there is no compelling argument, NO COMPELLING ARGUMENT to use anything more than
1. pencil 2. paper 3. optical scanner I'd extend your point 3 to require that the optical scanner's be installed in someones skull. In Scotland we replaced our age old, reliable 10,000 grannies in sportshalls approach to vote counting with optical scanners this year. The result was a complete farce with thousands of votes being declared void. Human eyeballs are the best scanner and with appropriate oversight the fairest vote counting mechanism IMHO.
Welcome to Scotland! I work about 200 yds from HECToR's home. It's a nice part of the world to work just next to the Pentland Hills outside Edinburgh. Quite a trek to get into Edinburgh itself though.
I guess that since the demise of the INMOS Transputer the UK hasn't had much to do with the design of supercomputers...though the architecture of the XT4 does seem to bear some similarity to the Transputer systems of the late 80s/early 90s.
So, wow, but still, whats so special about it? It does not make me any wiser or more important than any other user of slashdot. Apart from the ability to see through ESR and leap over Jon Katz stories in a single bound that is...
Your comments on ring mains are utter bollocks. If you had read the IEE Wiring Regulations you would know that the ring design is more than capable of dealing with a cable break without overloading.
That's actually quite an interesting idea (apart from the perpetual motion bit;). You'd never be able to generate steam but there is definitely a lot of waste heat from your datacenter (well 100% waste as virtually all the electricity that goes in is converted to heat). This could form the basis of a Combined Compute and Heat scheme where the heat is used to provide background heat (under floor heating say) to surrounding homes and offices.
It wouldn't handle the summer months where the demand for cooling is greatest. The cooling system could "charge up" a giant heat reservior during the summer for use during the winter. I guess the infrastructure would be somewhat complex and it would suffer from the same problems that blight CHP schemes...getting different organisations to cooperate on captial projects.
Maybe when energy gets to cost 5x what it does at the moment.
I build networked CCTV equipment as my day job. According the people who install our stuff the best way to get a potential thief or vandal to stop what they are doing is to say "Stop immediately and stay where you are the police have been notified". They usually turn and flee straight away...which is really the best option (at least for private property) where preventing too much damage is usually more important than apprehending the culprit. Sad but true.
I understand why people are wary of CCTV but there is a lot of very unnecessary negative feeling towards it. It could be used for bad but it is used for a lot of good. There are a lot of crimes carried out against people which just could not be solved without the CCTV evidence or leads obtained from CCTV. CCTV really does make the world a safer place (if it didn't I wouldn't be working in the industry trust me).
I work for a company (shameless plug, sorry) that specialises in IP network video surveillance: IndigoVision
Most of our product line is based on embedded Linux. We provide video transmitter boxes (or racks) that encode video from standard analog cameras (including Pelco PTZ) into MPEG-4. This is then recorded on either an embedded Linux Networked Video Recorder or on a suitably configured Windows server. All of our Linux boxes have built in iptables firewalls so you can restrict access to the devices in any way you choose. We also have state of the art Windows based Control Center software to control and monitor the solution.
The technology is very secure and many of our customers use it for internet based security monitoring. It also scales to very large installations...one of our partners used our products to provide CCTV coverage for the Winter Olympics in Torino which required several thousand video transmitters.
I can fully appreciate why Pelco don't recommend using anti-virus software on machines that record video. We don't recommend that either for our Windows based solution. The amount of data flowing through the system can be immense and virus-scanners are really not suited to the load (even if you exclude the actual video store). However you can do a *LOT* to avoid security problems on Windows by locking down the system as per Microsofts guidelines and running the video recorder software as a low privilege network service. Also avoiding the use of potentially vulnerable protocols like DCOM is a good idea too (we have our own cross-platform security audited protocols).
That said I do like our Embedded Linux servers. Very hassle free and almost nothing to manage...
Don't discount batch injection too quickly. This is the technique used by the cheaper aftermarket EFI units (Edelbrock, etc) costing many $$$. It gives you quite a boost over a carb with a lot better idling, throttle response, emissions and gas mileage.
The hard bit, as always, is getting all of the plumbing in place to get per-cylinder injectors. I intend to convert my carb SB Chevy to run with megasquirt and batch injection as a stepping stone to a full EFI332 based sequential solution. The megasquirt is definitely a useful technology for upgrading old EFI or mech FI to modern fueling algorithms as well as a stepping stone for those that want to progress to more expensive/harder FI systems (which is what the Megasquirt project grew out of BTW).
