Well, part of the analogy was, people are more likely to be simple masochists and run Windows and OS X for the fun of it, than chub-chasers who like the smell of body odour or will put up with it just because they're cheaper to maintain and less likely to run off with someone else:D
It started getting a bit complicated though so I left it.
In the end, Linux doesn't offer any greater efficiency, better features, over the fact that it is Software Libre rather than Software Gratis.
Linux is more likely to take off in the hands of consumers in the embedded device space - routers, phones, UMPC, and migrate to larger and larger systems.
When these systems in turn talk to the Big Iron running virtualized Xen, VMware on hardware by IBM or Sun, all running Linux or something similar, there will start to appear a kind of squashing effect.
Windows took the desktop space and spread outwards to handhelds and servers, but it's slowly been pushed back into it's little desktop box.
Eventually Windows will get popped out of the middle in favor of a homogenous, entirely Linux (or at least UNIXy) environment, rather than heterogenous environment of Windows, VxWorks, AIX, Solaris, QNX, MorphOS, MacOS, or whatever you "dare" to choose otherwise.
At least that's my take on it, if you're into the Big Crunch theory. I suppose it is just as likely that you'll have a MacOS handheld, a MacOS desktop and a MacOS server.
Here is the problem: you can't convince people that "Peer reviewed source, therefore more security, and oh it's free" is a good reason to switch to Linux.
Most people don't understand what peer reviewed source means, have no idea of the security of their PC (and not a care in the world anyway if they can just drop a virus checker on it and "solve" it) and, Windows and MacOS came with their system anyway, so are ostensibly free.
Linux has to actually expose a feature people want and do it so that it increases productivity and feels better than Windows or MacOS X. There was a podcast on The Register the other week with Mark Shuttleworth - the basic premise of part of it was that Compiz is cool, but useless, and it's the hope that enabling it by default means developers will turn it from a cool whizzy 3D smooth suave thing into something that improves user's experience, and their lives.
And that's why MacOS X and Windows win, because MacOS has Genie Effects (this is the carrot) but it also has Spotlight, and iTunes, and iPhoto, and Quicktime, and all the other stuff people want and need every day (this is the stick). Where MacOS has a soft, warm and inviting stick, brandished by a really hot chick in leather and a penchant for candle wax, Linux's stick has a poo on the end, and is brandished by a 300lb atheist liberal.
You're going to start a business based around commodity hardware *AND* commodity software?
You're not going to make any money. As soon as your B2B relationships are formed they can be broken again - either by your customers working out they can use your code for free (GPL rocks!!!) and buying their own equipment, or by directly enabling your competitors. The only way you're going to make money is by perceived vendor lock-in through "certification" of your POS device.
At least one portion needs to be only sourcable by you, or difficult for a customer to source themselves - using embedded boards (even if they are x86) for example. The PicoITX stuff is still rare as rocking horse shit and offers all the connections you might want. Putting it in a custom little box and collecting some "known good" peripherals would work. The entire solution (the words "inexpensive, off-the-shelf peripherals" had better not be in your product brochure) is your sales pitch.
The only way commodity hardware and commodity GPL software works for you here is that it will become impossible for your customers to steal your software and go it alone without contributing their work back to the world; but that doesn't mean they have to particularly make any effort to work with you once they have kicked you and your support contract into the dust.
What you need, then, is a morally pristine customer who is lazy, and everything will work just fine. They'll pay for everything.
You could also say that Microsoft should not support virus authors and DDoS-happy blackhats and so on by giving them 80% of the world's computers to easily crack. It's not always easy to find the guy who is hammering your servers or your corporation with spam, packet floods or spreading credit card information from your customers or whatever. Maybe you can find out what systems he used, but his address, to prosecute him? That could take forever.. or be never.
It's a lot more responsible to stop him from doing it so easily than to just say "our systems are insecure as shipped, if you did not buy them then you are free to contribute to the problem, and if someone cracks our insecure software on your pirate systems, why don't YOU go find them and shop them in to the police?"
