The thing with lasers is that the precautions needed depend heavilly on the particular laser. Some lasers don't really need any precautions at all. Others will cause blindness unless handled very carefully.
We have the laser classification system which is supposed to tell users roughly how dangerous a laser is. The higher the class the higher the precautions needed from basically none at all for classes I and II to careful planning of where the beam will end up, special protective gear for those working with the laser and interlocks so that if the laser area is voilated things get cut off instantly for class IV
This development does concern me. Essentially they are marketing a class 4 laser as a "big boys toy". Yes there are warnings on their site but i'm not cinvinced their warnings are sufficiant.
Ubuntu seems to have decided that the best time to make risky decisions is the release immediately after a LTS because if it ends up sucking people can stay on the lts without any worries about support disappearing.
I'm a beliver that AJAX and the like have thier place but are often misused.
In particular i'm against anything that works against "deep linking" of material on the web whether that be framesets where if you link to an individual page the site becomes unnavigable. AJAX that pulls all content into the same page, deliberate anti deep linking measures or whatever. Linking whether manually or by search engines is the whole point of the web!/. for all it's web design faults is actually a good example of how you can use AJAX to make a more friendly discussion system while preserving the ability to link to individual posts.
The problem is when someone makes a statement like that you don't know if they are using the terminology correctly or incorrectly. This means that whatever assumption you make you stand a significant chance of being wrong.
Which is a good reason as an author to avoid that terminology altogether just like you should avoid ambiguous date formats.
Hmm though at least so far none of the big journals seem to be bothering with any form of DRM, everything just comes down as unprotected pdf (though some of the pdfs are watermarked so they can trace which institution was the source of a piracy incident).
Re:Time to change your OS to OSX or BSD
on
Time To Dump XP?
·
· Score: 1
Here's a hint: Microsoft is headquartered in Washington state. Which is why I was somewhat surprised to see a date that only made sense in British format.
I guess they must be localising the pages or something.
Re:Time to change your OS to OSX or BSD
on
Time To Dump XP?
·
· Score: 1
Dates taken from the Windows XP Lifecycle page [microsoft.com]. hmm
April 14, 2009 (already in the past) for XP or XP SP1 The page you linked (unfortunately it doesn't say whether those dates are in american or english format but one of the dates is invalid if interpreted as american format) says the 30/08/2005 and 10/10/2006 for SP1/1a
after July 13, 2010 (next month) for XP SP2 Support ends 24 months after the next service pack releases or at the end of the product's support lifecycle, whichever comes first.
April 14, 2011 (next year) for XP SP3. If you have an enterprise contract/license, support for security fixes after April 8, 2014. Under SP3 the page you have linked says is
"Support ends 24 months after the next service pack releases or at the end of the product's support lifecycle, whichever comes first. For more information, please see the service pack policy at http://support.microsoft.com/lifecycle/#ServicePackSupport."
Assuming there isn't an XP SP4 that means we need to reffer to the lifecycle entry for the product as a whole which says 08/04/2014 for the end of extended support agreeing with your final date.
I cannot see anything on their corroberating your claim that security support will end sooner for those without enterprise licenses nor any mentions of the date you claim this will happen on.
A* is more complicated and varies by subject (it's NOT just another higher threshold on the overall mark for the subject). IIRC in maths it's an average of 90% on C3 and C4 but i'm not positive on that.
FWIW, my Mercedes diesel stationwagon uses about 5.5 L/100km for mixed city/rural driving, which is 42.8 mpg(US) or 51.4 mpg(UK). IIRC diesel has a higher energy density and higher price than petrol though so figures for petrol and diesel vehicles aren't directly comparable.
Don't assume that the only information available is only what's fed to you in TV ads. when trying to develop policies to make customers in general better informed about the efficiacy of thier purchases you have to realise that most customers aren't the smart informed type like you.
hence how the information is presented makes a HUGE difference to how the customer interprets it.
One thing I note about that system is that the stars are given based on efficiancy for it's size.
For many types of appliance this makes sense but with cars I worry it could give the impression that buying a bigger car to show off isn't such a bad idea.
I belive in some countries they use liters per 100km for measuring cars fuel efficiancy and this seems like a resonablly sensible system to me.
edit: I was wrong, it seems 2K and before had the autorun system completely disabled for removable media drives while XP enabled it for them to allow for the autoplay dialog.
IIRC the problem is the GUI options are per-drive.
This made some sense in the days when the feature was introduced when drives were generally things that didn't get added or removed very often but is something of a problem with USB sticks where the drive and the media are one device that is hot-plugged.