Urgh. Have you used a paper tape punch. My company maintains a couple of them for a customer who needs a secure unhackable write-only data transfer mechanism. They break down all the time and getting spares is a nightmare....
This stuff should be regulated. If you store personal info on an HDD for business purposes, you should have a legal responsibility (i.e. one that comes with repricussions if not met) to ensure that even after a drive is retired, the data is safe.
In the UK the Data Protection Act makes anyone who collects data on someone responsible for its safe keeping and destruction. It's backed up with heavy fines, prison sentences and raids on people who are suspected of breaching it....and I'll bet the same stupidity happens just as much here...but having the law probably helps.
This is one of the reasons why you should always clear any memory in your program that contained passwords, encryption keys, etc after you have finished using them. That way you minimise the chance of it leaking out if there is a fault in your app, or another one if your OS doesn't zero memory before use by another program (Win95 doesn't but NT does(?)).
Eric Raymond is running a project to create a bayesian spam filter written in C. I've been experimenting with it today and it looks pretty good. It's written as a client-server daemon that makes it *very* quick compared with a perl or other scripted approach. It's also pretty easy to integrate into any unix mail system.
The program, bogofilter, is stores its word lists using HP Judy dynamic arrays. Pretty interesting way of storing stuff and new to me anyway: http://sf.net/projects/judy
Those racecars share more in common with my daily-driver 1976 Dodge pickup truck than does a typical ricer's car. My '76 Ram has a 400 (6.6L) V8 driving the rear wheels. With a curb weight of 4,000lb, it's about twice the weight of a Honda Civic. But 6.6L / 1.6L = 4.125 times more engine, and all other things being equal, 4.125 times the power. Into only twice the weight.
Now imagine if you put something similar (a Chevy 350 say) in something that weighs less than a Honda Civic. If you did that you might come up with something like one of these.
After you've been for a test drive in one it takes about a week before you stop grinning....
Find out what it takes to "hack" your own one here.
The trash 80 has a perfectly good emulator: http://www.tim-mann.org/xtrs.html
Unfortunately I haven't been able to try out all of my old TRS-80 Model I software on it because the box with my tape of Space Invaders is right at the back of my loft and I'm too lazy;)
Tux is not just for kernel space. It has a very clever loadable module API for writing dynamic weblications (it is being actively researched by ad agencies judging by the Tux mailing list BTW). The modules can either run in kernel mode or user mode but obviously are much safer in userland. There is a clever mechanism for handing off control to the userland code whilst minimising context switches (as far as I understand). This allows for blazingly fast CGI like behaviour.
It also has a clever mechanism for running CGI apps more quickly than something like Apache.
If Tux didn't have support for dynamic content then it wouldn't pass the SPECWeb benchmark which has a large dynamic content from what I've seen.
In all other respects though you are right. It isn't as fully featured as a web server like Apache. I *am* hoping however that it will mature into a rock solid platform for dynamic weblication developemnt which really needs the drop in latency (rather than raw throughput) that you get from Tux's clever CGI architecture. Much smarter than Apache modules or ISAPI.
I saw statistics on this a couple of years ago. The traffic balance is around 50:50 but us Europeans pay soemthing like 80:20 of the cost of the link to connect to the "american" internet;)
I guess he's never heard of/used QNX, ChorusOS
Nucleus, or ThreadX. I did however like the gadgets,
but taking a look at the last week, with all the
Linux related companies going to the dogs , and 4
distributions going "kaput" within less than 6
months time, I would be looking at other
alternatives to Linux, especially if my business
were going to depend on them.
Of course you could look at RTEMS. The longstanding (1988, way before Linux existed) Open Source real-time operating system. It is as good as (many say better than) vxWorks, ChorusOS, Nucleus and ThreadX. http://www.rtems.com
The one thing you wont get with RTEMS is the hype and geeky admiration of Linux geeks (or vxWorks, QNX, Windows CE, PalmOS,etc luvvies). You just put it in things you *REALLY* want to work.
If you give people interesting ads, they will click. It's different to the real world. Since the success of ads is measured by clickthroughs (i.e. internet advertising is a direct sales route), then what you have to do is give people things they are directly interested in.
Amen to that.
Please, OSDN, give me:
- relevant adverts for me
- relevant adverts for the geographic region I come from. Cheap DSL in Utah is not really very interesting to a Scotsman living in Scotland!!! Everybodies time and money is wasted if you try to sell me something I can't buy.
- the ability to turn off adverts that I hate. Show me something irritating too often and I can guarantee that I will actively buy a competitors product.