Well, I think this is a push to increase security; it's harder to get any of those crappy ActiveX exploits through IE7 because of it's insane "please confirm installing a plugin 3 times" methodology. With WGA enabled you have all your legitimate Windows users using IE7 (or at least having it installed, remember IE7's browser components are used throughout XP - help files, embedded in other apps..) but everyone pirating it still uses the previous versions with no security updates installed.
You could easily claim (and be right) that disallowing the vast majority of pirated Windows copies the latest security updates contributes to the spread of viruses, trojans and generally misappropriation of networks.
After all, until Firefox implements some kind of MSHTML.DLL replacement scheme (would this be so difficult, really?), it is not possible to completely remove Internet Explorer from a standard Windows system (WinXP Lite etc. notwithstanding) and have it still function the same way.
Someone should port the Wine MSHTML.DLL back to Windows.. and have it use Gecko, in order that we completely reduce the requirement of Windows on the obvious things. I think it'd have to be modified to use ActiveX controls though, there was a project for this once, I really can't work out why they abandoned it though (ActiveX security policies may be easily broken etc. but it would have the happy benefit of enabling everyone with IE-requiring internet banking etc. to use those sites, too!)
I basically think if the guys at Firefox were really serious about putting themselves as a true alternative to IE, they would focus a little more on truly replacing IE rather than just being installed side-by-side.
You can't just say "oh it flopped, let's throw it away and start again". There is no precedent for this development model whatsoever.
What if they released a "crappy" version of Ubuntu and then say, fuck it, let's start again and use Fedora as a base instead of Debian, or let's base Ubuntu on QNX instead? They wouldn't do it. The development time would be in the order of half a decade or more.
In fact, the same time it took Apple to move from MacOS Classic to MacOS X. MacOS 9 lasted 3 years before the change, and is still bundled with MacOS X as a Classic environment even now. The only impetus for "ditching" MacOS 9 would be the move to Intel, where it cannot run in the same kind of emulation environment. This is an OS that lasted 8 years past it's time on earth, and was only a stopgap between 8.x (which was terrible) and the oft-delayed X anyway, somewhat allayed by the inclusion of the Carbon application framework.
Given how complicated that was, do they expect Microsoft do to the same and spend the next decade making a brand new OS? I really fucking doubt anyone with even a single brain cell engaged would...
The problem with filling your city with security cameras is, if your local police force (per-precinct) are incompetant, it doesn't fix anything. There is a careful balancing act of how well your local coppers do on catching criminals, and how many cameras they need.
What the article describes is a lackadaisical police force trying to solve their crime problem with cameras, which won't work. There are many instances where the local police are GOOD at what they do (after all they're league tabled just like everything in the UK) and where adding cameras to that cuts crime or improves the crime solving rate even further. That just isn't the case here.
After all, one of the 'advantages' of a camera-filled society is that crime is discouraged since it's all too easy for criminals to show their face or follow the criminal back to their city center council flat - and yes there are stupid ones who graffiti their name on the walls of the places they steal from who don't require cameras to drop themselves in it - but, if the police force are a bunch of idiots they won't be catching the guys that DO commit crimes in the first place. Having their face on camera does not mean an instant arrest..
I think IBM weighing in will make it easier (more developers) but I'd be sad to see the commercially-oriented development structure go away. I doubt IBM are planning that considering their focus is a Microsoft-competitive Lotus suite and not entirely freedom-oriented.
Taking some of the control from Sun, and having IBM give in some effort and direction will mean the product can only get better. Wresting control from them and doing a design-by-committee open-source movement might fundamentally destroy the package.
There are only very few projects which have been spawned from a commercial development and moved to a true open source, open development and open community design model and survived with a great product. One might say Firefox is one of them, but would it have even gotten there if Netscape/AOL hadn't been pushing their buttons to produce browsers? It's perfectly possible that, given the way most open source projects are run, we would still be running Mozilla 1.8 beta right now and Firefox would never have been spawned from it.
I guess, if you want to fork Open Office, you're free to. Go ahead, make Open Open Office and see how far you get. The best parts of it might be rolled into Lotus Suite and Star Office, or.. they might not.
MS will fix the dumb "now my Gbit adapter only runs at 40Mbit/s" bug and people will whine that they only get 600Mbit/s performance out of it, even though that might be down to their router.
