For the record, the back end uses the Amazon API PHP library and has boilerplate code to set it up, including the super sekrit API keys that I can't include in an open source client
hmm, but would it have been better to include a placeholder config where the user puts their own super secret Amazon keys in the front end and runs it entirely for themselves?
I agree that the petition is going to go nowhere too but the way to fix it is only partly come up with an alternative.
We need to ensure IT grunts in big corporations know about UEFI and that it is bad, so when the boss signs off on the budget for new kit, they can go to Dell and say "we'll have 10000 of those, but we will reject them if they come with UEFI and we'll see what HP has to offer instead" (or vice versa), these guys can also tell the boss that they can't use UEFI in their organisation as it means they won't be able to reallocate the machines in the future (meaning - reallocate them as linux machines; use your own wording).
We also need the companies who might sign this petition also know to write to Dell and tell him UEFI is crap and he shouldn't entertain boards with it, and in turn to tell the mobo manufacturers that UEFI chipset boards will not sell. If Google said "we won't be buying Dell any more due to boards running UEFI" publicly, you can guarantee Dell stock would drop quite a bit and the execs at Dell would sit up and take a lot of notice.
All in all, the only way to really defeat this is via the money. If UEFI boards meant no sales, the whole thing would become a non-issue in no time.
It's like if Balmer dragged a programmer up on stage and proceeded to flog the crap out of him
you mean, threw a chair at him.
I believe he did something like that, though for far less reason, and now has notoriety for 3 things: throwing chairs, dancing like a monkey shouting 'developers', and leading Microsoft through a decade of no share price growth.
Linus will get a reputation for being belligerent and rude... oh wait!
"many eyes" making the code work seem to be obsolete when you have a club that you have to be a member of before they take notice of you. Its always been this way in so many fields of human endeavour - if you're expecting the community around Linux to be different and based on technical excellence only then you have a lot to learn about people.
The trick is to acknowledge that you're not "one of them" and approach the task of submitting your patch like this - submit the patch as normal, and then contact Linus or someone with an email humbly telling them that you've noticed a little bug that slipped through and that you've submitted a patch that you think will fix the problem - once you have their attention enough to look at the fix, they'll be ready to accept it, as if it was something they'd noticed themselves. I guess that once you've done this a couple of times, they'd accept your future patches without review.
I know its not supposed to work that way, but it does... I'd be interested to know how you went about the submission process. It could be that they were just too busy to review all the patches that were submitted, so yours languished, un-noticed for so long.
actually, you're probably paying way more tax that you're required to pay - as seen by the recent scandals in the UK where various celebrities simply pay their money into an offshore account owned by a privately-held company and then take out a loan from said company, thus meaning their income is roughly 0, and therefore they don't have any tax to pay.
See, these schemes are quite legal, and the celebrities involved weren't required to pay any tax on an income on nothing, so why do you pay tax?
'course, said schemes are incredibly dodgy and caused a lot of backlash from the public who do see tax as a necessary evil, and rich people being able to scam their way to not paying anything as an even greater evil. The only real solution is to simplify the tax laws considerably so clever accountants cannot come up with these workarounds and loopholes. Oh, and to refuse to recognise the tax status of countries that have 0% tax systems, or to make companies that do "set up shop" (usually a post-office box) in these countries have a certain percentage of their workforce be employed there.
yes, I remember those days. I used to code OS/2 (it was a long time ago) and I had 2 binders full of everything you could ever want to know about the APIs available to me. When I had an Amiga you could buy 3 rather large books with everything, down to wiring and chipset information. Even Windows NT (3.51) had a set of 5 books in hardbound covers that told you everything.
Today, the closest you get to support is:
"hello my name is Vijay how many I help you. I see what the problem might be being, please try to turn your device off and then on again and call back if the problem remains, now please tell me how well this support information has been to you today"... and when you call back, you get someone else giving you the same 'advice' and asking to be rated in helpfulness for a little bonus.
Languages like.NET allocate a massive amount of memory - objects are new'ed all over the place, and internally even more so.
Take a look at the GC allocations counters in perfmon to see just how "few" allocations are actually done, you'll be surprised. Then take a look at the GC collection counters for an even bigger surprise.
Locality of reference is the huge thing today - my 3ghz CPU only runs that fast if it can read memory, and it can only read at that speed from L1 cache (which is quite tiny). Sure they have MBs of cache at L2 or L3, but that's still slow, just faster than main RAM. So unless your app can fit into the L1 cache, its going to run slowly (relatively), and I mean slowly like its swapping (as it will be - swapping to the other caches or main RAM).
