To my horror, once I finally got the thing installed I learned that it doesn't even come with a compiler. Sure you can add GCC to it,
Right, and indeed I think gcc might even be on the companion 'sfw' CD, if you get full Solaris 10 media. I'm not sure, I've never installed Solaris 10 from CD media;) Dont forget, that if you install a Solaris 9 packaged gcc, you'll need to run the not-really-aptly-named "fix-headers" (or somesuch) command. Gcc likes to munge system headers and sneakily use the munged copies instead, and you'll hit problems if your gcc is using S9 munged headers and you try compile on S10.
there must be some art to making GNU's tools work properly with Sun's libc that is beyond me. The biggest problem I had was libtool seems to be completely broken with respect to shared libraries on Solaris.
What problems did you have? I regularly use libtool on Solaris, and have not had problems. Libtool, despite its complexity (as with all the auto* GNU build system tools) when things go wrong, does a good job 99% of the time.
one problem with scaling the opteron up to such large systems is the physical address space in current implementations. It's 40 bits.
I'm sure AMD could be persuaded to build opteron's with more phys address space if demand were there. Though, I have no idea how ingrained the 40 bit assumption is in the AMD64 MMU structures. But at worst, with a CPU feature bit, it could be enabled selectively, if it is not completely configurable in setup of the page directory already. AMD would have been idiots not to allow for expansion of the phys address space.;)
The other thing the hub chip does on SGI machines is to handle the page directory stuff, counting remote vs local accesses, and handling automatically migrating pages to nodes that are making heavy use of them. Most of this cool stuff is patented, so I doubt the Opteron is doing it.
AIUI, its left to software on Opteron. So it's all down to your OS. Linux has some support for software balanced ccNUMA (mostly process CPU affinity, and allocating pages on the affine CPU, if possible - dont think it can migrate pages between nodes yet, but i could be wrong).
Ironically Opteron would be a less good fit than, say, Xeon, for the engineering that SGI have done for MIPS and Itanium, exactly because it has native support for ccNUMA and has integral memory controllers.
Newisys have a chipset that extends the CC and addressing of Opterons so that you can put upto 32 in a system.
Well, I didnt mean a drop-in fit to Origin, but more building a large ccNUMA machine with Opteron. That someone has already done with Opteron what it took SGI many many years to engineer with Origin (and reeingeer for Itanic) demonstrates my point that Opteron would be less effort, no?;)
BTW, as for reimplementation, check out what DEC did with the Alpha Wildfire machines (the big GS series machines), indeed with the 21364 DEC^WCompaq actually integrated routing protocols into the CPUs. The opteron ccNUMA stuff derives more from DECs approach than SGIs (namely CPU as actively aware participant in a ccNUMA topology, compared to SGI and others approach to Itanic and Xeon of building the smarts around 'dumb' CPUs)
SGI is being very successfull with it's 512 itanium machines running Linux.
Note that SGI are doing this with very very special hardware. IIRC each CPU brick in an Origin has 4 itanics. All these bricks are then interconnected with very very special CPU interconnect routers.
That these machines go to 512 CPUs has *nothing* to do with the CPUs being Itanic, it's all down to the ccNUMA interconnect technology (which SGI initially acquired from Cray). If you need further convincing of this, note that the Origin 3k architecture SGIs machines have essentially the same architecture, but use MIPS CPUs. This architecture could be applied to Opteron too, and probably with less effort, as Opteron natively supports ccNUMA and comes with CPU networking built-in.
Pentium Pro, another hyped up CPU that never really delivered.
Hang on, you are joking, right?
PPro has probably been Intel's best chip architecture to date. The initial P6 had bad 16bit performance, which made it a bad choice for consumers are that time, but it was very competitive in normal 32bit mode, idea for NT, Linux and other PC Unixen. The 2nd iteration of the P6 architecture fixed the 16bit issue and was enormously successful. The latest iteration of that arch (Pentium-M) is quietly outperforming the architecture designed to replace it, the P4, at nearly half the clock speed and far less power usage. Indeed, it looks like Intel will be going *back* to the P6 family in future as its 'frontline' PC architecture.