You can't win when users are furiously fiddling to increase numbers by fractions of a percent anyway, maybe MS should have just ignored the needs of customers playing high definition content, and just made it run like shit unless you have a PC that costs more than a lot of people earn in 6 months?
I don't think the audio transfer bug is to do with whether or not you can play HD content though, which is what Peter was whining about.
As I think I said on the original article, it is VERY common in system design to prioritise certain components of the system in favor of others, when it will improve the user experience. I have worked on a bunch of SoC's and northbridges and they all have bus priorities you can tweak, the recommendation is basically that ethernet is not as important as audio or video, since skips in audio or video you will notice, and network bandwidth you will not.
However such system design is very specific - and most systems are designed for very specific purposes. On a 400MHz SoC you may cut priority from disk and network to service audio, but you would never notice due to the lack of use of disk and network. Windows is a desktop operating system - server operating system sometimes. When you cut disk and network performance to allow audio to run undisturbed and video to be decoded without skipping or glitches, people notice because they're using the disk and network (copying files, playing games..)
Windows XP had a network reservation system (which could reserve something like 20% of maximum bandwidth) which people misunderstood too; changing registry settings and removing QoS handlers from the system for something which did NOT affect network performance in the slightest during normal use of the system. It was well coded but, in the end, nobody bothered to use it because most users tweak it out and most developers refused to use such an API (media players and network apps worked well enough without it). Without using that API, your bandwidth is the same as it always was, and with using that API, your bandwidth was probably the same as it always was (I have a little tool which watches Media Player do it, and the reservation was never more than the maximum bandwidth required to play the video stream; 1% maybe. If you were streaming from the network at full rate for the media file, you would lose that bandwidth to other uses anyway. The difference is, copying a file and using the traditional queueing mechanism will NOT get in the way of video streaming because Media Player has asked to be guaranteed enough bandwidth and priority service on that stream! This is how QoS is meant to work, and QoS is not a Windows Extended Standard, but an IEEE one:)
Basically, media playback is a deterministic thing. You need to get audio when you ask for it, and you need to get video when you ask for it. If this service is to be guaranteed stable, you need to reprioritise the rest of the system to do it. What is happening on Vista is that it has taken it to an extreme on some systems and unfortunately hasn't been implemented too well with regards to multiple network adapters or systems which have very high specifications. You do not need to reprioritise networking if you have a gigabit network on a Core Quad 2.6GHz system with Intel's latest chipset streaming to a Creative Labs soundcard, but Windows Vista doesn't know that.
However Vista could be made to be more deterministic about it; measure system performance as it does with the Games etc. (after all, if I need a 1.0 score to play Hearts and I have a 4.5, it needn't worry) - and work out quite simply, do I have enough system performance to acheive this goal, and if not, set a reservation or reprioritise.
I wonder how they could do that though. The only way you can accurately measure the need is to experience it; you may play 2 minutes of an HD movie and notice the audio skips because you are copying a file, and Windows might adjust interrupt servicing in the networking subsystem to allow audio interrupts priority, or adjust buffer sizes to compensate, and it will be fine for the 2 hour duration of the HD movie. But this would always be AFTER the user was frustrated by a video or audio skip, maybe in a scene-setting action portion or narration.
All in all, it should never just drop your network perfo
Compiling x86 code now brings them zero steps closer to the goal;)
There is MIPS, VAX, PDP10 code generators in the source code but, OpenBSD never worked on any of the weirder ones and the MIPS support is busted. I reckon they should not focus so much on the "architectures that gcc has dropped" but on the architectures gcc sucks at (everything except x86).
Well, that's fairly pointless that it only compiles x86 code. So, it basically means they're competing to make a compiler which is BSD licensed and compiles the most common processor out there, against the biggest GPL licensed compiler on the most common processor?
I really think some effort needs to be put in to support other processors like ARM, MIPS and PPC - after all, GCC is a hardly a diva for any of these (code generation is terrible) and they are not the fastest platforms either. This is where actual performance improvements and quality code generation would actually count for something.
I think it's even LESS relevant considering Apple bundle and "tie" these features to their operating system too by bundling iLife and encouraging.Mac subscriptions (http://www.apple.com/dotmac/).