The other aspect of a GC language is that it makes it easy for allocation to occur, so all these languages (which admittedly might be due to other factors, not the GC) tend to use a lot more ram than other types of system. More RAM = less that can fit in that cache = slower.
Letting the GC fill up a gig of ram just means you have to wait a little longer before you see it stall, which won't be very long on today's systems that gobble up a gig in no time. Then it just has more work to do compacting the large heap - which is why one performance fix to GCs was generational collection - ie running the GC on a smaller set of the heap. This is why you don't see the GC pauses you used to on old Java programs.
deallocation is only basically free when you don't do it. Otherwise, you have to compact the heap - which is most certainly not free - or you'll either run out of memory, or you'll have to start allocating using a next-free-block algorithm, just like malloc does.
I hear this a lot - that it is faster, and my observations ( of a fair few GC-based language systems) are that is is tosh.
It is faster to allocate - any system that doesn't care about memory usage is, and given an infinite block of memory that you always allocate from the end is super fast. The problem comes when you have to compact - this is a pretty slow operation. You'd never think about writing an app that copied large blocks of memory about every second, yet this is exactly what the GC does once your app gets going. De-allocation is also slower that most other systems as it has to consider finalisation (which, as you noted is shite) but it is necessary - you cannot have the system deallocate an object while it contains other resources that require freeing, and you cannot have a high-level language that mandates you to manually free those things (C does that, its a fine language for its purposes, but its not considered as high a level as you want).
Then locality of reference, this is also mostly meaningless. I know you want memory blocks to be contiguous (imagine a string where each character was in wildly different memory locations), but compare what the GC gives you compared to the C style of allocating a block of memory and filling it with data - in a GC system you can allocate memory all over the place, and the GC will compact it to a point, but those blocks will never be exactly next to each other - which is what you want for the performance benefits. If you have a c style array, or a c++ vector, then you do get this kind of performance gain (at a cost of having to copy the memory into the pre-allocated buffer, but that's still no slower than the GC copying the same data when it does a compaction).
So think about what the GC really gives you - the ability to pretend you don't care about memory.
I'd say they made a huge mistake when designing GC systems, in that they always make it work with memory, but then apply it to objects. Object lifecycle is different to memory, you want objects to disappear when they're no longer used, only in some cases can you tell the system to destroy the object when it feels like it, but in most cases you want the object to clean itself up properly. There's no real reason why they couldn't have put RAII style semantics in (ie by making the system generate automatic calls to a destructor - like Microsoft now does with the using() syntax) and let the GC manage the memory allocation only.
But still, for performance, a GC is nowhere near as fast as a deterministic program. Just look at the performance counters in a.NET app to see the GC going bonkers. I've seen it compacting the heap hundreds of times a second in a decently written large app.
so what you're saying is that when there's a fuckup in the code that causes a big problem, you'd rather it happened at random rather than in a deterministic place where you can see it occur, easily see what the problem is and fix it.
There are plenty of cases where GC is a joke, not being able to clean up your code until it feels like it doesn't make for better or more reliable systems. (an example, I have a WCF service, it uses log4net to log stuff, and I use its property system to add the caller's ip. But where there's an exception, the pop doesn;t occur in the service Dispose method because the framework has decided to hang on to it and free it up later. So the next log gets 2 ip addresses logged.. So what would be nice clean code in C++ becomes a mess of manually handling Exception exceptions in order to bypass the system so it works). GC is for lazy-ass programmers who don't know better.. RAII is for lazyass programmers who know and love what it does for them.
However, all Stallman had to offer on the topic was "We still prefer C to C++, because C++ is so ugly" (sic).
well, score one for technical argument and intellectual reasoning.
If ugliness was a consideration, then surely GNU should be moving to a beautiful language, such as C#. If no, then WTF does subjective qualities of a language have to do with anything, there's no problem with any language as long as it is understandable (and I contend that means individual coding style rather than anything in-built - C can be so obfuscated they have competitions for it!).
the problem with all of this is the usual geek tendency of arrogance - where one always thinks he has the mental superiority to everyone else, and this is the biggest single detriment to the OSS community of all.
possibly, the trouble is you get a lego set that has all the special pieces for the puzzle that you get in the box. Its not like my day when a curved piece of (I suppose pavement) was a fancy piece that became the edge of the gun emplacement on my space-battleship.