We're talking around each other now, saying the exact same thing essentially.
To specify nominal value ± tolerance (or even with seperate + and - tolerance, as in UK) is same thing as specifying a range. In the EU, it's specified as nominal/tolerance - maybe in USA it's specified as a range, but it's two ways of expressing same values ultimately!
Urr, yes of course it is standardised - dont be silly. Note that standardised figures can include a specified tolerance hence covering a range. The nominal EU AC domestic supply voltage is 230V ± whatever V of tolerance @ 50Hz. Probably about ± 5% of nominal, as that covers everything from 220 to 240V. However, note there is a nominal single standardised voltage.
In reality, supply voltage is dependent on load, the power company actually has to monitor and change the supply as the load changes (eg from day to night time when people use power differently), to maintain acceptable supply voltage levels (and efficient power transfer). Also, the power companies tend to monitor and modulate supply based at a far more macroscopic level of many many thousands of loads aggregated. The local view can be very different (eg, you might have a DIY crazy neighbour who regularly uses heavy duty electric drills, or you might have a geek neighbour who runs a whole bunch of computers 24x7 and draws a kW or more of continious resistive power from your local loop).
I thought the UK ran at 240 volts, not 250. I thought South Africa was the only country that ran 250. I though most of Europe was 220 and Europe and the UK were going to unify their standards at 230.
I thought UK was 240V too. I used to live there about 15 years ago and distinctly remember the figure of 240 mentioned in everything from physics classes to electrical safety literature. I now live (and have even before living in UK) in Ireland which for historical reasons long had identical consumer-visible standards/arrangements for electricity (probably deeper than just consumer-visible too, but i dont know). We used to be 240V here too (same sturdy and safe 3pin british sockets too, thanks be $DEITY - cant stand those damned continental 2-pins). However, my UPS has reported ~230V AC supply RMS voltage ever since i bought it in 1999.
Ie, I think you're right, most of EC has probably long standardised on 230V nominal supply voltage, but no one remembers anything but the figure they were taught in school 15+ years ago (240V for UK/Eire, 220V for most of Europe).
However, if you use bandwidth you're not simply using it as you would use a radio signal. Your intrusion (because that's what it is) is causing other users' bandwidth to decrease. Not only that, but you're active on a network that you shouldn't have access to, which could be considered illegal. I'm not familiar with US laws.
The problem with this idea is that simply by turning on your 802.11b wireless NIC you'll interfere with bandwidth. If you connect to one AP (legitimately), you'll definitely interfere with bandwidth available to other 802.11 networks.
See, that's the whole point of the 2.4GHz band, it's a public, unregulated band and we're free to transmit as we want (within power limits), whether it reduces bandwidth for others or not. So the "they're stealing bandwidth!" argument simply does not wash, because I can "steal" this bandwidth at the physical layer without ever communicating with the AP at the 802.11 layer and the AP operator can do nothing about it, because it is my right to transmit on 2.4GHz as much as anyone elses (power limits and public nuisance laws not withstanding).
2.4GHz is free for all. The comms authorities of the world got together at ITU and decided to reserve 2.4GHz band for public, unregulated use (within power limits). The FCC in the USA and ETSI in europe (implemented by each countries radio regulatory authority) mandate 2.4GHz as public property. You have *0* right to exclusive use of 2.4GHz, it's no more yours than mine. If you want guaranteed bandwidth, you need to go licence spectrum from FCC or whatever body it is in your country (eg ComReg in mine).
Accessing computer networks without authorisation is a different thing. It has nothing to do with bandwidth usage of the wireless network. If you access their internet connection and use that bandwidth, then yes you may be breaking a law. However, whether or not accessing an open AP counts as unauthorised network access is a different matter, and would depend on your jurisdiction and the opinion of a judge.