Remember USB sticks are a fairly new phenomenon. Yes they existed when XP was released but they were far from common.
with windows XP they did actually improve things for removable media drives (what most USB sticks enumerate as) by displaying the autoplay dialog rather than automatically performing the autorun action. But they didn't close down the autorun.inf's control of what happens when the user clicks the drives icon in windows explorer. Also some USB sticks enumerate as both a cdrom AND a removable media drive.
Disabling the autorun system completely is the safe answer but unfortunately there is no easy way to do this (regedit and gpedit.msc are the only ways i'm aware of).
Like I dunno, limited user accounts Limited user accounts have little to do with this unless they are VERY limited (far more limited than any linux system i'm aware of does by default).
and non-executable mounts? You don't need to go that far, just not running stuff without being explicitly told to would be sufficiant to block most of this sort of crap.
don't ask me how a UPS is a "human interface device"; but that's what it reported itself as.. This is a result of the way windows handles USB devices.
If you report as a custom USB device (that is a device not supporting any of the standard USB device classes) windows will go looking for a driver that matches your VID/PID and prompt the user if it doesn't find one. Sometimes that is desirable but sometimes you don't really want a driver, you just want your custom app to talk directly to your custom device.
OTOH if you report as a HID device windows will just load the standard HID driver. Your app can then use the functionality of that driver to talk to the device directly.
The BIOS really is a PC oddity. You don't see much like it in scope and complexity on other computer platforms. I disagree, afaict the likes of openfirmware (sun and newworld powermac) and EFI (itanium, intel mac and used in compatbility mode on some PC systems) are even greater in scope and complexity (for example they add an interepreted language for drivers in option roms rather than just running native code) albiet less crufty.
Often booting does as little as possible and then gets out of the way. When designing a computer system you have to put sufficiant code in rom or similar* to load the rest of the OS from some other medium. On a PCI based system that means:
1: enumerate all the PCI busses and give all the devices in the system addresses. 2: load driver code from somewhere (whether the main firmware or an "option rom" on the PCI device) to activate any device that may be the controller for the boot device 3: decide what device to get the next part of the system from and load and run code from it (possiblly providing that code with services of some form).
Most embeded linux systems i'm aware of take the route of putting the whole linux kernel and initrd in the onboard flash. This keeps things simple for the code doing the loading but also means you are limited in how many kernels you can keep arround at once and you are screwed bigtime if your kernels/initrds grow beyond a certain size.
*by or similar I mean stuff like paralell flash or serial flash that is read into ram by a tiny rom in the CPU.
A lot depends on the topology. Afaict one of the reasons that DSL was a commercial sucess was that it could be deployed gradually. When a customer orders DSL you just add a patch jumper at the exchange to the DSL gear. IIRC there were some places in the UK and germany (I dunno if any were in america) that had fiber to the cabinet and then equipment to convert that to pots/ISDN in the cabinet and afaict these were among the last places to get DSL because of the huge cost of upgrading all those cabinated.
Does anyone know how FTTH systems are generally being installed? are they taking every house direct to a big "telephone exchange" or are they concentrating things more locally?
Just got to make sure that fiber network is designed in such a way that it's easilly upgradable as newer tech comes along.
Many years ago (long before DSL) telcos in some places ran fiber to the cabinet and then pots/isdn from there to individual properties. This was touted as being the way of the future.
Then DSL came along and those areas were the last to get it because the decentralised system made upgrading a few customers at a time far more expensive (with a traditional exchange you can just put your DSL gear in a seperate rack and jumper the phone lines to it)
If I was designing a system like this I'd probablly do the conversion at the mail processing box and send something lower level (maybe PCL, maybe something even lowere level) to the printer.
Why put all that smarts in the printer where it will only be used occasionally when you can centralise it and therefore keep it busy all the time?
Most network printers afaict default to accepting print jobs and even adminitstration control from anyone who can directly connect to them. Usually this isn't too much of a problem because home users and small buisnesses are usually on NATed networks and larger companies hopefully have someone who knows what they are doing.
These printers OTOH presumablly connect outbound to some HP controlled server that accepts emails on thier behalf. That means if HP don't get this right they could be very vulerable to attack.
Note that a "performance index" like that is including a whole mix of tasks some more threaded than others but it would appear relatively few of them can take much advantage of more than 2 cores.
Which is great if that's what you have but I would guess that most people considering a 6-core chip would have a task in mind that is going to max the chip out.
Slashdot Mirror
The thing with lasers is that the precautions needed depend heavilly on the particular laser. Some lasers don't really need any precautions at all. Others will cause blindness unless handled very carefully.