Simple user profiles for adverts sound like a brilliant idea...and unlike doubleclick, OSDN sites have the neural capability to actually *use* the information for _their_ gain.
Am I alone in finding this one of the most interesting slashot stories ever (well since I started reading Slashdot in late 97)?
No mindless rhetoric, no INAL comments, no anti-MS(or xBSD) Linux rocks comments. Just no-nonsense, no BS, approachable technical (and idealogical) information and commentary.
Why don't guys like Intel and AMD put a teenie weanie bit of PROM in their chips (like they had for the serial number) that gets burned with the bus speed/multiplier? Then software (like the BIOS) could read this back when it does its usual CPUID trick and verify that the chip was running in spec.
The BIOS (and/or OS) could report that the system was being over/under clocked. The OCers would love this (especially if it came up in large red and orange letters;) and innocents wouldn't get burned by dodgy resellers.
This strikes me as being pretty easy to implement particularly if you consider that Intel has already done this with its CPU serial number.
Rambus is still lame and DDR-SDRAM is still the future of PC/Workstation memory.
Rambus is not dying the death solely because the manufacturers have to pay licence fees. It's slow, unreliable and difficult to make and thus nobody should recommend it for their systems. period.
>Note that Cygnus makes most of its revenue from >distributing proprietary tools.
Not so. They make most of their money from development and support of custom, well tested, versions of gcc, gdb and binutils.
Cygnus contributes a *LOT* of man hours back to the community mainly through gcc, binutils and gdb. The vast majority of new CPU ports for gcc come from Cygnus for example. They have a lot of very talented engineers that are only too eager to help with difficult problems. What RedHat does is, by comparison, chicken feed.
An intersting thing to note is that Cygnus has been making honest money out of Open Source for the past 10 years or so...just how long is it going to be till RedHat make a single cent?
And, yes I do know that RedHat owns Cygnus now...;)
Not all programmers are highly paid... It'd be nice to know what was worth throwing the money at and what wasn't (apart from not bothering with fancy gfx cards;). This is especially true if you have a stingy boss and/or a lot of development machines.
Slashdot Mirror
1. pencil
2. paper
3. optical scanner I'd extend your point 3 to require that the optical scanner's be installed in someones skull. In Scotland we replaced our age old, reliable 10,000 grannies in sportshalls approach to vote counting with optical scanners this year. The result was a complete farce with thousands of votes being declared void. Human eyeballs are the best scanner and with appropriate oversight the fairest vote counting mechanism IMHO.
Welcome to Scotland! I work about 200 yds from HECToR's home. It's a nice part of the world to work just next to the Pentland Hills outside Edinburgh. Quite a trek to get into Edinburgh itself though.
I guess that since the demise of the INMOS Transputer the UK hasn't had much to do with the design of supercomputers...though the architecture of the XT4 does seem to bear some similarity to the Transputer systems of the late 80s/early 90s.
Your comments on ring mains are utter bollocks. If you had read the IEE Wiring Regulations you would know that the ring design is more than capable of dealing with a cable break without overloading.
That's actually quite an interesting idea (apart from the perpetual motion bit ;). You'd never be able to generate steam but there is definitely a lot of waste heat from your datacenter (well 100% waste as virtually all the electricity that goes in is converted to heat). This could form the basis of a Combined Compute and Heat scheme where the heat is used to provide background heat (under floor heating say) to surrounding homes and offices.
It wouldn't handle the summer months where the demand for cooling is greatest. The cooling system could "charge up" a giant heat reservior during the summer for use during the winter. I guess the infrastructure would be somewhat complex and it would suffer from the same problems that blight CHP schemes...getting different organisations to cooperate on captial projects.
Maybe when energy gets to cost 5x what it does at the moment.
I build networked CCTV equipment as my day job. According the people who install our stuff the best way to get a potential thief or vandal to stop what they are doing is to say "Stop immediately and stay where you are the police have been notified". They usually turn and flee straight away...which is really the best option (at least for private property) where preventing too much damage is usually more important than apprehending the culprit. Sad but true.
I understand why people are wary of CCTV but there is a lot of very unnecessary negative feeling towards it. It could be used for bad but it is used for a lot of good. There are a lot of crimes carried out against people which just could not be solved without the CCTV evidence or leads obtained from CCTV. CCTV really does make the world a safer place (if it didn't I wouldn't be working in the industry trust me).