Many people allow public access to their 802.11 APs, and leave them open deliberately and do not mind them being used. How do you tell which APs are open by intent and which are open accidently? Really, the only way reliable way indicator of whether public access is allowed is whether or not the AP has been configured to allow open access.
Me, I have my APs set to use WEP (for my own clients) but to still allow open access.;)
When I discovered that it was because my laptop had an built-in wifi card that had a binary only driver (and I had to jump through hoops to install 3rd party support for it) I instantly just wiped Suse out and put Windows back on. A laptop is COMPLETELY USELESS to me without a functioning wireless interface.. and I don't want to have to jump through hoops to get the damn thing to work either - it's just frustrating.
I'm sure it is.
Whose fault is it though? The kernel guys' fault for not being able to write drivers for hardware they can not get specs for? SuSe's fault for not including the binary-only version driver, the licence for which almost certainly prohibits redistribution?
Why dont you blame these hardware manufacturers for not releasing programming specifications (openly or under NDA to kernel hackers) and/or not writing a GPL driver? They're the only ones who can change the situation. They're the ones to pin the blame on for making your Linux experience frustrating.
Complaining on/. about SuSe, who can not do anything about it is senseless.
I remember reading before the internet that France had some internet for their country. It was much like our gopher system in the early days of the internet. But everyone was identifiable, and they could remove useless content. I think I remember reading it is still popular and is in use. I wish I could remember the name of it.
1. Router LSA, fundamental LSA. Used to describe each router-node, particularly their links, in SPF
2. Network LSA, fundamental LSA. Generated by the designated router on each network, to describe that network for SPF.
3. Summary LSA, used to aggregrate networks between areas
4. ASBR summary, OSPF AS global LSAs to describe ASBRs, in particular, without a corresponding ASBR-summary for an originating router, an AS-External route is not valid.
5. AS external, to describe routes external to the OSPF domain (OSPF AS), ie routes which are distributed into OSPF
6. Multicast Group Membership, for use by the mostly defunct MOSPF multicast routing protocol.
7. NSSA LSA, used to allow ASBRs to exist in otherwise stubby areas (NSSA areas), only valid within NSSA areas, must be translated at the NSSA ABR if one wishes to propogate the routing information to rest of OSPF domain
8. External attributes, defunct
9, 10, 11. Opaque LSAs, with a distribution scope of link, area, AS respectively. Used to propogate arbitrary data, opaque to OSPF itself, over the OSPF protocol, eg traffic-engineering information, information to allow graceful restart of OSPF, even HA clustering information.
I never understood why nobody made an Athlon=>Alpha shim board to run to run an K7 in an Alpha EV6 box or vice versa.
I doubt it'd have been pin-compatible.
However the defunct API Networks, the company into which HP and Samsung divested marketing of Alpha, did make an Alpha/AMD box, the UP100 and UP1100 (best link I could find), which was a Alpha EV67 with an AMD AMD-751 "Irongate" "northbridge" as memory/pci/agp/ev6 controller, as used on early K7 boards. The UP1x00 wasnt too popular though, it's memory performance sucked compared to the EV67 boxes which used the DEC 21272 "Tsunami" northbridge, like the DEC DP264 and API Networks UP2000.
To be specific, their 256 cpu Itanic cluster hence can find collisions in just:
80000/256/24 = 13 days
Also, SHA computation is simple integer arithmetic so, assuming the attack also is integer only, you dont need expensive Itanics (which have an edge on floating point only), a cheap P4 or Athlon will do the job as well. Finding collisions in SHA-0 within a few weeks is then likely quite possible for less than 500m (ie available even to smallish companies) or for any group that has enough members to give it access to a few hundred PCs.
If the attack extends to SHA-1, this is fairly bad news..
