I just have the opinion that if a person is not intelligent enough to find a screwdriver, they aren't smart enough to upgrade their computer.
The point is really not about intelligence, any idiot can wield a screwdriver and fiddle inside his PC. If normal people could not use tools, Bill Gates would still be the richest man in the world.
The point is about able to do work inside the computer fast. Inserting memory can be done in less than a minute on an Apple case, it takes a least five if you have a generic no-name box. When you are upgrading a bunch of machines, it makes a difference...
Actually, Apple also had to do a transition between machines without open-firmware to the one with it, so there was some backward compatibility. On the other hand, I suspect the newer macs could not boot older oses (I don't know if EFI machines would seriously boot Windows 95, and why anybody would try that...)
The sad thing about intel doing their own stuff is that Open-firmware is here and standard. One of the most interesting ideas of having the F-code engine was to have processor independent drivers on the card. I.e you plug-in the card and it works, regardless of the fact the processor is PPC, sparc, or i686.
One funny trivia fact about Apple's open-firmware is that the firmware understands certain file-systems (HFS+,Ext2) and executable formats (PEF,ELF). The funning thing is the firmware does not understand OS X's executable format (Mach-O) so on every OS X machine, there is an ELF format bootloader.
yes, it is 'port-friendly'... in a similar way to OSX, actually.
I am not so sure. OS X was an Unix from start, and with each version, libraries for simplifying the port of Unix applications are added, so most Unix ports can compile directly or with minimal changes. Plus OS X has to two APIs that were used to build programs at some time, Carbon (classical Macintosh applications) and Cocoa (Next step applications).
From what I gathered, BeOS has Posix support and that's it.
Until GUIs start being made more with vectors and less with bitmaps they are ALWAYS going to be memory hogs.
MMh, things are not so simple. Vector data has to be rasterised to be displayed. While it would be technically possible to have the GPU do this, this is not the case for the moment. Thus the CPU has to do the rasterising and this takes time. For this reason it is common to keep cached versions of rasterized vector data in memory (typically font glyphs).
The reason lightweight display managers are fast is because the graphical elements can be represented by simple sequences of fast drawing operations (i.e vectors), typically horizontal and vertical lines and boxes. In clear, the vectorial representation is compact and the rasterising fast.
So simply relying on vectors GUI elements will not help. If each the window close button is a vector image with thousands of vector operations it will not be fast.
Rendering advanced vector operations like splines is quite CPU intensive, and if you want to save on memory, you have to make sure that the rendering occurs late and deep in the graphical system, ideally on the GPU, or at least in the graphical server. This means supporting things like splines in the X Server.
Also note that the same logic can be applied to compressed bitmaps. If you can defer the decompression of JPEG data until it is in the graphical server or the GPU, you would save quite some memory. Apple uses such a trick to a certain extent, by RLE-compressing the window buffers.
I think any GUI is going to get "heavier" over time, as more features and functioanlity are added.
You are right to a certain extent, but normally, there is a cycle. Features are added and the system slows down, then the code is optimised and redesigned to bring back some speed, and the cycle restarts.
My impression is that the second part of the cycle is weak in Linux. People do what is fun, optimising code, specially someone else's is not fun. In particular, people like to optimise their code for their spanking new machine so as to be able to boast impressive number. Also many people also think that optimising simply means compiling with -O2.
Of course another solving the problem also implies acknowledging that the problem exist, and look at how this could be solved. There was recently an article on slashdot on how to make an OS faster, look how far down you have to scroll to find some people who actually discuss how such tricks could be used for Linux instead of bitching that X is faster than Y...
My impression, is that one the main issues in Linux is redundant code and data. How many window drawing routines are lingering in memory? How many routines do decode bitmaps? How many toolbars icons with arrows?
as well as knowing the best drivers to use, best graphics options in, say, xf86config, the best compiler options in GCC, etc, etc, all contribute to the speed of the system - caching and dynamic defragmentation is just a small part of that.