I downloaded the Windows Live installer and ran it on XP. Yes, it requires IE7, but that's pretty much a required upgrade here and it's probably far more to do with security updates and browser technology required to do it than it is to do with "we will never make it work on Firefox".
After all, everyone has to have IE installed on Windows, but they do not have Firefox when they first boot up. Microsoft couldn't use Firefox as a hard dependency. Removing IE from Windows means, removing internet access needed to download Firefox. So, you can't get rid of it any more than you'd want to get rid of Safari from MacOS.
In what way does any tax money go to pay for this guy's research?
We didn't pay him to do the research, so saying "we paid our taxes so we demand to see it!" is irrelevant. What is relevant here is that he released it under Creative Commons.
They're not so much idiots, but I would find it INCREDIBLY difficult to name a function "functionSigniture". It's just WRONG.
There is absolutely no reason for spelling it that way, even if you're Hooked On Phonics, it isn't pronounced that way in any language approaching English.
If this is an exported API function then it would cause a huge problem. Now, consider that a guy who is writing a function which calculates some kind of signature (perhaps a hash or a certification routine) cannot even spell the word which describes what he is doing. Does that give you confidence that the signature function is correct?
Still, the point stands; if your developers can't form a coherent sentence using well-spelled function names I'd fear for their code in the first place. It only takes a couple of typos to make code readability drop through the floor. You don't want automated tools you want to hire developers who can write.
But, the quality of education seems to have declined in recent years. I remember writing stuff for English class at school and you'd get your work scribbled in red ink for making spelling mistakes, all the time. I've looked at my brother's marked English homework (he's 15) and even the glaring mistakes are missed. Having to type everything rather than hand write it seems to be the source.
People need to be able to write, and not just trust a spell checker. But then again, this ALSO falls down when you don't have native English speakers on staff.
I've got a couple of projects on the radar right now where tiny spelling mistakes are in production code - API definitions, symbols that are exported - that just appear in every version. If someone had been reviewing and had an eye for it they would be fixed. What doesn't help is none of the guys on the projects are English or American besides me..
Review your code, review their core, spell check everything by hand, from the design document down.
Actually, if you wrote a design document and a functional prototype of your software in the first place, a great deal of the user-visible API will be set in stone anyway. Including naming. You can use any word processor spell check in that case.
Truth hurts, doesn't it? There is absolutely no correlation between high IQ and how liberal you are. There are hundreds of studies to this effect, I could probably go to the local university and look up a doctoral thesis covering it and come back with more than a handful. Some of the most intelligent people on the planet are right wing nutjobs. Nutjobs, yes, but still highly intelligent.
Only a geek would run into a forum and say "I hold my political beliefs because I am more intelligent than you".
I like to liken the geek world to the mud-packing "annoying peasant" in Monty Python and the Holy Grail. As a comment on this little segment of society I think it really, really works. Graham Chapman and Michael Palin really have gotten you pegged in that scene, whether they knew it or not, it fits.
"Help, help, I'm being repressed!"
I just wonder what watery bint threw a scimitar at Eric Raymond and Richard Stallman.
Actually the Qt licensing got better last month, now it's GPL with some great exceptions; however using it in a commercial application still requires licensing, I don't figure this is too bad. Anyone who was using it commercially anyway (before it was GPL) was paying for it anyway. Nothing changed there.
After all, the licensing money goes to pay Trolltech developers to improve the toolkit. That is a lot more than you can say about GTK+. I have a deep distrust of something that was born out of the need for a single application to have a GUI and then spread across the rest of the application world. It smacks of an incremental design process which I don't think works in software - the glorious exception being Linux, but then Linux is very, very well managed by a BDFL. Who's the BDFL for GNOME and GTK?
I wonder why KDM got to be classed as 'abandonware' by some people here? It's a great login manager..
Slashdot Mirror
Well, part of the analogy was, people are more likely to be simple masochists and run Windows and OS X for the fun of it, than chub-chasers who like the smell of body odour or will put up with it just because they're cheaper to maintain and less likely to run off with someone else :D
It started getting a bit complicated though so I left it.
In the end, Linux doesn't offer any greater efficiency, better features, over the fact that it is Software Libre rather than Software Gratis.