3. Ok, seemingly hard lockups are out - but these are more and more uncommon with 2.6 - but nearly everything else is dumpable, unless it screws with the NIC driver itself? I'll have to check out that watchdog, that sounds useful.
What do you mean by "hard lockups" exactly? NMI cant be disabled AIUI, so unless hardware were hung, there's a good chance you can get a panic.
1. Correct 2. Correct, the purpose of netconsole being to capture kernel messages that otherwise would not be sent due to a hung kernel - all other messages, syslog can take of. 3. Unlikely. If the kernel can panic, it can use netconsole. Used with nmi_watchdog, you can even get a stack trace from a stuck interrupt handler 4. No, netconsole uses polling-mode and drives the nic driver directly.
Note that netconsole can also dump the entire contents of RAM.
You can't have a system of proportional representation in a presidential election, it's a contradiction in terms!
It is sort of yes;)
But despite the contradiction in the name, PR-STV can be used for single-winner, AFAIK. Set the quota to 50%+ of votes and knock out all candidates bar the top 2 after the first round and transfer the surplus votes, or somesuch. I'm reasonably sure we use a form of PR-STV here in Ireland for our presidential elections.
I'm not 100% sure, but iirc we use PR-STV for our presidential election here in Ireland (as we do for our dail (parliament), county council and EU elections). So PR-STV can be used for single-winner elections, AFAIK.
What you want is a Proportional Representation voting system. Then you could could give your first preference vote to $3rd_party preference and give $mainstream-lesser-evil your second preference vote.
If you want to keep track of finances and you do it with integers (better than floating point for money), then you max out at ~65000 pennies. That $650. It's not that hard to wrap around a 16bit number in real life.
Jeebus... Are you trolling, or do you really think software on computers with 16bit word size cant count past 65k? multiple word arithmetic is quite easily possible. (anyone with primary school maths should be able to work out how to do multiple word addition, subtract, multiply at least..)
Very restrictive licence. Noone is allowed to distribute qmail modified in *any* way, you cant even change the install paths, you *must* accept DJB's unrelated-to-SMTP ideas on where software should be installed. You cant add patches, etc..
Qmail is not Free Software, and a maintainance nightmare because of the licence.
Ok, I'm giving up here. We're basically saying the exact same thing, except you keep saying I'm not.
Then I've misunderstood you somehow.
one as how it would be if you accept certain conclusions (e.g. from the GPL FAQ)
Ah, well ok. I've basically ignored the GPL faq, or at least not drawn from it in my own arguments. The answers you've quoted from it seem to suggest the GPL FAQ is possibly not a terribly good interpretation of the GPL in places. I got the impression we both were of that opinion. If you were extrapolating from the GPL FAQ, then I've been knocking down the FAQ.
(note that in many places where I wrote "you" or attacked "your opinion" I did realise that it was merely an argument you were presenting rather than your actual opinion. Attributing the argument to you directly was a convenient shorthand).
Or, I make a statement with certain conditions (e.g. a 3a distribution), and you say "no, that's not true, since (changed condition, e.g. 3b distribution)".
Possible. I'd have to read back again. However, for some of the arguments, the rebuff required "given those conditions, x can not be so, it can only be y". It did seem to become circular though.
Thanks for the exchange!
ditto!
I'd say thanks for the debate, but I'm not really sure it was one.
LOL. Yeah, agreed, as I said in other post, one or the other of us or both is missing the others points. I'm not quite sure which way(s) or to what extent though. Maybe we've been in violent agreement.;)
Slashdot Mirror
To my horror, once I finally got the thing installed I learned that it doesn't even come with a compiler. Sure you can add GCC to it,
;) Dont forget, that if you install a Solaris 9 packaged gcc, you'll need to run the not-really-aptly-named "fix-headers" (or somesuch) command. Gcc likes to munge system headers and sneakily use the munged copies instead, and you'll hit problems if your gcc is using S9 munged headers and you try compile on S10.