Today, lego kits are really just big 3d jigsaw puzzles for 20/30 somethings who used to have lego when they were kids.
Lego needs to make big boxes of generic bricks cheaply available, though you're better off looking for these in garage sales.
however, that just masks the problem, doesn't solve it.
With the current Firefoxes, you can go to the help menu and see what tabs are using what memory from the 'troubleshooting information' item. then you can see a rogue tab has gobbled all your ram, and close it. That's a solution.
fuckwittery is like obscenity - I know it when I see it.
and when I find such a fuckwittery in practice, my tactic on resolving it depends on the quality of the fuckwit - if he's my peer, I point it out and expect him to realise it himself, if he's my junior then he gets a code review session. If he's my boss, I offer to polish the bits his greatness didn't have time to complete himself:)
that's becuase it isn't a design suite really. Its a developer tool that they thought - either through greed or naievety - that they could sell twice.
Once upon a time, all the developer tooling was inside visual Studio, so a dev could knock up a GUI and hook it into the code he'd written. Then Microsoft invented XAML and next thing you know visual studio's XAML editing capabilities were extremely poor (ie often would crash if you tried to do anything remotely complicated, couldn't review what the GUI looked like if you edited the underlying xaml code, and could barely design the gui without hand-edited bits anyway) but its was ok as they released a separate GUI design tool you could use alongside VS.
the idea was that your designer could write XAML and then give it to the coder who would write the code and tie the two together... shame that, in true Microsoft fashion, separation of concerns like this is always broken by the need to "integrate" everything, so pretty soon you had to write code ("code behind") in your XAML to make it do anything useful, like binding a variable to a GUI element - see, simples!
So all in all, I can see why they wanted to put this stuff back into Visual Studio - its easier for them, and it means your designer will have to buy a copy of VS instead.
the only coding standard that is required there is the one that says:
Code MUST not be written like you are a fuckwit.
Even without a standard, I'd look at that and either gently rebuke the coder (who I hope would agree that it was bad and needed formatting - we all do stuff like this occasionally), or would just reformat it to be readable. No drama, problem solved.
The problem with the standard is that it refuses the places where the: "if (foo) bar(); " style is appropriate - eg. times when you have a large block of repetitive code like this, where formatting it onto 2 (or 4 with bracket placement demands) would make it unreadable.
All in all I find that style matters very little. So what if my colleague writes underscores between names in variables, and another uses camel case.. as long as everyone uses the same style within the file, I'm happy (actually, happier than a stupid standard that you have to refer to to keep your code from failing the automated style nazis)
However, the parts of a standard that really do matter are the bits no-one seems to care about (when they're all arguing about the trivialities of bracket placement): project layout. Its no good having a file with a certain style of naming convention if I cannot find the flipping file that has the code I want to look at - we've all seen funny layouts with bits of file here and there, not to mention libraries or other dependencies tucked away in a sub-sub-folder, or documentation that is a text file hidden away somewhere.
The thing that really annoys me is that 'professional' coders spend so much time arguing the toss about style that they miss the stuff that matters. Maybe its because I have a foot in the release management side of things that this matters more to me, or maybe I just like things to be neat.
Style.. well, no-one can read your code even with a style-based standard as they will lay classes and namespaces differently, and their functions will always require investigation - so it realy doesn't save any time.A standard that says to put all your private variables at the end of the class, that might make a difference compared to say scattering them throughout the class - but who would do that, that would respond to a slap anyway.
It's not like they own Phonegap for creating naive Mobile Apps with HTML5 and JS,/i>
you're referring to Apache Cordoba right?
Kudos to Adobe for giving it to the Apache Foundation though, they get a lot of bad press - mainly because of crappy Flash-based adverts - but they deserve a bit better. Only a bit mind, if they release a version of Flash without security flaws or Reader that opens quickly without a mass of 'features'.
you're also not missing the benchmarks comparing the performance to a Windows box running the same benchmarks (where applicable) or the benchmarks showing the proprietary driver's performance.
Slashdot Mirror
For the record, the back end uses the Amazon API PHP library and has boilerplate code to set it up, including the super sekrit API keys that I can't include in an open source client
hmm, but would it have been better to include a placeholder config where the user puts their own super secret Amazon keys in the front end and runs it entirely for themselves?
free markets are good - which is why we don't allow monopolies to distort them.