Linux is more likely to take off in the hands of consumers in the embedded device space - routers, phones, UMPC, and migrate to larger and larger systems.
When these systems in turn talk to the Big Iron running virtualized Xen, VMware on hardware by IBM or Sun, all running Linux or something similar, there will start to appear a kind of squashing effect.
Windows took the desktop space and spread outwards to handhelds and servers, but it's slowly been pushed back into it's little desktop box.
Eventually Windows will get popped out of the middle in favor of a homogenous, entirely Linux (or at least UNIXy) environment, rather than heterogenous environment of Windows, VxWorks, AIX, Solaris, QNX, MorphOS, MacOS, or whatever you "dare" to choose otherwise.
At least that's my take on it, if you're into the Big Crunch theory. I suppose it is just as likely that you'll have a MacOS handheld, a MacOS desktop and a MacOS server.
Here is the problem: you can't convince people that "Peer reviewed source, therefore more security, and oh it's free" is a good reason to switch to Linux.
Most people don't understand what peer reviewed source means, have no idea of the security of their PC (and not a care in the world anyway if they can just drop a virus checker on it and "solve" it) and, Windows and MacOS came with their system anyway, so are ostensibly free.
Linux has to actually expose a feature people want and do it so that it increases productivity and feels better than Windows or MacOS X. There was a podcast on The Register the other week with Mark Shuttleworth - the basic premise of part of it was that Compiz is cool, but useless, and it's the hope that enabling it by default means developers will turn it from a cool whizzy 3D smooth suave thing into something that improves user's experience, and their lives.
And that's why MacOS X and Windows win, because MacOS has Genie Effects (this is the carrot) but it also has Spotlight, and iTunes, and iPhoto, and Quicktime, and all the other stuff people want and need every day (this is the stick). Where MacOS has a soft, warm and inviting stick, brandished by a really hot chick in leather and a penchant for candle wax, Linux's stick has a poo on the end, and is brandished by a 300lb atheist liberal.
Firefox and Opera and Safari have the same problem here..
You're going to start a business based around commodity hardware *AND* commodity software?
You're not going to make any money. As soon as your B2B relationships are formed they can be broken again - either by your customers working out they can use your code for free (GPL rocks!!!) and buying their own equipment, or by directly enabling your competitors. The only way you're going to make money is by perceived vendor lock-in through "certification" of your POS device.
At least one portion needs to be only sourcable by you, or difficult for a customer to source themselves - using embedded boards (even if they are x86) for example. The PicoITX stuff is still rare as rocking horse shit and offers all the connections you might want. Putting it in a custom little box and collecting some "known good" peripherals would work. The entire solution (the words "inexpensive, off-the-shelf peripherals" had better not be in your product brochure) is your sales pitch.
The only way commodity hardware and commodity GPL software works for you here is that it will become impossible for your customers to steal your software and go it alone without contributing their work back to the world; but that doesn't mean they have to particularly make any effort to work with you once they have kicked you and your support contract into the dust.
What you need, then, is a morally pristine customer who is lazy, and everything will work just fine. They'll pay for everything.
You could also say that Microsoft should not support virus authors and DDoS-happy blackhats and so on by giving them 80% of the world's computers to easily crack. It's not always easy to find the guy who is hammering your servers or your corporation with spam, packet floods or spreading credit card information from your customers or whatever. Maybe you can find out what systems he used, but his address, to prosecute him? That could take forever.. or be never.
It's a lot more responsible to stop him from doing it so easily than to just say "our systems are insecure as shipped, if you did not buy them then you are free to contribute to the problem, and if someone cracks our insecure software on your pirate systems, why don't YOU go find them and shop them in to the police?"
Well, I think this is a push to increase security; it's harder to get any of those crappy ActiveX exploits through IE7 because of it's insane "please confirm installing a plugin 3 times" methodology. With WGA enabled you have all your legitimate Windows users using IE7 (or at least having it installed, remember IE7's browser components are used throughout XP - help files, embedded in other apps..) but everyone pirating it still uses the previous versions with no security updates installed.
You could easily claim (and be right) that disallowing the vast majority of pirated Windows copies the latest security updates contributes to the spread of viruses, trojans and generally misappropriation of networks.