Right, and indeed I think gcc might even be on the companion 'sfw' CD, if you get full Solaris 10 media. I'm not sure, I've never installed Solaris 10 from CD media
there must be some art to making GNU's tools work properly with Sun's libc that is beyond me. The biggest problem I had was libtool seems to be completely broken with respect to shared libraries on Solaris.
What problems did you have? I regularly use libtool on Solaris, and have not had problems. Libtool, despite its complexity (as with all the auto* GNU build system tools) when things go wrong, does a good job 99% of the time.
one problem with scaling the opteron up to such large systems is the physical address space in current implementations. It's 40 bits.
;)
I'm sure AMD could be persuaded to build opteron's with more phys address space if demand were there. Though, I have no idea how ingrained the 40 bit assumption is in the AMD64 MMU structures. But at worst, with a CPU feature bit, it could be enabled selectively, if it is not completely configurable in setup of the page directory already. AMD would have been idiots not to allow for expansion of the phys address space.
The other thing the hub chip does on SGI machines is to handle the page directory stuff, counting remote vs local accesses, and handling automatically migrating pages to nodes that are making heavy use of them. Most of this cool stuff is patented, so I doubt the Opteron is doing it.
AIUI, its left to software on Opteron. So it's all down to your OS. Linux has some support for software balanced ccNUMA (mostly process CPU affinity, and allocating pages on the affine CPU, if possible - dont think it can migrate pages between nodes yet, but i could be wrong).
Ironically Opteron would be a less good fit than, say, Xeon, for the engineering that SGI have done for MIPS and Itanium, exactly because it has native support for ccNUMA and has integral memory controllers.
;)
Newisys have a chipset that extends the CC and addressing of Opterons so that you can put upto 32 in a system.
Well, I didnt mean a drop-in fit to Origin, but more building a large ccNUMA machine with Opteron. That someone has already done with Opteron what it took SGI many many years to engineer with Origin (and reeingeer for Itanic) demonstrates my point that Opteron would be less effort, no?
BTW, as for reimplementation, check out what DEC did with the Alpha Wildfire machines (the big GS series machines), indeed with the 21364 DEC^WCompaq actually integrated routing protocols into the CPUs. The opteron ccNUMA stuff derives more from DECs approach than SGIs (namely CPU as actively aware participant in a ccNUMA topology, compared to SGI and others approach to Itanic and Xeon of building the smarts around 'dumb' CPUs)
Itanium cpu limit: 512 cpus.
SGI is being very successfull with it's 512 itanium machines running Linux.
Note that SGI are doing this with very very special hardware. IIRC each CPU brick in an Origin has 4 itanics. All these bricks are then interconnected with very very special CPU interconnect routers.
That these machines go to 512 CPUs has *nothing* to do with the CPUs being Itanic, it's all down to the ccNUMA interconnect technology (which SGI initially acquired from Cray). If you need further convincing of this, note that the Origin 3k architecture SGIs machines have essentially the same architecture, but use MIPS CPUs. This architecture could be applied to Opteron too, and probably with less effort, as Opteron natively supports ccNUMA and comes with CPU networking built-in.
Pentium Pro, another hyped up CPU that never really delivered.
Hang on, you are joking, right?
PPro has probably been Intel's best chip architecture to date. The initial P6 had bad 16bit performance, which made it a bad choice for consumers are that time, but it was very competitive in normal 32bit mode, idea for NT, Linux and other PC Unixen. The 2nd iteration of the P6 architecture fixed the 16bit issue and was enormously successful. The latest iteration of that arch (Pentium-M) is quietly outperforming the architecture designed to replace it, the P4, at nearly half the clock speed and far less power usage. Indeed, it looks like Intel will be going *back* to the P6 family in future as its 'frontline' PC architecture.
So you must be joking.
We're talking around each other now, saying the exact same thing essentially.