We have the laser classification system which is supposed to tell users roughly how dangerous a laser is. The higher the class the higher the precautions needed from basically none at all for classes I and II to careful planning of where the beam will end up, special protective gear for those working with the laser and interlocks so that if the laser area is voilated things get cut off instantly for class IV
This development does concern me. Essentially they are marketing a class 4 laser as a "big boys toy". Yes there are warnings on their site but i'm not cinvinced their warnings are sufficiant.
Ubuntu seems to have decided that the best time to make risky decisions is the release immediately after a LTS because if it ends up sucking people can stay on the lts without any worries about support disappearing.
I'm a beliver that AJAX and the like have thier place but are often misused.
In particular i'm against anything that works against "deep linking" of material on the web whether that be framesets where if you link to an individual page the site becomes unnavigable. AJAX that pulls all content into the same page, deliberate anti deep linking measures or whatever. Linking whether manually or by search engines is the whole point of the web! /. for all it's web design faults is actually a good example of how you can use AJAX to make a more friendly discussion system while preserving the ability to link to individual posts.
The problem is when someone makes a statement like that you don't know if they are using the terminology correctly or incorrectly. This means that whatever assumption you make you stand a significant chance of being wrong.
Which is a good reason as an author to avoid that terminology altogether just like you should avoid ambiguous date formats.
Hmm though at least so far none of the big journals seem to be bothering with any form of DRM, everything just comes down as unprotected pdf (though some of the pdfs are watermarked so they can trace which institution was the source of a piracy incident).
Here's a hint: Microsoft is headquartered in Washington state.
Which is why I was somewhat surprised to see a date that only made sense in British format.
I guess they must be localising the pages or something.
Dates taken from the Windows XP Lifecycle page [microsoft.com].
hmm
April 14, 2009 (already in the past) for XP or XP SP1
The page you linked (unfortunately it doesn't say whether those dates are in american or english format but one of the dates is invalid if interpreted as american format) says the 30/08/2005 and 10/10/2006 for SP1/1a
after July 13, 2010 (next month) for XP SP2
Support ends 24 months after the next service pack releases or at the end of the product's support lifecycle, whichever comes first.
April 14, 2011 (next year) for XP SP3. If you have an enterprise contract/license, support for security fixes after April 8, 2014.
Under SP3 the page you have linked says is
"Support ends 24 months after the next service pack releases or at the end of the product's support lifecycle, whichever comes first. For more information, please see the service pack policy at http://support.microsoft.com/lifecycle/#ServicePackSupport ."
Assuming there isn't an XP SP4 that means we need to reffer to the lifecycle entry for the product as a whole which says 08/04/2014 for the end of extended support agreeing with your final date.
I cannot see anything on their corroberating your claim that security support will end sooner for those without enterprise licenses nor any mentions of the date you claim this will happen on.
Indeed A is 80% and above
A* is more complicated and varies by subject (it's NOT just another higher threshold on the overall mark for the subject). IIRC in maths it's an average of 90% on C3 and C4 but i'm not positive on that.
GCSEs have had A* for years, A levels have only recently aquired them.
FWIW, my Mercedes diesel stationwagon uses about 5.5 L/100km for mixed city/rural driving, which is 42.8 mpg(US) or 51.4 mpg(UK).
IIRC diesel has a higher energy density and higher price than petrol though so figures for petrol and diesel vehicles aren't directly comparable.
Don't assume that the only information available is only what's fed to you in TV ads.
when trying to develop policies to make customers in general better informed about the efficiacy of thier purchases you have to realise that most customers aren't the smart informed type like you.
hence how the information is presented makes a HUGE difference to how the customer interprets it.
One thing I note about that system is that the stars are given based on efficiancy for it's size.
For many types of appliance this makes sense but with cars I worry it could give the impression that buying a bigger car to show off isn't such a bad idea.
I belive in some countries they use liters per 100km for measuring cars fuel efficiancy and this seems like a resonablly sensible system to me.
edit: further for completely turning off autorun to be effective you must make sure you have a particular security update installed.
http://support.microsoft.com/kb/967715
the whole thing is a gigantic mess!
edit: I was wrong, it seems 2K and before had the autorun system completely disabled for removable media drives while XP enabled it for them to allow for the autoplay dialog.
IIRC the problem is the GUI options are per-drive.
This made some sense in the days when the feature was introduced when drives were generally things that didn't get added or removed very often but is something of a problem with USB sticks where the drive and the media are one device that is hot-plugged.
