IMO length is not the only important thing in making a url easy to type
For your example I would suggest something like
classes/maths/college-algebra/chapter-1/1-5.xml
Same basic structure A little shorter but far less annoying to type due to the elimination of pointless capitals and the consistant use of dashes where words need to be broken within a part.
I'm not convinced being short is that much of an advantage. Typing english words is not that slow and a sequence of english words that makes sense is likely much easier to remember than some meaninless alphanumeric code.
What is bad for usability is including lots of irrelevent (or only relavent for tracking users habbits) crap in a url. e.g.
The character encodings are irrelevent since the search terms as ascii. The source of the query is only relevent for tracking what tools people are using to search it doesn't provide any utility to the user.
We geeks can probablly work out how to trim that down to a more minimal url that works as needed ( http://www.google.com/search?q=BS+EN+60238 ) but ordinary people would just see a wall of meaningless text with thier query burried in it somewhere.
C, on the other hand, is a miserable language for almost everything except system-level (kernel, driver) hacking. Its widespread use for many other functions is a stupid fashion. It is annoying that such a crappy language became the de-facto standard. Still unfortunately "network effects" make it the most sensible choice for many applications.
1: C is probablly the most portable language arround. There are C compilers for everything from tiny pics up to the biggest mainframes. On some platforms it's the only high level language with even a half-decent compiler. 2: there are lots of libraries designed for use from C. It's certainly possible to use them from other languages but it can be a PITA and may well require either rewriting a load of macros or creating your own C shim layer. Then you get a new version of the library with a compatible C API but an incompatible ABI and have to re-do your custom imports to match (or worse not realise they need changing and end up with strange bugs) 3: On linux at least the standard libraries for other langauges (particularlly C++) have traditionally had far worse ABI breakage issues than the C standard library.
most of them come formatted with patent encumbered [earthweb.com] FAT16/32 file system. nitpick: it's not the filesystem itself that is patented but the hack to add long filenames.
So even if linux is forced to pull support for the patented features you should still be able to access the files using thier 8.3 aliases.
That looks like it would be great at protecting against accidents in storage/transportation but not so great at protecting against accidents in use.
Personally i've found the way I damage USB sticks is to bend them while they are sticking out the front of a computer (one time by kicking it, the other by rolling the computer forward to get it out for installing an upgrade) the other time by. Lukilly the ones i've had this happen with have worked long enough after the damage to allow for recovery (one of them is still working).
The default on XP and later is to reboot automatically, sometimes the BSOD seems to pop up briefly but either it doesn't always or I often miss it, not sure which (whether you see it briefly may may also depend on how quick your monitor can switch to text mode)
Afaict there are a few common causes of BSODs/autoreboots.
Shitty software (while userland software is not supposed to be able to cause BSODs a lot of software ships with drivers of some sort) Shitty hardware drivers Shitty hardware (particularlly dodgy ram, I'd reccomend running a memory test as one of the first diagnostics on a machine giving a lot of BSODs)
How many home/small buisness users upgrade windows anyway? The fact is such users generally get the new version of windows when they get a new PC. Big corp/institutional users tend to deploy new versions by reimaing everything with thier new standard image. So it seems the proportion of users needing an upgrade install would be fairly small.
a 95 upgrade CD will do a clean install as long as you give it the 3.1 install floppy when it asks for it.
What is more fun is convincing a windows 95 OEM cd to do an upgrade (say to upgrade a machine to OSR2 from 95 original release), afaict to do that you have to delete win.com first.
The incandescent in my bedroom light has been replaced twice in 6 years. I put in a 100 watt bulb and dim it 75% and this one is lasting almost as long as the CFl's. Be aware that when dimming an incandescent the light output goes down much faster than the power consumption.
EU is making incandescents illegal (to sell). IIRC they are only banning conventional incandescents. Halogens are to remain for the moment (though it wouldn't surprise me if they target those before too long)
Some lighting manufacturers are now making halogens in a form to directly replace ordinary incandescents though they aren't widely availible yet.