To specify nominal value ± tolerance (or even with seperate + and - tolerance, as in UK) is same thing as specifying a range. In the EU, it's specified as nominal/tolerance - maybe in USA it's specified as a range, but it's two ways of expressing same values ultimately!
Urr, yes of course it is standardised - dont be silly. Note that standardised figures can include a specified tolerance hence covering a range. The nominal EU AC domestic supply voltage is 230V ± whatever V of tolerance @ 50Hz. Probably about ± 5% of nominal, as that covers everything from 220 to 240V. However, note there is a nominal single standardised voltage.
In reality, supply voltage is dependent on load, the power company actually has to monitor and change the supply as the load changes (eg from day to night time when people use power differently), to maintain acceptable supply voltage levels (and efficient power transfer). Also, the power companies tend to monitor and modulate supply based at a far more macroscopic level of many many thousands of loads aggregated. The local view can be very different (eg, you might have a DIY crazy neighbour who regularly uses heavy duty electric drills, or you might have a geek neighbour who runs a whole bunch of computers 24x7 and draws a kW or more of continious resistive power from your local loop).
I thought the UK ran at 240 volts, not 250. I thought South Africa was the only country that ran 250. I though most of Europe was 220 and Europe and the UK were going to unify their standards at 230.
I thought UK was 240V too. I used to live there about 15 years ago and distinctly remember the figure of 240 mentioned in everything from physics classes to electrical safety literature. I now live (and have even before living in UK) in Ireland which for historical reasons long had identical consumer-visible standards/arrangements for electricity (probably deeper than just consumer-visible too, but i dont know). We used to be 240V here too (same sturdy and safe 3pin british sockets too, thanks be $DEITY - cant stand those damned continental 2-pins). However, my UPS has reported ~230V AC supply RMS voltage ever since i bought it in 1999.
Ie, I think you're right, most of EC has probably long standardised on 230V nominal supply voltage, but no one remembers anything but the figure they were taught in school 15+ years ago (240V for UK/Eire, 220V for most of Europe).
Its going to be a mail/file server. I think you don't need a ton of horsepower.......
Until you try run Spamassassin and Clamav to filter spam and windows-virus-cruddage and wonder why email takes days to arrive...
You forgot GAA football / Gaelic football, played in Ireland. Which is mostly what the Aussie rules football derives from.
However, if you use bandwidth you're not simply using it as you would use a radio signal. Your intrusion (because that's what it is) is causing other users' bandwidth to decrease. Not only that, but you're active on a network that you shouldn't have access to, which could be considered illegal. I'm not familiar with US laws.
;)
The problem with this idea is that simply by turning on your 802.11b wireless NIC you'll interfere with bandwidth. If you connect to one AP (legitimately), you'll definitely interfere with bandwidth available to other 802.11 networks.
See, that's the whole point of the 2.4GHz band, it's a public, unregulated band and we're free to transmit as we want (within power limits), whether it reduces bandwidth for others or not. So the "they're stealing bandwidth!" argument simply does not wash, because I can "steal" this bandwidth at the physical layer without ever communicating with the AP at the 802.11 layer and the AP operator can do nothing about it, because it is my right to transmit on 2.4GHz as much as anyone elses (power limits and public nuisance laws not withstanding).
2.4GHz is free for all. The comms authorities of the world got together at ITU and decided to reserve 2.4GHz band for public, unregulated use (within power limits). The FCC in the USA and ETSI in europe (implemented by each countries radio regulatory authority) mandate 2.4GHz as public property. You have *0* right to exclusive use of 2.4GHz, it's no more yours than mine. If you want guaranteed bandwidth, you need to go licence spectrum from FCC or whatever body it is in your country (eg ComReg in mine).
Accessing computer networks without authorisation is a different thing. It has nothing to do with bandwidth usage of the wireless network. If you access their internet connection and use that bandwidth, then yes you may be breaking a law. However, whether or not accessing an open AP counts as unauthorised network access is a different matter, and would depend on your jurisdiction and the opinion of a judge.
