issue. Drivers are drivers, this much is true. Just because they are video drivers doesn't mean they don't belong in the kernel. (Hell my nVidia card has to have kernel drivers.)
At the same time, however, I think it hampers the rate of development for newer, faster and better drivers (video or otherwise). Video drivers especially change quickly as newer cards come out. It's not like SCSI or IDE or sound or network interfaces... they all seem fairly "standard" -- Video still seems to be a fast moving target.
EEK! Mike Harris is the premier of Ontario, Canada. And also I guy I hate. Now he's working on Free Software?!?! What is the world coming to....
If memory serves there used to be a Mike Harris hanging around either FidoNet or DragoNet in the 519 area code. His .sig was always "No I am NOT the premier of Ontario".
FidoNet... Christ... 1:221/[something].77 was my point.. I forget how the addressing works now. That was YEARS ago.
XFree86 4.0.3 does not include Mach64 render code, so there is no anti-aliasing or anything.
While XFree86 4.0.3 does not have RENDER for Mach64, CVS does. I am running right now on this laptop. Used the antialias howto on dot.kde.org -- everything under 8 and above 12 is antialiased, it is sweet. And since I'm on an LCD subpixel antialiasing makes small text MUCH clearer than standard greylevel antialias.
I would love to know how to turn on antialiasing for any italics regardless of size; I tried a few variations of the configuration outlined on dot.kde.org but nothing seems to work. If anyone has any success here, please let me know.
Maybe you do this, but most of the internet does despite efforts to the contrary -- MFS once threatened to disconnect people who weren't blocking spoofed traffic entering and leaving their network.
I tend to think that far less of the edge routers are doing this than you may think. As you said, filtering becomes expensive as the routing tables become larger and companies are not wanting to spend the money on new equipment. That's why I'm suggesting that only the edges perform this type of filtering. You have a limited number of interfaces and the IPs flowing between them are far smaller than the number of acceptable addresses flying through a core router's interfaces.
Also, I'm not talking about the dialup providers. The amount of bandwidth a single (or even 1000) dialup users can provide doesn't even register on the map of these DDoS attacks. But every DSL and cable provider should be filtering their incoming traffic.
These attacks are verydifficult to defend against, because they simulate actual valid traffic. And you can't block the source address, because it's invalid.
If the address is invalid it should have been blocked by the outgoing routers or the router at your edge. i.e. my firewall won't pass any of the private IP blocks and the routers at my ISP drop packets from IPs in those same private blocks.
Now spoofing routable address is a different story but again if the edge routers simply dropped traffic from IPs it shouldn't see on that particular interface you'd see a hell of a lot less DDoS than you do now.
ou gotta read how he complains about the standart socket implementation of Win2k / XP. First people complain about the lack of, now because its there, damn give MS a break.
I don't know about you, but my routers simply drop traffic which doesn't come from an IP that isn't from the interface it's coming in on. I don't care what they users run because the edge routers won't allow spoofed traffic... well not unless they spoof another IP within the block(s) on the interface. :-)