Can you elaborate on that? Electric heat is virtually 100% efficient Yes if you only look at the end part of the chain.
OTOH if you compare the efficiancy of fuel based heating to the efficiancy of powerplant+electrical distribution network+resistive electric heater then resisitive electric heating loses big time.
Heat pumps on the other hand can get efficiancies over 100% (since they pump in heat from outside) which offsets the inefficiancies in the generation chain.
Still, as demonstrated on the Myth Busters a while back, LED bulbs can be turned on and off many more times than other kind of bulb. Hardly surprising but few bulbs used for general lighting are switched anywhere near as often as those on that mythbusters rig.
Still lifetime is an advantage of LEDs in some applications but it must be balanced against the very high upfront costs. Also for consumers it is difficult to get good lifetime information in advance.
One thing to watch for with both flouresecents and LEDs is that thier light output tends to go down as they age. They will drop below 50% of thier original brightness LONG before they actually blow.
using near UV emitting LEDs in a similar fashion that you described, which results in a better color spectrum but with the risk of emitting UV light if there's a manufacturing flaw. This is how the better (in terms of spectral quality) white LEDs (the ones that are actually reasonable for accomodation lighting rather than torches etc) are made. Unfortunately the result is an efficiancy only barely better than flourescent lighting and a FAR higher capital cost.
As you are probablly aware mains is an AC system. We can assume that the mains waveform is a sinewave (not perfect but a pretty good approximation)
First lets consider the case where the load is purely resistive. * instantanious current is instantanious voltage devided by resistance * Therefore instantanious power is the instantanious voltage squared then divided by the resistance. * Therefore average power is the RMS voltage squared then divided by the resistance.
However for many real devices current is not directly proportional to voltage and as such the current waveform may be either out of phase with the voltage waveform, have harmonics or both.
It turns out (I won't do the math to prove it here but it's not too hard) only the component of the fundamental frequency of the current waveform that is in phase with the voltage waveform results in a net transfer of power. The component of the fundamental frequency of the current waveform that is in quadrature phase with the voltage waveform and any components of the current waveform at higher frequencies simply result in power oscilating between source and load.
However while those components don't represent transfer of power to the load they DO create volt drop and power loss in the wiring. Furthermore those currents have to come from somewhere (either the generators or power factor correction)
So we have "real power" which is the net power transfered to the load and "apparent power" which is the product of RMS current and RMS voltage. For a purely resistive load theese are the same but for other types of load they often aren't. (there is also a related term called reactive power but I won't bother to go into it here)
Power factor is "apparent power" divided by "real power"
Residential customers are metered and billed by thier "real power" consumption but utilities have to plan thier distribution arround the "apparent power" consumption. So low power factor is a PITA for utilites.
e.g. with 3 you can get broadband at £15 per moth for 15GB per month (which is more than most terrestrial broadband but tollerable IMO) but the overage rate is 10p per MB (which works out to £100 per GB which is IMO insanely expensive)
This means that any user of a mobile broadband contract has to be EXTREMELY carefull to keep an eye on thier usage.
It's not a true captcha IMO because test generation is not automated. That means anyone with sufficiant resources can compile a "test database" that allows thier bots to pass every time.
Still it's probablly effective for small sites provided a custom question list is used.
For small sites the best strategy is generally to do something *different* from everyone else. What that something is doesn't matter too much as long as it's different and you are small enough not to be an explicit target.
For larger sites things look less rosy, recpatcha may be an option but i'm not sure how well it will stand up to a combination of OCR and brute force (keep trying new recaptchas until you get one right)
if you remember the changes from 2 years ago where the summer time changed a few weeks, many company discovered that "extended support" didn't cover that sorta thing and had to pay many thousand $ for the patch. I'm a european so I avoided the american DST fiasco but IIRC MS was actually quite nice about that one. While you did have to pay for the hotfix they also told you how to fix the issue manually and some people produced third party patchers.