Many people allow public access to their 802.11 APs, and leave them open deliberately and do not mind them being used. How do you tell which APs are open by intent and which are open accidently? Really, the only way reliable way indicator of whether public access is allowed is whether or not the AP has been configured to allow open access.
Me, I have my APs set to use WEP (for my own clients) but to still allow open access.
When I discovered that it was because my laptop had an built-in wifi card that had a binary only driver (and I had to jump through hoops to install 3rd party support for it) I instantly just wiped Suse out and put Windows back on. A laptop is COMPLETELY USELESS to me without a functioning wireless interface.. and I don't want to have to jump through hoops to get the damn thing to work either - it's just frustrating.
/. about SuSe, who can not do anything about it is senseless.
I'm sure it is.
Whose fault is it though? The kernel guys' fault for not being able to write drivers for hardware they can not get specs for? SuSe's fault for not including the binary-only version driver, the licence for which almost certainly prohibits redistribution?
Why dont you blame these hardware manufacturers for not releasing programming specifications (openly or under NDA to kernel hackers) and/or not writing a GPL driver? They're the only ones who can change the situation. They're the ones to pin the blame on for making your Linux experience frustrating.
Complaining on
Sun keyboards have cut/paste/etc. keys and they're very useful.
I remember reading before the internet that France had some internet for their country. It was much like our gopher system in the early days of the internet. But everyone was identifiable, and they could remove useless content. I think I remember reading it is still popular and is in use. I wish I could remember the name of it.
minitel
1. Router LSA, fundamental LSA. Used to describe each router-node, particularly their links, in SPF
2. Network LSA, fundamental LSA. Generated by the designated router on each network, to describe that network for SPF.
3. Summary LSA, used to aggregrate networks between areas
4. ASBR summary, OSPF AS global LSAs to describe ASBRs, in particular, without a corresponding ASBR-summary for an originating router, an AS-External route is not valid.
5. AS external, to describe routes external to the OSPF domain (OSPF AS), ie routes which are distributed into OSPF
6. Multicast Group Membership, for use by the mostly defunct MOSPF multicast routing protocol.
7. NSSA LSA, used to allow ASBRs to exist in otherwise stubby areas (NSSA areas), only valid within NSSA areas, must be translated at the NSSA ABR if one wishes to propogate the routing information to rest of OSPF domain
8. External attributes, defunct
9, 10, 11. Opaque LSAs, with a distribution scope of link, area, AS respectively. Used to propogate arbitrary data, opaque to OSPF itself, over the OSPF protocol, eg traffic-engineering information, information to allow graceful restart of OSPF, even HA clustering information.
that's about it..
I never understood why nobody made an Athlon=>Alpha shim board to run to run an K7 in an Alpha EV6 box or vice versa.
I doubt it'd have been pin-compatible.
However the defunct API Networks, the company into which HP and Samsung divested marketing of Alpha, did make an Alpha/AMD box, the UP100 and UP1100 (best link I could find), which was a Alpha EV67 with an AMD AMD-751 "Irongate" "northbridge" as memory/pci/agp/ev6 controller, as used on early K7 boards. The UP1x00 wasnt too popular though, it's memory performance sucked compared to the EV67 boxes which used the DEC 21272 "Tsunami" northbridge, like the DEC DP264 and API Networks UP2000.
So yes, the "Athlon" Alpha did exist..
To be specific, their 256 cpu Itanic cluster hence can find collisions in just:
80000/256/24 = 13 days
Also, SHA computation is simple integer arithmetic so, assuming the attack also is integer only, you dont need expensive Itanics (which have an edge on floating point only), a cheap P4 or Athlon will do the job as well. Finding collisions in SHA-0 within a few weeks is then likely quite possible for less than 500m (ie available even to smallish companies) or for any group that has enough members to give it access to a few hundred PCs.