Besides, he might change his mind about not caring about technology after he starts using Windows 8 :-)
I agree that the petition is going to go nowhere too but the way to fix it is only partly come up with an alternative.
We need to ensure IT grunts in big corporations know about UEFI and that it is bad, so when the boss signs off on the budget for new kit, they can go to Dell and say "we'll have 10000 of those, but we will reject them if they come with UEFI and we'll see what HP has to offer instead" (or vice versa), these guys can also tell the boss that they can't use UEFI in their organisation as it means they won't be able to reallocate the machines in the future (meaning - reallocate them as linux machines; use your own wording).
We also need the companies who might sign this petition also know to write to Dell and tell him UEFI is crap and he shouldn't entertain boards with it, and in turn to tell the mobo manufacturers that UEFI chipset boards will not sell. If Google said "we won't be buying Dell any more due to boards running UEFI" publicly, you can guarantee Dell stock would drop quite a bit and the execs at Dell would sit up and take a lot of notice.
All in all, the only way to really defeat this is via the money. If UEFI boards meant no sales, the whole thing would become a non-issue in no time.
It's like if Balmer dragged a programmer up on stage and proceeded to flog the crap out of him
you mean, threw a chair at him.
I believe he did something like that, though for far less reason, and now has notoriety for 3 things: throwing chairs, dancing like a monkey shouting 'developers', and leading Microsoft through a decade of no share price growth.
Linus will get a reputation for being belligerent and rude... oh wait!
"many eyes" making the code work seem to be obsolete when you have a club that you have to be a member of before they take notice of you. Its always been this way in so many fields of human endeavour - if you're expecting the community around Linux to be different and based on technical excellence only then you have a lot to learn about people.
The trick is to acknowledge that you're not "one of them" and approach the task of submitting your patch like this - submit the patch as normal, and then contact Linus or someone with an email humbly telling them that you've noticed a little bug that slipped through and that you've submitted a patch that you think will fix the problem - once you have their attention enough to look at the fix, they'll be ready to accept it, as if it was something they'd noticed themselves. I guess that once you've done this a couple of times, they'd accept your future patches without review.
I know its not supposed to work that way, but it does... I'd be interested to know how you went about the submission process. It could be that they were just too busy to review all the patches that were submitted, so yours languished, un-noticed for so long.
actually, you're probably paying way more tax that you're required to pay - as seen by the recent scandals in the UK where various celebrities simply pay their money into an offshore account owned by a privately-held company and then take out a loan from said company, thus meaning their income is roughly 0, and therefore they don't have any tax to pay.
See, these schemes are quite legal, and the celebrities involved weren't required to pay any tax on an income on nothing, so why do you pay tax?
'course, said schemes are incredibly dodgy and caused a lot of backlash from the public who do see tax as a necessary evil, and rich people being able to scam their way to not paying anything as an even greater evil. The only real solution is to simplify the tax laws considerably so clever accountants cannot come up with these workarounds and loopholes. Oh, and to refuse to recognise the tax status of countries that have 0% tax systems, or to make companies that do "set up shop" (usually a post-office box) in these countries have a certain percentage of their workforce be employed there.
yes, I remember those days. I used to code OS/2 (it was a long time ago) and I had 2 binders full of everything you could ever want to know about the APIs available to me. When I had an Amiga you could buy 3 rather large books with everything, down to wiring and chipset information. Even Windows NT (3.51) had a set of 5 books in hardbound covers that told you everything.
Today, the closest you get to support is:
"hello my name is Vijay how many I help you. I see what the problem might be being, please try to turn your device off and then on again and call back if the problem remains, now please tell me how well this support information has been to you today"... and when you call back, you get someone else giving you the same 'advice' and asking to be rated in helpfulness for a little bonus.
Languages like .NET allocate a massive amount of memory - objects are new'ed all over the place, and internally even more so.
Take a look at the GC allocations counters in perfmon to see just how "few" allocations are actually done, you'll be surprised. Then take a look at the GC collection counters for an even bigger surprise.
Locality of reference is the huge thing today - my 3ghz CPU only runs that fast if it can read memory, and it can only read at that speed from L1 cache (which is quite tiny). Sure they have MBs of cache at L2 or L3, but that's still slow, just faster than main RAM. So unless your app can fit into the L1 cache, its going to run slowly (relatively), and I mean slowly like its swapping (as it will be - swapping to the other caches or main RAM).
The other aspect of a GC language is that it makes it easy for allocation to occur, so all these languages (which admittedly might be due to other factors, not the GC) tend to use a lot more ram than other types of system. More RAM = less that can fit in that cache = slower.