Actually, this part (caching, defragmenting) is what the article is about and I suspect it is not a small part of perceived performance. Most of the optimisations presented try to minimise the time needed to have the code in memory ready to be executed (by pre-loading and pre-linking). You can fiddle gcc compilation flags all you want, it won't affect this metric much (except in regard to the size of the binary).
Now you can argue that this metric is not a good one, and does not represent responsiveness, or that those optimisations are mostly geared towards desktop usage (and I would agree). The core thing is simply that the article presented interesting way of making a desktop more responsive.
Your reaction annoyed me, because dismissing interesting concepts because Apple only has to support limited hardware instead at thinking if and how those concepts could be adapted to Linux, is not the best attitude to improve Linux.
Actually, if you had bothered with reading the article instead of repeating the old Apple has it easy with limited hardware cliché, you would have noticed that this is absolutely not related to driver performance.
Only one optimisation presented is related to hardware drivers, and it is cache of what kernel extensions will probably be loaded. Most of the optimisations (basically lots of caching and dynamic defragmentation) could be implemented in Linux, regardless of the amount of supported hardware.
Middle-click-pastes is just a more evolved interface to drag-and-drop, which has extremely similar functionality.
Except for the fact that drag and drop is not persistent. The content of the drag session will only exist during the drag session. If the session stops (successfully or not) the content of the pasteboard is cleared.
If you select, you can insert with the middle button as long as you have not selected something else.
I am not saying that select - insert is not a valid paradigm, simply claiming that is a more advanced form of drag drop seem quite a stretch to me. Additionally, this only works for text, while drag drop works for more advanced data types).
He is mentionning a Laserjet III that was introduced in 1990, so we are talking equipement that runs over ten years. In that light, three year is not so impressive.
Myself, I'm still using a Laserjet 6MP that is still working well.
How about power cords? I have about 50. Phone cords?
If you travel, keeping those things can be quite nice, so you can trade them. This is how I now have a US powercord for my laptop: I traded it in Japan...
I agree with you, as far as I know, Macs don't sleep to disk. One nice feature is that certain applications, like iChat are notified when the machine goes to sleep and automatically disconnect. I wish ssh would be able to somehow handle a connection when the machine goes to sleep.
The problem in disconnecting devices like drives while the machine is sleeping is that the device has been shutdown (in particular if, like for an iPod it gets the power from the bus) and parked, but the filesystem has not been unmounted. The data is probably safe (the OS probably does a sync on the filesystem), but force-unmounting a filesystem can confuse applications.
A thing which I found interesting is that certain buses stay powered on, while others not (USB on, Firewire off). For instance, the buit-in USB port stay powered on, which make sense on a desktop as it is the mechanism to wake up the machine. On the other hand, I have a PCI card with additional USB ports, and this one gets powered off, so if you connect you keyboard mouse on this board, you cannot wake up the machine (it is logical in a way, still took me five minutes to figure it out).
Please, go back and re-read what I wrote. You clearly to not fit into the class of users who have changed the industry by bringing in new viewpoints and tasks in the last 10 years.
Err, yes possibly, then again, who exactly has brought new viewpoints in the word-processing scene?
Rather, you were an early adopter, and the tools were suited to the needs of you and those like you 10 years ago.
I would not call myself an early adopter because I used a word-processor in the nineties, but if you wish to define me as such, fine...
Also, don't assume I mean MS Word when I say word processing.
Could you point me to this word processor that is radically different from Mac Write / Word?
It's only true that when those people are done throwing away all of the "bells and whistles" that they feel they don't need, the same infrastructure can still be found.
Yes to a certain extent, but there is also a point where a tool has matured enough so that the design of said tool becomes stable. The problem with the word-processor example, is that it is a tool with a certain goal (preparing texts for printing), it can be extended to some other taks (complex page layout), but from there on, adding more bells and whistles won't really improve the tool.
Well, true, but those bells and whistles were added precisely because the needs of the next generation of users was different, and the platform grew to encompas their needs.
Honestly the bells were also often added because the designer thought it would be a good idea, or cool, or whatever, to come back to MS Word, it also has lost some features (publish-subscribe), and some task went to the background (printing mass mailings). Could you give me an example of a new feature for a word processor of this 'new generation', which is related to the goal of the tool (i.e something which has to do with word processing).