HP seem to be just plain weird in thier handling of this, they make the XP option the default option in the selector on thier small buisness site yet afaict they will only sell you machines with XP installed under the following conditions
"To qualify for this downgrade an end user must be a business (including governmental or educational institutions) and is expected to order annually at least 25 customer systems with the same custom image." unlike dell who make vista the default will happilly sell you XP on one off orders.
XP will have security fixes until 2014, placing it at around 13 years. Almost double of RHEL. True but it only has that extremly long lifecycle because of huge delays to vista (which caused the "2 years past the successor product" part of the lifecycle policy to be invoked).
IMO what matters more is not so much the total length of the lifecycle as the level of overlap between one release and the next. The windows lifecycle policy states that MS will provide at least a 7 year overlap (2 years mainstream, 5 extended).
RHEL seems to release roughly every 2 years and doesn't seem to have any specific provisions for overlap. So that gives an overlap of roughly 5 years though it will vary from release to release.
good point, just the other day someone from MS was saying how they've taken over 95% of the netbook market, and last I looked, those little netbooks weren't running Vista. True but the version of windows on those netbooks will still be getting security updates for longer than the linux distros that are availible on those same netbooks (yes I know a couple of the big enterprise linux distros have support lifecycles only slightly worse than those from MS but the enterprise distros are NOT what is shipping on netbooks).
Does this mean MSFT engineers will no longer "talk users through" the downgrade process. I suspect they will still provide activation for downgrades performed using existing retail/system builder media as that is probablly considered to be activation service rather than support.
If you want help getting your XP downgrade to work with your hardware you will have to either get that from your OEM or pay for it from MS.
Slashdot Mirror
IMO length is not the only important thing in making a url easy to type
For your example I would suggest something like
classes/maths/college-algebra/chapter-1/1-5.xml
Same basic structure A little shorter but far less annoying to type due to the elimination of pointless capitals and the consistant use of dashes where words need to be broken within a part.
I'm not convinced being short is that much of an advantage. Typing english words is not that slow and a sequence of english words that makes sense is likely much easier to remember than some meaninless alphanumeric code.
What is bad for usability is including lots of irrelevent (or only relavent for tracking users habbits) crap in a url. e.g.
http://www.google.com/search?q=BS+EN+60238&ie=utf-8&oe=utf-8&aq=t&rls=org.debian:en-GB:unofficial&client=iceweasel-a
The character encodings are irrelevent since the search terms as ascii. The source of the query is only relevent for tracking what tools people are using to search it doesn't provide any utility to the user.
We geeks can probablly work out how to trim that down to a more minimal url that works as needed ( http://www.google.com/search?q=BS+EN+60238 ) but ordinary people would just see a wall of meaningless text with thier query burried in it somewhere.
C, on the other hand, is a miserable language for almost everything except system-level (kernel, driver) hacking. Its widespread use for many other functions is a stupid fashion.
It is annoying that such a crappy language became the de-facto standard. Still unfortunately "network effects" make it the most sensible choice for many applications.
1: C is probablly the most portable language arround. There are C compilers for everything from tiny pics up to the biggest mainframes. On some platforms it's the only high level language with even a half-decent compiler.
2: there are lots of libraries designed for use from C. It's certainly possible to use them from other languages but it can be a PITA and may well require either rewriting a load of macros or creating your own C shim layer. Then you get a new version of the library with a compatible C API but an incompatible ABI and have to re-do your custom imports to match (or worse not realise they need changing and end up with strange bugs)
3: On linux at least the standard libraries for other langauges (particularlly C++) have traditionally had far worse ABI breakage issues than the C standard library.
most of them come formatted with patent encumbered [earthweb.com] FAT16/32 file system.
nitpick: it's not the filesystem itself that is patented but the hack to add long filenames.
So even if linux is forced to pull support for the patented features you should still be able to access the files using thier 8.3 aliases.
That looks like it would be great at protecting against accidents in storage/transportation but not so great at protecting against accidents in use.