Hi,
I work for a company (shameless plug, sorry) that specialises in IP network video surveillance: IndigoVision
Most of our product line is based on embedded Linux. We provide video transmitter boxes (or racks) that encode video from standard analog cameras (including Pelco PTZ) into MPEG-4. This is then recorded on either an embedded Linux Networked Video Recorder or on a suitably configured Windows server. All of our Linux boxes have built in iptables firewalls so you can restrict access to the devices in any way you choose. We also have state of the art Windows based Control Center software to control and monitor the solution.
The technology is very secure and many of our customers use it for internet based security monitoring. It also scales to very large installations...one of our partners used our products to provide CCTV coverage for the Winter Olympics in Torino which required several thousand video transmitters.
I can fully appreciate why Pelco don't recommend using anti-virus software on machines that record video. We don't recommend that either for our Windows based solution. The amount of data flowing through the system can be immense and virus-scanners are really not suited to the load (even if you exclude the actual video store). However you can do a *LOT* to avoid security problems on Windows by locking down the system as per Microsofts guidelines and running the video recorder software as a low privilege network service. Also avoiding the use of potentially vulnerable protocols like DCOM is a good idea too (we have our own cross-platform security audited protocols).
That said I do like our Embedded Linux servers. Very hassle free and almost nothing to manage...
hope this helps,
Dave
Don't discount batch injection too quickly. This is the technique used by the cheaper aftermarket EFI units (Edelbrock, etc) costing many $$$. It gives you quite a boost over a carb with a lot better idling, throttle response, emissions and gas mileage.
The hard bit, as always, is getting all of the plumbing in place to get per-cylinder injectors. I intend to convert my carb SB Chevy to run with megasquirt and batch injection as a stepping stone to a full EFI332 based sequential solution. The megasquirt is definitely a useful technology for upgrading old EFI or mech FI to modern fueling algorithms as well as a stepping stone for those that want to progress to more expensive/harder FI systems (which is what the Megasquirt project grew out of BTW).
Urgh. Have you used a paper tape punch. My company maintains a couple of them for a customer who needs a secure unhackable write-only data transfer mechanism. They break down all the time and getting spares is a nightmare....
This stuff should be regulated. If you store personal info on an HDD for business purposes, you should have a legal responsibility (i.e. one that comes with repricussions if not met) to ensure that even after a drive is retired, the data is safe.
In the UK the Data Protection Act makes anyone who collects data on someone responsible for its safe keeping and destruction. It's backed up with heavy fines, prison sentences and raids on people who are suspected of breaching it....and I'll bet the same stupidity happens just as much here...but having the law probably helps.This is one of the reasons why you should always clear any memory in your program that contained passwords, encryption keys, etc after you have finished using them. That way you minimise the chance of it leaking out if there is a fault in your app, or another one if your OS doesn't zero memory before use by another program (Win95 doesn't but NT does(?)).
Eric Raymond is running a project to create a bayesian spam filter written in C. I've been experimenting with it today and it looks pretty good. It's written as a client-server daemon that makes it *very* quick compared with a perl or other scripted approach. It's also pretty easy to integrate into any unix mail system.
Find out about it http://tuxedo.org/~esr/bogofilter/
The program, bogofilter, is stores its word lists using HP Judy dynamic arrays. Pretty interesting way of storing stuff and new to me anyway: http://sf.net/projects/judy
Those racecars share more in common with my daily-driver 1976 Dodge pickup truck than does a typical ricer's car. My '76 Ram has a 400 (6.6L) V8 driving the rear wheels. With a curb weight of 4,000lb, it's about twice the weight of a Honda Civic. But 6.6L / 1.6L = 4.125 times more engine, and all other things being equal, 4.125 times the power. Into only twice the weight.
Now imagine if you put something similar (a Chevy 350 say) in something that weighs less than a Honda Civic. If you did that you might come up with something like one of these.
After you've been for a test drive in one it takes about a week before you stop grinning....
Find out what it takes to "hack" your own one here.
The trash 80 has a perfectly good emulator: ;)
http://www.tim-mann.org/xtrs.html
Unfortunately I haven't been able to try out all of my old TRS-80 Model I software on it because the box with my tape of Space Invaders is right at the back of my loft and I'm too lazy
Recent versions of Mozilla allow you to just select the URL text (that isn't a hyper link) and click the middle mouse button to go to it.
;). It saves bucket loads of time!