Actually, I fully agree that certain software tools will evolve radically, but I don't think word-processors are among them.
If you think about it, word processor were designed to prepare texts, letters and the nice feature the Mac Write paradigm brought us was WISWYG, but as paper become less the main media, this task will drift away. Now if you want to call 'word processor' a very advanced XML editor that is used as a front-end to a back-end that produces some printed outputs, we agree, except on the definition of 'word processor', e.g. I would not call Latex a word processor, but I agree this is open to discussion.
I'm not sure how to respond to this...you've mixed clearly good things with the hideous, with a mixed blessing (the fact that simple files are large or the ability to handle large files?)
The point I was trying to make is that there were a few changes, but not all for the best and nothing really ground breaking. By large files, I mean that Word files have gotten bigger and bigger for the same text.
I should emphasise that the good things (drag-drop, contextual menus) are more related to changes in the OS than core changes in the design of Word. Actually I now realise that publish-subscribe also dissapeared.
Even something as "basic" as word processing has changed radically in the last 10 years as a wider variety of people have gained access to computers.
What?
Sorry, no. More than ten years ago, I was doing word-processing using Mac Write II on a Macintosh classic, now I'm using Word X on a Dual G4. Would you please tell me what radical changes happened in word processing? The interface is basically the same, the text window, the ruler above with the tab stops and some buttons for getting bold etc. Many aspects of computing have changed dramatically but not word processors.
Actually, Macwrite II was a very lightweight program, Microsoft Word for Macintosh had more features. Here is a list of improvement I have seen in Word since version 4.
Drag-drop
Contextual menus
Clippy
Macro-viruses
Large files
Others features have disappeared since, like the latex-like equation code, the mailing subsystem and the integrated screen saver .
Some people are probably doing word-processing using really radical tools (vi?) but the overall design of the mainstream application is quite stable and will not change, the current model is good enough and what people are used to.
Maybe for some non-roman scripts, something completely different will be designed, and maybe back-ported for roman systems, but until then, word processors will look like Mac Write, spreadsheets will look like Visicalc, etc.
Many ad companies already do use location targetted ads - I often see Sainsbury's, Switch etc ads on US sites. Of course this isn't 100% foolproof, but it did even work with my old (pre-RIPE) IP.
In my experience this is quite random. In my case (I'm Swiss), I get quite a variety of adds, beside the one for the US, I get some for France and some for Germany, nothing for Italy yet (Italian is the third langage spoken in Switzerland).
At least I seem not to get proposals to move my money to a Swiss bank account anymore...
I suspect the system used to select adds is quite primitive, and I'm simply too strange a case to get tailored adds.
Actually, I have been wondering about this for some time. We now have quite complex financial tools. Would it be possible to build financial tools that 'simulate' a lottery? Are there explicit financial laws to prevent his?
For instance could you build 'exclusive pool funds' which basically give control / ownership of the whole pool to one of fund owners based on some random condition? If this condition were random, you would get a lottery.
Somehow, if you consider a given stock as a random variable, you can 'code' some logic using basically less than and greater than operations with futures, does somebody know if can code more complex logic? I would think that you would need something that does mutual exclusion (only a few people win) and random (for instance by betting on the least significant 'bits' of a stock value).
1) How you describe video games as beeing the same can also be applied to movies. The same "stories" are done over and over.
The characters, places,... are all that changes. Will the movie industry ever be complete wiped out?
You are right, the same movie is done again and again, but the interesting thing is the period. For instance there are many titanic movies but on average, there is one every five years. Take Terminator movies, on average there is five years between each. Even if the you take broader categories, like catastrophic movies (airplane,ship,whatever) how many come out every year? (I admit there are recurring categories, like Bruce Willis saving the world, but well).
If you take this analogy, the gaming industry can come out with a new iteration of the same game every five years. So while the gaming industry would probably never be wiped out, it would have to cool down a bit: there are not so many different game types, and the current upgrade frequency might not be sustainable once the novelty wears off.