If the attack extends to SHA-1, this is fairly bad news..
3. Ok, seemingly hard lockups are out - but these are more and more uncommon with 2.6 - but nearly everything else is dumpable, unless it screws with the NIC driver itself? I'll have to check out that watchdog, that sounds useful.
What do you mean by "hard lockups" exactly? NMI cant be disabled AIUI, so unless hardware were hung, there's a good chance you can get a panic.
--paulj
1. Correct
2. Correct, the purpose of netconsole being to capture kernel messages that otherwise would not be sent due to a hung kernel - all other messages, syslog can take of.
3. Unlikely. If the kernel can panic, it can use netconsole. Used with nmi_watchdog, you can even get a stack trace from a stuck interrupt handler
4. No, netconsole uses polling-mode and drives the nic driver directly.
Note that netconsole can also dump the entire contents of RAM.
You can't have a system of proportional representation in a presidential election, it's a contradiction in terms!
;)
It is sort of yes
But despite the contradiction in the name, PR-STV can be used for single-winner, AFAIK. Set the quota to 50%+ of votes and knock out all candidates bar the top 2 after the first round and transfer the surplus votes, or somesuch. I'm reasonably sure we use a form of PR-STV here in Ireland for our presidential elections.
I'm not 100% sure, but iirc we use PR-STV for our presidential election here in Ireland (as we do for our dail (parliament), county council and EU elections). So PR-STV can be used for single-winner elections, AFAIK.
What you want is a Proportional Representation voting system. Then you could could give your first preference vote to $3rd_party preference and give $mainstream-lesser-evil your second preference vote.
If you want to keep track of finances and you do it with integers (better than floating point for money), then you max out at ~65000 pennies. That $650. It's not that hard to wrap around a 16bit number in real life.
Jeebus... Are you trolling, or do you really think software on computers with 16bit word size cant count past 65k? multiple word arithmetic is quite easily possible. (anyone with primary school maths should be able to work out how to do multiple word addition, subtract, multiply at least..)
It is not GPL'd, it has a BSD lic. But it's source is freely avail. And it is FREE.
Have you actually read the qmail licence? Indeed, can you even find the licence for qmail? The closest thing I can find for a qmail licence is:
http://cr.yp.to/qmail/dist.html
Very restrictive licence. Noone is allowed to distribute qmail modified in *any* way, you cant even change the install paths, you *must* accept DJB's unrelated-to-SMTP ideas on where software should be installed. You cant add patches, etc..
Qmail is not Free Software, and a maintainance nightmare because of the licence.
Ok, I'm giving up here. We're basically saying the exact same thing, except you keep saying I'm not.
;)
Then I've misunderstood you somehow.
one as how it would be if you accept certain conclusions (e.g. from the GPL FAQ)
Ah, well ok. I've basically ignored the GPL faq, or at least not drawn from it in my own arguments. The answers you've quoted from it seem to suggest the GPL FAQ is possibly not a terribly good interpretation of the GPL in places. I got the impression we both were of that opinion. If you were extrapolating from the GPL FAQ, then I've been knocking down the FAQ.
(note that in many places where I wrote "you" or attacked "your opinion" I did realise that it was merely an argument you were presenting rather than your actual opinion. Attributing the argument to you directly was a convenient shorthand).
Or, I make a statement with certain conditions (e.g. a 3a distribution), and you say "no, that's not true, since (changed condition, e.g. 3b distribution)".
Possible. I'd have to read back again. However, for some of the arguments, the rebuff required "given those conditions, x can not be so, it can only be y". It did seem to become circular though.
Thanks for the exchange!
ditto!
I'd say thanks for the debate, but I'm not really sure it was one.
LOL. Yeah, agreed, as I said in other post, one or the other of us or both is missing the others points. I'm not quite sure which way(s) or to what extent though. Maybe we've been in violent agreement.
See ya around!