Letting the GC fill up a gig of ram just means you have to wait a little longer before you see it stall, which won't be very long on today's systems that gobble up a gig in no time. Then it just has more work to do compacting the large heap - which is why one performance fix to GCs was generational collection - ie running the GC on a smaller set of the heap. This is why you don't see the GC pauses you used to on old Java programs.
"is basically free".
lolol.
deallocation is only basically free when you don't do it. Otherwise, you have to compact the heap - which is most certainly not free - or you'll either run out of memory, or you'll have to start allocating using a next-free-block algorithm, just like malloc does.
I hear this a lot - that it is faster, and my observations ( of a fair few GC-based language systems) are that is is tosh.
It is faster to allocate - any system that doesn't care about memory usage is, and given an infinite block of memory that you always allocate from the end is super fast. The problem comes when you have to compact - this is a pretty slow operation. You'd never think about writing an app that copied large blocks of memory about every second, yet this is exactly what the GC does once your app gets going. De-allocation is also slower that most other systems as it has to consider finalisation (which, as you noted is shite) but it is necessary - you cannot have the system deallocate an object while it contains other resources that require freeing, and you cannot have a high-level language that mandates you to manually free those things (C does that, its a fine language for its purposes, but its not considered as high a level as you want).
Then locality of reference, this is also mostly meaningless. I know you want memory blocks to be contiguous (imagine a string where each character was in wildly different memory locations), but compare what the GC gives you compared to the C style of allocating a block of memory and filling it with data - in a GC system you can allocate memory all over the place, and the GC will compact it to a point, but those blocks will never be exactly next to each other - which is what you want for the performance benefits. If you have a c style array, or a c++ vector, then you do get this kind of performance gain (at a cost of having to copy the memory into the pre-allocated buffer, but that's still no slower than the GC copying the same data when it does a compaction).
So think about what the GC really gives you - the ability to pretend you don't care about memory.
I'd say they made a huge mistake when designing GC systems, in that they always make it work with memory, but then apply it to objects. Object lifecycle is different to memory, you want objects to disappear when they're no longer used, only in some cases can you tell the system to destroy the object when it feels like it, but in most cases you want the object to clean itself up properly. There's no real reason why they couldn't have put RAII style semantics in (ie by making the system generate automatic calls to a destructor - like Microsoft now does with the using() syntax) and let the GC manage the memory allocation only.
But still, for performance, a GC is nowhere near as fast as a deterministic program. Just look at the performance counters in a .NET app to see the GC going bonkers. I've seen it compacting the heap hundreds of times a second in a decently written large app.
so what you're saying is that when there's a fuckup in the code that causes a big problem, you'd rather it happened at random rather than in a deterministic place where you can see it occur, easily see what the problem is and fix it.
There are plenty of cases where GC is a joke, not being able to clean up your code until it feels like it doesn't make for better or more reliable systems. (an example, I have a WCF service, it uses log4net to log stuff, and I use its property system to add the caller's ip. But where there's an exception, the pop doesn;t occur in the service Dispose method because the framework has decided to hang on to it and free it up later. So the next log gets 2 ip addresses logged.. So what would be nice clean code in C++ becomes a mess of manually handling Exception exceptions in order to bypass the system so it works). GC is for lazy-ass programmers who don't know better.. RAII is for lazyass programmers who know and love what it does for them.
However, all Stallman had to offer on the topic was "We
still prefer C to C++, because C++ is so ugly" (sic).
well, score one for technical argument and intellectual reasoning.
If ugliness was a consideration, then surely GNU should be moving to a beautiful language, such as C#. If no, then WTF does subjective qualities of a language have to do with anything, there's no problem with any language as long as it is understandable (and I contend that means individual coding style rather than anything in-built - C can be so obfuscated they have competitions for it!).
the problem with all of this is the usual geek tendency of arrogance - where one always thinks he has the mental superiority to everyone else, and this is the biggest single detriment to the OSS community of all.
possibly, the trouble is you get a lego set that has all the special pieces for the puzzle that you get in the box. Its not like my day when a curved piece of (I suppose pavement) was a fancy piece that became the edge of the gun emplacement on my space-battleship.
Today, lego kits are really just big 3d jigsaw puzzles for 20/30 somethings who used to have lego when they were kids.