Personally i've found the way I damage USB sticks is to bend them while they are sticking out the front of a computer (one time by kicking it, the other by rolling the computer forward to get it out for installing an upgrade) the other time by. Lukilly the ones i've had this happen with have worked long enough after the damage to allow for recovery (one of them is still working).
The default on XP and later is to reboot automatically, sometimes the BSOD seems to pop up briefly but either it doesn't always or I often miss it, not sure which (whether you see it briefly may may also depend on how quick your monitor can switch to text mode)
Afaict there are a few common causes of BSODs/autoreboots.
Shitty software (while userland software is not supposed to be able to cause BSODs a lot of software ships with drivers of some sort)
Shitty hardware drivers
Shitty hardware (particularlly dodgy ram, I'd reccomend running a memory test as one of the first diagnostics on a machine giving a lot of BSODs)
How many home/small buisness users upgrade windows anyway? The fact is such users generally get the new version of windows when they get a new PC. Big corp/institutional users tend to deploy new versions by reimaing everything with thier new standard image. So it seems the proportion of users needing an upgrade install would be fairly small.
a 95 upgrade CD will do a clean install as long as you give it the 3.1 install floppy when it asks for it.
What is more fun is convincing a windows 95 OEM cd to do an upgrade (say to upgrade a machine to OSR2 from 95 original release), afaict to do that you have to delete win.com first.
I would assume so, presumablly you would have to work along the cable fixing each cut and then taking a new measurement.
Quite a good way to inflict a lot of financial damage on a telco really. Afaict fixing fibers is FAR more expensive than cutting them.
The incandescent in my bedroom light has been replaced twice in 6 years. I put in a 100 watt bulb and dim it 75% and this one is lasting almost as long as the CFl's.
Be aware that when dimming an incandescent the light output goes down much faster than the power consumption.
EU is making incandescents illegal (to sell).
IIRC they are only banning conventional incandescents. Halogens are to remain for the moment (though it wouldn't surprise me if they target those before too long)
Some lighting manufacturers are now making halogens in a form to directly replace ordinary incandescents though they aren't widely availible yet.
Can you elaborate on that? Electric heat is virtually 100% efficient
Yes if you only look at the end part of the chain.
OTOH if you compare the efficiancy of fuel based heating to the efficiancy of powerplant+electrical distribution network+resistive electric heater then resisitive electric heating loses big time.
Heat pumps on the other hand can get efficiancies over 100% (since they pump in heat from outside) which offsets the inefficiancies in the generation chain.
Still, as demonstrated on the Myth Busters a while back, LED bulbs can be turned on and off many more times than other kind of bulb.
Hardly surprising but few bulbs used for general lighting are switched anywhere near as often as those on that mythbusters rig.
Still lifetime is an advantage of LEDs in some applications but it must be balanced against the very high upfront costs. Also for consumers it is difficult to get good lifetime information in advance.
One thing to watch for with both flouresecents and LEDs is that thier light output tends to go down as they age. They will drop below 50% of thier original brightness LONG before they actually blow.
using near UV emitting LEDs in a similar fashion that you described, which results in a better color spectrum but with the risk of emitting UV light if there's a manufacturing flaw.
This is how the better (in terms of spectral quality) white LEDs (the ones that are actually reasonable for accomodation lighting rather than torches etc) are made. Unfortunately the result is an efficiancy only barely better than flourescent lighting and a FAR higher capital cost.
As you are probablly aware mains is an AC system. We can assume that the mains waveform is a sinewave (not perfect but a pretty good approximation)
First lets consider the case where the load is purely resistive.
* instantanious current is instantanious voltage devided by resistance
* Therefore instantanious power is the instantanious voltage squared then divided by the resistance.
* Therefore average power is the RMS voltage squared then divided by the resistance.
However for many real devices current is not directly proportional to voltage and as such the current waveform may be either out of phase with the voltage waveform, have harmonics or both.