All browsers should support this (if they don't already
Dave
Tux is not just for kernel space. It has a very clever loadable module API for writing dynamic weblications (it is being actively researched by ad agencies judging by the Tux mailing list BTW). The modules can either run in kernel mode or user mode but obviously are much safer in userland. There is a clever mechanism for handing off control to the userland code whilst minimising context switches (as far as I understand). This allows for blazingly fast CGI like behaviour.
It also has a clever mechanism for running CGI apps more quickly than something like Apache.
If Tux didn't have support for dynamic content then it wouldn't pass the SPECWeb benchmark which has a large dynamic content from what I've seen.
In all other respects though you are right. It isn't as fully featured as a web server like Apache. I *am* hoping however that it will mature into a rock solid platform for dynamic weblication developemnt which really needs the drop in latency (rather than raw throughput) that you get from Tux's clever CGI architecture. Much smarter than Apache modules or ISAPI.
Dave
Cha!
;)
I saw statistics on this a couple of years ago. The traffic balance is around 50:50 but us Europeans pay soemthing like 80:20 of the cost of the link to connect to the "american" internet
I guess he's never heard of/used QNX, ChorusOS Nucleus, or ThreadX. I did however like the gadgets, but taking a look at the last week, with all the Linux related companies going to the dogs , and 4 distributions going "kaput" within less than 6 months time, I would be looking at other alternatives to Linux, especially if my business were going to depend on them.
Of course you could look at RTEMS. The longstanding (1988, way before Linux existed) Open Source real-time operating system. It is as good as (many say better than) vxWorks, ChorusOS, Nucleus and ThreadX. http://www.rtems.com
The one thing you wont get with RTEMS is the hype and geeky admiration of Linux geeks (or vxWorks, QNX, Windows CE, PalmOS,etc luvvies). You just put it in things you *REALLY* want to work.
...that we can look forward to victims of Outlook virii being culled and their bodies burned as a grim reminder to others?
Shame. That's what's happening a few miles from my house to Foot and Mouth victims.
If you give people interesting ads, they will click. It's different to the real world. Since the success of ads is measured by clickthroughs (i.e. internet advertising is a direct sales route), then what you have to do is give people things they are directly interested in.
Amen to that.
Please, OSDN, give me:
- relevant adverts for me
- relevant adverts for the geographic region I come from. Cheap DSL in Utah is not really very interesting to a Scotsman living in Scotland!!! Everybodies time and money is wasted if you try to sell me something I can't buy.
- the ability to turn off adverts that I hate. Show me something irritating too often and I can guarantee that I will actively buy a competitors product.
Simple user profiles for adverts sound like a brilliant idea...and unlike doubleclick, OSDN sites have the neural capability to actually *use* the information for _their_ gain.
Am I alone in finding this one of the most interesting slashot stories ever (well since I started reading Slashdot in late 97)?
No mindless rhetoric, no INAL comments, no anti-MS(or xBSD) Linux rocks comments. Just no-nonsense, no BS, approachable technical (and idealogical) information and commentary.
Thanks guys!
Why don't guys like Intel and AMD put a teenie weanie bit of PROM in their chips (like they had for the serial number) that gets burned with the bus speed/multiplier? Then software (like the BIOS) could read this back when it does its usual CPUID trick and verify that the chip was running in spec.
;) and innocents wouldn't get burned by dodgy resellers.
The BIOS (and/or OS) could report that the system was being over/under clocked. The OCers would love this (especially if it came up in large red and orange letters
This strikes me as being pretty easy to implement particularly if you consider that Intel has already done this with its CPU serial number.
Rambus is still lame and DDR-SDRAM is still the future of PC/Workstation memory.
Rambus is not dying the death solely because the manufacturers have to pay licence fees. It's slow, unreliable and difficult to make and thus nobody should recommend it for their systems. period.
>Note that Cygnus makes most of its revenue from
;)
>distributing proprietary tools.
Not so. They make most of their money from development and support of custom, well tested, versions of gcc, gdb and binutils.
Cygnus contributes a *LOT* of man hours back to the community mainly through gcc, binutils and gdb. The vast majority of new CPU ports for gcc come from Cygnus for example. They have a lot of very talented engineers that are only too eager to help with difficult problems. What RedHat does is, by comparison, chicken feed.
An intersting thing to note is that Cygnus has been making honest money out of Open Source for the past 10 years or so...just how long is it going to be till RedHat make a single cent?
And, yes I do know that RedHat owns Cygnus now...
Not all programmers are highly paid... It'd be nice to know what was worth throwing the money at and what wasn't (apart from not bothering with fancy gfx cards ;). This is especially true if you have a stingy boss and/or a lot of development machines.