Lego needs to make big boxes of generic bricks cheaply available, though you're better off looking for these in garage sales.
if I had a nail stuck in my arm, I'll (manually) pull it out. You, it seems, would leave it stuck in you and say "its ok, I have another arm".
however, that just masks the problem, doesn't solve it.
With the current Firefoxes, you can go to the help menu and see what tabs are using what memory from the 'troubleshooting information' item. then you can see a rogue tab has gobbled all your ram, and close it. That's a solution.
fuckwittery is like obscenity - I know it when I see it.
and when I find such a fuckwittery in practice, my tactic on resolving it depends on the quality of the fuckwit - if he's my peer, I point it out and expect him to realise it himself, if he's my junior then he gets a code review session. If he's my boss, I offer to polish the bits his greatness didn't have time to complete himself :)
that's becuase it isn't a design suite really. Its a developer tool that they thought - either through greed or naievety - that they could sell twice.
Once upon a time, all the developer tooling was inside visual Studio, so a dev could knock up a GUI and hook it into the code he'd written. Then Microsoft invented XAML and next thing you know visual studio's XAML editing capabilities were extremely poor (ie often would crash if you tried to do anything remotely complicated, couldn't review what the GUI looked like if you edited the underlying xaml code, and could barely design the gui without hand-edited bits anyway) but its was ok as they released a separate GUI design tool you could use alongside VS.
the idea was that your designer could write XAML and then give it to the coder who would write the code and tie the two together... shame that, in true Microsoft fashion, separation of concerns like this is always broken by the need to "integrate" everything, so pretty soon you had to write code ("code behind") in your XAML to make it do anything useful, like binding a variable to a GUI element - see, simples!
So all in all, I can see why they wanted to put this stuff back into Visual Studio - its easier for them, and it means your designer will have to buy a copy of VS instead.
the only coding standard that is required there is the one that says:
Code MUST not be written like you are a fuckwit.
Even without a standard, I'd look at that and either gently rebuke the coder (who I hope would agree that it was bad and needed formatting - we all do stuff like this occasionally), or would just reformat it to be readable. No drama, problem solved.
The problem with the standard is that it refuses the places where the: "if (foo) bar(); " style is appropriate - eg. times when you have a large block of repetitive code like this, where formatting it onto 2 (or 4 with bracket placement demands) would make it unreadable.
All in all I find that style matters very little. So what if my colleague writes underscores between names in variables, and another uses camel case.. as long as everyone uses the same style within the file, I'm happy (actually, happier than a stupid standard that you have to refer to to keep your code from failing the automated style nazis)
However, the parts of a standard that really do matter are the bits no-one seems to care about (when they're all arguing about the trivialities of bracket placement): project layout. Its no good having a file with a certain style of naming convention if I cannot find the flipping file that has the code I want to look at - we've all seen funny layouts with bits of file here and there, not to mention libraries or other dependencies tucked away in a sub-sub-folder, or documentation that is a text file hidden away somewhere.
The thing that really annoys me is that 'professional' coders spend so much time arguing the toss about style that they miss the stuff that matters. Maybe its because I have a foot in the release management side of things that this matters more to me, or maybe I just like things to be neat.
Style.. well, no-one can read your code even with a style-based standard as they will lay classes and namespaces differently, and their functions will always require investigation - so it realy doesn't save any time.A standard that says to put all your private variables at the end of the class, that might make a difference compared to say scattering them throughout the class - but who would do that, that would respond to a slap anyway.
but what if they are performing a background task? what then?
so the General Electric XM134 minigun (as seen in Predator) is a legal weapon! w00t!
At least the wimpy antisocial loners who seem to be the ones shooting up their schools won't be able to lift it, so I guess it is a safe option there.
well, I think a gun barrel made of layers of resin will work fine - go ahead guys, be sure to test it yourselves first though.
It's not like they own Phonegap for creating naive Mobile Apps with HTML5 and JS,/i>
you're referring to Apache Cordoba right?
Kudos to Adobe for giving it to the Apache Foundation though, they get a lot of bad press - mainly because of crappy Flash-based adverts - but they deserve a bit better. Only a bit mind, if they release a version of Flash without security flaws or Reader that opens quickly without a mass of 'features'.
you're also not missing the benchmarks comparing the performance to a Windows box running the same benchmarks (where applicable) or the benchmarks showing the proprietary driver's performance.
performance per watt - yeah, sure, give it to Nvidia.
Or they could have chosen ARM to design faster processors that still use less power. Or give it to AMD who sure need the money.