It turns out (I won't do the math to prove it here but it's not too hard) only the component of the fundamental frequency of the current waveform that is in phase with the voltage waveform results in a net transfer of power. The component of the fundamental frequency of the current waveform that is in quadrature phase with the voltage waveform and any components of the current waveform at higher frequencies simply result in power oscilating between source and load.
However while those components don't represent transfer of power to the load they DO create volt drop and power loss in the wiring. Furthermore those currents have to come from somewhere (either the generators or power factor correction)
So we have "real power" which is the net power transfered to the load and "apparent power" which is the product of RMS current and RMS voltage. For a purely resistive load theese are the same but for other types of load they often aren't. (there is also a related term called reactive power but I won't bother to go into it here)
Power factor is "apparent power" divided by "real power"
Residential customers are metered and billed by thier "real power" consumption but utilities have to plan thier distribution arround the "apparent power" consumption. So low power factor is a PITA for utilites.
Another issue is insane overage rates.
e.g. with 3 you can get broadband at £15 per moth for 15GB per month (which is more than most terrestrial broadband but tollerable IMO) but the overage rate is 10p per MB (which works out to £100 per GB which is IMO insanely expensive)
This means that any user of a mobile broadband contract has to be EXTREMELY carefull to keep an eye on thier usage.
So just what is the price of a botnet node with a unique public IP address on the black market?
You don't think spammers buy thier IPs legitimately do you?
It's not a true captcha IMO because test generation is not automated. That means anyone with sufficiant resources can compile a "test database" that allows thier bots to pass every time.
Still it's probablly effective for small sites provided a custom question list is used.
not impossible but you would have to convince users to install it. Effectively it would be a system that discriminated against smart users.
And how do you plan to use those license codes to get non-secuirty hotfixes released after XP goes into extended support?
Depends on the size of the site.
For small sites the best strategy is generally to do something *different* from everyone else. What that something is doesn't matter too much as long as it's different and you are small enough not to be an explicit target.
For larger sites things look less rosy, recpatcha may be an option but i'm not sure how well it will stand up to a combination of OCR and brute force (keep trying new recaptchas until you get one right)
if you remember the changes from 2 years ago where the summer time changed a few weeks, many company discovered that "extended support" didn't cover that sorta thing and had to pay many thousand $ for the patch.
I'm a european so I avoided the american DST fiasco but IIRC MS was actually quite nice about that one. While you did have to pay for the hotfix they also told you how to fix the issue manually and some people produced third party patchers.
HP seem to be just plain weird in thier handling of this, they make the XP option the default option in the selector on thier small buisness site yet afaict they will only sell you machines with XP installed under the following conditions
"To qualify for this downgrade an end user must be a business (including governmental or educational institutions) and is expected to order annually at least 25 customer systems with the same custom image." unlike dell who make vista the default will happilly sell you XP on one off orders.
XP will have security fixes until 2014, placing it at around 13 years. Almost double of RHEL.
True but it only has that extremly long lifecycle because of huge delays to vista (which caused the "2 years past the successor product" part of the lifecycle policy to be invoked).
IMO what matters more is not so much the total length of the lifecycle as the level of overlap between one release and the next. The windows lifecycle policy states that MS will provide at least a 7 year overlap (2 years mainstream, 5 extended).
RHEL seems to release roughly every 2 years and doesn't seem to have any specific provisions for overlap. So that gives an overlap of roughly 5 years though it will vary from release to release.
good point, just the other day someone from MS was saying how they've taken over 95% of the netbook market, and last I looked, those little netbooks weren't running Vista.
True but the version of windows on those netbooks will still be getting security updates for longer than the linux distros that are availible on those same netbooks (yes I know a couple of the big enterprise linux distros have support lifecycles only slightly worse than those from MS but the enterprise distros are NOT what is shipping on netbooks).
Does this mean MSFT engineers will no longer "talk users through" the downgrade process.
I suspect they will still provide activation for downgrades performed using existing retail/system builder media as that is probablly considered to be activation service rather than support.
If you want help getting your XP downgrade to work with your hardware you will have to either get that from your OEM or pay for it from MS.