After all, until Firefox implements some kind of MSHTML.DLL replacement scheme (would this be so difficult, really?), it is not possible to completely remove Internet Explorer from a standard Windows system (WinXP Lite etc. notwithstanding) and have it still function the same way.
Someone should port the Wine MSHTML.DLL back to Windows.. and have it use Gecko, in order that we completely reduce the requirement of Windows on the obvious things. I think it'd have to be modified to use ActiveX controls though, there was a project for this once, I really can't work out why they abandoned it though (ActiveX security policies may be easily broken etc. but it would have the happy benefit of enabling everyone with IE-requiring internet banking etc. to use those sites, too!)
I basically think if the guys at Firefox were really serious about putting themselves as a true alternative to IE, they would focus a little more on truly replacing IE rather than just being installed side-by-side.
Can you get any simpler and in-spirit than 3-clause BSD? Are we just going to get the "Googlized MIT License" out of this?
I was under the impression that Leopard for PPC will still have the MacOS 9 environment.
Okay so new iMacs won't have it but everyone who still has anything above an 866MHz G4, can...
I heartily agree.
You can't just say "oh it flopped, let's throw it away and start again". There is no precedent for this development model whatsoever.
What if they released a "crappy" version of Ubuntu and then say, fuck it, let's start again and use Fedora as a base instead of Debian, or let's base Ubuntu on QNX instead? They wouldn't do it. The development time would be in the order of half a decade or more.
In fact, the same time it took Apple to move from MacOS Classic to MacOS X. MacOS 9 lasted 3 years before the change, and is still bundled with MacOS X as a Classic environment even now. The only impetus for "ditching" MacOS 9 would be the move to Intel, where it cannot run in the same kind of emulation environment. This is an OS that lasted 8 years past it's time on earth, and was only a stopgap between 8.x (which was terrible) and the oft-delayed X anyway, somewhat allayed by the inclusion of the Carbon application framework.
Given how complicated that was, do they expect Microsoft do to the same and spend the next decade making a brand new OS? I really fucking doubt anyone with even a single brain cell engaged would...
The problem with filling your city with security cameras is, if your local police force (per-precinct) are incompetant, it doesn't fix anything. There is a careful balancing act of how well your local coppers do on catching criminals, and how many cameras they need.
What the article describes is a lackadaisical police force trying to solve their crime problem with cameras, which won't work. There are many instances where the local police are GOOD at what they do (after all they're league tabled just like everything in the UK) and where adding cameras to that cuts crime or improves the crime solving rate even further. That just isn't the case here.
After all, one of the 'advantages' of a camera-filled society is that crime is discouraged since it's all too easy for criminals to show their face or follow the criminal back to their city center council flat - and yes there are stupid ones who graffiti their name on the walls of the places they steal from who don't require cameras to drop themselves in it - but, if the police force are a bunch of idiots they won't be catching the guys that DO commit crimes in the first place. Having their face on camera does not mean an instant arrest..
I think IBM weighing in will make it easier (more developers) but I'd be sad to see the commercially-oriented development structure go away. I doubt IBM are planning that considering their focus is a Microsoft-competitive Lotus suite and not entirely freedom-oriented.
Taking some of the control from Sun, and having IBM give in some effort and direction will mean the product can only get better. Wresting control from them and doing a design-by-committee open-source movement might fundamentally destroy the package.
There are only very few projects which have been spawned from a commercial development and moved to a true open source, open development and open community design model and survived with a great product. One might say Firefox is one of them, but would it have even gotten there if Netscape/AOL hadn't been pushing their buttons to produce browsers? It's perfectly possible that, given the way most open source projects are run, we would still be running Mozilla 1.8 beta right now and Firefox would never have been spawned from it.
I guess, if you want to fork Open Office, you're free to. Go ahead, make Open Open Office and see how far you get. The best parts of it might be rolled into Lotus Suite and Star Office, or.. they might not.
MS will fix the dumb "now my Gbit adapter only runs at 40Mbit/s" bug and people will whine that they only get 600Mbit/s performance out of it, even though that might be down to their router.