Remember USB sticks are a fairly new phenomenon. Yes they existed when XP was released but they were far from common.
with windows XP they did actually improve things for removable media drives (what most USB sticks enumerate as) by displaying the autoplay dialog rather than automatically performing the autorun action. But they didn't close down the autorun.inf's control of what happens when the user clicks the drives icon in windows explorer. Also some USB sticks enumerate as both a cdrom AND a removable media drive.
Disabling the autorun system completely is the safe answer but unfortunately there is no easy way to do this (regedit and gpedit.msc are the only ways i'm aware of).
Like I dunno, limited user accounts
Limited user accounts have little to do with this unless they are VERY limited (far more limited than any linux system i'm aware of does by default).
and non-executable mounts?
You don't need to go that far, just not running stuff without being explicitly told to would be sufficiant to block most of this sort of crap.
don't ask me how a UPS is a "human interface device"; but that's what it reported itself as..
This is a result of the way windows handles USB devices.
If you report as a custom USB device (that is a device not supporting any of the standard USB device classes) windows will go looking for a driver that matches your VID/PID and prompt the user if it doesn't find one. Sometimes that is desirable but sometimes you don't really want a driver, you just want your custom app to talk directly to your custom device.
OTOH if you report as a HID device windows will just load the standard HID driver. Your app can then use the functionality of that driver to talk to the device directly.
The BIOS really is a PC oddity. You don't see much like it in scope and complexity on other computer platforms.
I disagree, afaict the likes of openfirmware (sun and newworld powermac) and EFI (itanium, intel mac and used in compatbility mode on some PC systems) are even greater in scope and complexity (for example they add an interepreted language for drivers in option roms rather than just running native code) albiet less crufty.
Often booting does as little as possible and then gets out of the way.
When designing a computer system you have to put sufficiant code in rom or similar* to load the rest of the OS from some other medium. On a PCI based system that means:
1: enumerate all the PCI busses and give all the devices in the system addresses.
2: load driver code from somewhere (whether the main firmware or an "option rom" on the PCI device) to activate any device that may be the controller for the boot device
3: decide what device to get the next part of the system from and load and run code from it (possiblly providing that code with services of some form).
Most embeded linux systems i'm aware of take the route of putting the whole linux kernel and initrd in the onboard flash. This keeps things simple for the code doing the loading but also means you are limited in how many kernels you can keep arround at once and you are screwed bigtime if your kernels/initrds grow beyond a certain size.
*by or similar I mean stuff like paralell flash or serial flash that is read into ram by a tiny rom in the CPU.
A lot depends on the topology. Afaict one of the reasons that DSL was a commercial sucess was that it could be deployed gradually. When a customer orders DSL you just add a patch jumper at the exchange to the DSL gear. IIRC there were some places in the UK and germany (I dunno if any were in america) that had fiber to the cabinet and then equipment to convert that to pots/ISDN in the cabinet and afaict these were among the last places to get DSL because of the huge cost of upgrading all those cabinated.
Does anyone know how FTTH systems are generally being installed? are they taking every house direct to a big "telephone exchange" or are they concentrating things more locally?
Just got to make sure that fiber network is designed in such a way that it's easilly upgradable as newer tech comes along.
Many years ago (long before DSL) telcos in some places ran fiber to the cabinet and then pots/isdn from there to individual properties. This was touted as being the way of the future.
Then DSL came along and those areas were the last to get it because the decentralised system made upgrading a few customers at a time far more expensive (with a traditional exchange you can just put your DSL gear in a seperate rack and jumper the phone lines to it)
It really depends on what functionality you want. If you just want to print plain text emails it probablly would be that easy.
If you want to print attatchments it gets a bit harder since you need to identify and convert them.
If I was designing a system like this I'd probablly do the conversion at the mail processing box and send something lower level (maybe PCL, maybe something even lowere level) to the printer.
Why put all that smarts in the printer where it will only be used occasionally when you can centralise it and therefore keep it busy all the time?
Most network printers afaict default to accepting print jobs and even adminitstration control from anyone who can directly connect to them. Usually this isn't too much of a problem because home users and small buisnesses are usually on NATed networks and larger companies hopefully have someone who knows what they are doing.
These printers OTOH presumablly connect outbound to some HP controlled server that accepts emails on thier behalf. That means if HP don't get this right they could be very vulerable to attack.
Sorry that should have said more than 4 cores. Though even between 2 and 4 the difference is a lot less than a doubling.
Note that a "performance index" like that is including a whole mix of tasks some more threaded than others but it would appear relatively few of them can take much advantage of more than 2 cores.
Which is great if that's what you have but I would guess that most people considering a 6-core chip would have a task in mind that is going to max the chip out.