If you think about it there is a huge variety of different movie types, video-game diversity is not that big. I don't think that video-gaming will go the way of the dodo, but there will be some adaptation from the current model. I hope this will bring out new interesting games.
Aqua and Quartz are libraries that run on top of Darwin (the kernel), once Darwin runs on a PPC machine, running OS X is quite easy. You don't need to recompile any of the frameworks. In fact this is the way OS X can be run on machine Apple does not support.
You can run OS X without Quartz Extreme, in fact this is the case on older Macs who don't have graphics cards that are powerful enough. This simply taxes the CPU more and disables some fancy effects. Of course, if the graphic card supports the right OpenGL operations, Quartz Extreme will run.
well it will probably lack some of the chips and stuff to run OS X,
Err, what chips and stuff? Darwin does not need any special chips, ROMs or anything, this is why it can run on x86 hardware. One simply would need to write drivers for the hardware and a bootloader to run Darwin (same for Linux as a matter of fact).
A lift algorithm isn't going to take into account the number of times you press,
An algorithm taking this into account could be interesting. It would bring the lift to the most motivated persons (the one pressing faster). If you have a resource that is free and precious, it makes sense to associate it a 'cost' to optimise its use.
With such algorithms people who only need to go down a few levels would use the stairs and not bother to call the lift and people who are really in a hurry would get the lift. Of course, the 'price' of the lift would increase at rush hours and go down during the night.
PHB would ask their secretaries to press the button multiple times for them. I suppose this would make a new job, lift button presser...
I never implied this and anyway it would not little sense, as there are not many linux games to warez.
If anything it's the Windows gamers that are more likely to be used to getting stuff for free (like Windows and Office).
Windows users are not the issue, they are the majority and the core market the games are targetted at it.
...But so is your insistence that Linux users won't shell out for games. I will and I do.
The real question is, will you shell out for a Linux version of a game that is:
More expensive than the Windows version.
Less complete.
It is one thing to claim that if game XXX is ported to Linux, I will buy it and another to buy it when it comes out and is more expensive that the Windows version and doesn't have all the features.
I recently bought Dungeon Siege for my OS X machine. It was more expensive than the Windows version (which was not so hot anymore and thus discounted), and some things are missing (the level editor). Still I paid for it, and I'm happily playing it.
I'm don't think that Linux users are cheap warez grabbing people and never implied this, but those I know are very cost-conscious and will resent buying over-priced, down-graded versions (especially when they can boot windows a play the first rate game).
Again this is anecdotal and based on the Linux users I know, but given the prominence of posts of Linux users on slashdot describing Mac users as idiots that buy overpriced stuff, I think this stereotype is quite common...
Remember also that porting games Windows->Linux x86 is a hell of a lot easier/cheaper than doing it Windows->OS X PPC. Indeed, some developers will already have some kind of Linux x86 toolchain because their dedicated multiplayer servers run on Linux.
Porting a Win x86 game to Linux x86 is certainly technically easier that porting it to OS X - although I think you minimise the issues associated with Linux- technical issues are just that, technical issues that have to be solved, what is important is the money.
The question is what makes better business sense, to do a port nobody will buy, or a slightly more difficult port that will be sold. TransGaming seems to think the later does.
The switch from porting games to OS X instead of Linux seem obvious to me: there is some chance of making money. While you can discuss the relative market shares of OS X versus Linux until the sun dies out, the two markets are quite different.
OS X users are used to pay for software, there is piracy of course, and open-source and free software, but companies, including Microsoft, are making money selling OS X software. Linux software by default is free.
Most linux users are using a x86 processor, and have thus a machine that can boot Windows. I have seen a lot of people using Linux for desktop use, but most of them have still some version of windows lying around, typically as a dual boot option or for running VM-ware. The only option for Mac users to run windows software is virtual PC and it is not suitable for gaming - not recent games at least.
If you consider this, making games for OS X makes sense, the people are more willing to pay, and you can sell them a more expensive version of the game (By the time a Mac version comes out, the PC version of said game is discounted). This would be a hard sell for Linux people, as most of them have the option of rebooting to play the cheap windows version. So you basically have a low volume version that has to be as cheap as the high volume version. Tough.