You can't win when users are furiously fiddling to increase numbers by fractions of a percent anyway, maybe MS should have just ignored the needs of customers playing high definition content, and just made it run like shit unless you have a PC that costs more than a lot of people earn in 6 months?
I don't think the audio transfer bug is to do with whether or not you can play HD content though, which is what Peter was whining about.
:)
As I think I said on the original article, it is VERY common in system design to prioritise certain components of the system in favor of others, when it will improve the user experience. I have worked on a bunch of SoC's and northbridges and they all have bus priorities you can tweak, the recommendation is basically that ethernet is not as important as audio or video, since skips in audio or video you will notice, and network bandwidth you will not.
However such system design is very specific - and most systems are designed for very specific purposes. On a 400MHz SoC you may cut priority from disk and network to service audio, but you would never notice due to the lack of use of disk and network. Windows is a desktop operating system - server operating system sometimes. When you cut disk and network performance to allow audio to run undisturbed and video to be decoded without skipping or glitches, people notice because they're using the disk and network (copying files, playing games..)
Windows XP had a network reservation system (which could reserve something like 20% of maximum bandwidth) which people misunderstood too; changing registry settings and removing QoS handlers from the system for something which did NOT affect network performance in the slightest during normal use of the system. It was well coded but, in the end, nobody bothered to use it because most users tweak it out and most developers refused to use such an API (media players and network apps worked well enough without it). Without using that API, your bandwidth is the same as it always was, and with using that API, your bandwidth was probably the same as it always was (I have a little tool which watches Media Player do it, and the reservation was never more than the maximum bandwidth required to play the video stream; 1% maybe. If you were streaming from the network at full rate for the media file, you would lose that bandwidth to other uses anyway. The difference is, copying a file and using the traditional queueing mechanism will NOT get in the way of video streaming because Media Player has asked to be guaranteed enough bandwidth and priority service on that stream! This is how QoS is meant to work, and QoS is not a Windows Extended Standard, but an IEEE one
Basically, media playback is a deterministic thing. You need to get audio when you ask for it, and you need to get video when you ask for it. If this service is to be guaranteed stable, you need to reprioritise the rest of the system to do it. What is happening on Vista is that it has taken it to an extreme on some systems and unfortunately hasn't been implemented too well with regards to multiple network adapters or systems which have very high specifications. You do not need to reprioritise networking if you have a gigabit network on a Core Quad 2.6GHz system with Intel's latest chipset streaming to a Creative Labs soundcard, but Windows Vista doesn't know that.
However Vista could be made to be more deterministic about it; measure system performance as it does with the Games etc. (after all, if I need a 1.0 score to play Hearts and I have a 4.5, it needn't worry) - and work out quite simply, do I have enough system performance to acheive this goal, and if not, set a reservation or reprioritise.
I wonder how they could do that though. The only way you can accurately measure the need is to experience it; you may play 2 minutes of an HD movie and notice the audio skips because you are copying a file, and Windows might adjust interrupt servicing in the networking subsystem to allow audio interrupts priority, or adjust buffer sizes to compensate, and it will be fine for the 2 hour duration of the HD movie. But this would always be AFTER the user was frustrated by a video or audio skip, maybe in a scene-setting action portion or narration.
All in all, it should never just drop your network perfo
I think I'm going to write a blog on this! It seems the cool thing to do.
I'm taking suggestions on how much bullshit and FUD I should put in!!!
*smirk*
:D
That would be nice, though, right? A mobile phone compiler that can actually run at a decent speed without using up a whole flash card?
Nintendo DS Homebrew could really benefit from pcc on ARM.
That was what I said wasn't it?
;)
Compiling x86 code now brings them zero steps closer to the goal
There is MIPS, VAX, PDP10 code generators in the source code but, OpenBSD never worked on any of the weirder ones and the MIPS support is busted. I reckon they should not focus so much on the "architectures that gcc has dropped" but on the architectures gcc sucks at (everything except x86).
Well, that's fairly pointless that it only compiles x86 code. So, it basically means they're competing to make a compiler which is BSD licensed and compiles the most common processor out there, against the biggest GPL licensed compiler on the most common processor?
I really think some effort needs to be put in to support other processors like ARM, MIPS and PPC - after all, GCC is a hardly a diva for any of these (code generation is terrible) and they are not the fastest platforms either. This is where actual performance improvements and quality code generation would actually count for something.
Yes and Grandma is really going to do that herself isn't she?
I think it's even LESS relevant considering Apple bundle and "tie" these features to their operating system too by bundling iLife and encouraging .Mac subscriptions (http://www.apple.com/dotmac/).
I downloaded the Windows Live installer and ran it on XP. Yes, it requires IE7, but that's pretty much a required upgrade here and it's probably far more to do with security updates and browser technology required to do it than it is to do with "we will never make it work on Firefox".
After all, everyone has to have IE installed on Windows, but they do not have Firefox when they first boot up. Microsoft couldn't use Firefox as a hard dependency. Removing IE from Windows means, removing internet access needed to download Firefox. So, you can't get rid of it any more than you'd want to get rid of Safari from MacOS.
In what way does any tax money go to pay for this guy's research?
We didn't pay him to do the research, so saying "we paid our taxes so we demand to see it!" is irrelevant. What is relevant here is that he released it under Creative Commons.
They're not so much idiots, but I would find it INCREDIBLY difficult to name a function "functionSigniture". It's just WRONG.
There is absolutely no reason for spelling it that way, even if you're Hooked On Phonics, it isn't pronounced that way in any language approaching English.
If this is an exported API function then it would cause a huge problem. Now, consider that a guy who is writing a function which calculates some kind of signature (perhaps a hash or a certification routine) cannot even spell the word which describes what he is doing. Does that give you confidence that the signature function is correct?
How wonderfully ironic :)
Still, the point stands; if your developers can't form a coherent sentence using well-spelled function names I'd fear for their code in the first place. It only takes a couple of typos to make code readability drop through the floor. You don't want automated tools you want to hire developers who can write.
But, the quality of education seems to have declined in recent years. I remember writing stuff for English class at school and you'd get your work scribbled in red ink for making spelling mistakes, all the time. I've looked at my brother's marked English homework (he's 15) and even the glaring mistakes are missed. Having to type everything rather than hand write it seems to be the source.
People need to be able to write, and not just trust a spell checker. But then again, this ALSO falls down when you don't have native English speakers on staff.
I've got a couple of projects on the radar right now where tiny spelling mistakes are in production code - API definitions, symbols that are exported - that just appear in every version. If someone had been reviewing and had an eye for it they would be fixed. What doesn't help is none of the guys on the projects are English or American besides me..
Review your code, review their core, spell check everything by hand, from the design document down.
Actually, if you wrote a design document and a functional prototype of your software in the first place, a great deal of the user-visible API will be set in stone anyway. Including naming. You can use any word processor spell check in that case.
Truth hurts, doesn't it? There is absolutely no correlation between high IQ and how liberal you are. There are hundreds of studies to this effect, I could probably go to the local university and look up a doctoral thesis covering it and come back with more than a handful. Some of the most intelligent people on the planet are right wing nutjobs. Nutjobs, yes, but still highly intelligent.
Only a geek would run into a forum and say "I hold my political beliefs because I am more intelligent than you".
I like to liken the geek world to the mud-packing "annoying peasant" in Monty Python and the Holy Grail. As a comment on this little segment of society I think it really, really works. Graham Chapman and Michael Palin really have gotten you pegged in that scene, whether they knew it or not, it fits.
"Help, help, I'm being repressed!"
I just wonder what watery bint threw a scimitar at Eric Raymond and Richard Stallman.
Actually the Qt licensing got better last month, now it's GPL with some great exceptions; however using it in a commercial application still requires licensing, I don't figure this is too bad. Anyone who was using it commercially anyway (before it was GPL) was paying for it anyway. Nothing changed there.
After all, the licensing money goes to pay Trolltech developers to improve the toolkit. That is a lot more than you can say about GTK+. I have a deep distrust of something that was born out of the need for a single application to have a GUI and then spread across the rest of the application world. It smacks of an incremental design process which I don't think works in software - the glorious exception being Linux, but then Linux is very, very well managed by a BDFL. Who's the BDFL for GNOME and GTK?
I wonder why KDM got to be classed as 'abandonware' by some people here? It's a great login manager..