I know more people who use Linux than OS X (I work in a academic setting), yet I only know one Linux user who bought a Linux game, it was Heroes of Might a Magic by Loki, and he bought it because it was discounted (basically at the time Loki was going out of buisnes). Most OS X users I know have bought a few games.
I think there might also be technical issues (variety of sound system in different Linux distributions) or legal (lawyers of gaming companies being nervous about the GPL), but for me, those are secondary, the core issue is the target market.
Slashdot Mirror
The point is about able to do work inside the computer fast. Inserting memory can be done in less than a minute on an Apple case, it takes a least five if you have a generic no-name box. When you are upgrading a bunch of machines, it makes a difference...
The sad thing about intel doing their own stuff is that Open-firmware is here and standard. One of the most interesting ideas of having the F-code engine was to have processor independent drivers on the card. I.e you plug-in the card and it works, regardless of the fact the processor is PPC, sparc, or i686.
One funny trivia fact about Apple's open-firmware is that the firmware understands certain file-systems (HFS+,Ext2) and executable formats (PEF,ELF). The funning thing is the firmware does not understand OS X's executable format (Mach-O) so on every OS X machine, there is an ELF format bootloader.
From what I gathered, BeOS has Posix support and that's it.
The reason lightweight display managers are fast is because the graphical elements can be represented by simple sequences of fast drawing operations (i.e vectors), typically horizontal and vertical lines and boxes. In clear, the vectorial representation is compact and the rasterising fast.
So simply relying on vectors GUI elements will not help. If each the window close button is a vector image with thousands of vector operations it will not be fast.
Rendering advanced vector operations like splines is quite CPU intensive, and if you want to save on memory, you have to make sure that the rendering occurs late and deep in the graphical system, ideally on the GPU, or at least in the graphical server. This means supporting things like splines in the X Server.
Also note that the same logic can be applied to compressed bitmaps. If you can defer the decompression of JPEG data until it is in the graphical server or the GPU, you would save quite some memory. Apple uses such a trick to a certain extent, by RLE-compressing the window buffers.
My impression is that the second part of the cycle is weak in Linux. People do what is fun, optimising code, specially someone else's is not fun. In particular, people like to optimise their code for their spanking new machine so as to be able to boast impressive number. Also many people also think that optimising simply means compiling with -O2.
Of course another solving the problem also implies acknowledging that the problem exist, and look at how this could be solved. There was recently an article on slashdot on how to make an OS faster, look how far down you have to scroll to find some people who actually discuss how such tricks could be used for Linux instead of bitching that X is faster than Y...
My impression, is that one the main issues in Linux is redundant code and data. How many window drawing routines are lingering in memory? How many routines do decode bitmaps? How many toolbars icons with arrows?
Now you can argue that this metric is not a good one, and does not represent responsiveness, or that those optimisations are mostly geared towards desktop usage (and I would agree). The core thing is simply that the article presented interesting way of making a desktop more responsive.
Your reaction annoyed me, because dismissing interesting concepts because Apple only has to support limited hardware instead at thinking if and how those concepts could be adapted to Linux, is not the best attitude to improve Linux.
Only one optimisation presented is related to hardware drivers, and it is cache of what kernel extensions will probably be loaded. Most of the optimisations (basically lots of caching and dynamic defragmentation) could be implemented in Linux, regardless of the amount of supported hardware.
If you select, you can insert with the middle button as long as you have not selected something else.
I am not saying that select - insert is not a valid paradigm, simply claiming that is a more advanced form of drag drop seem quite a stretch to me. Additionally, this only works for text, while drag drop works for more advanced data types).
Myself, I'm still using a Laserjet 6MP that is still working well.
The problem in disconnecting devices like drives while the machine is sleeping is that the device has been shutdown (in particular if, like for an iPod it gets the power from the bus) and parked, but the filesystem has not been unmounted. The data is probably safe (the OS probably does a sync on the filesystem), but force-unmounting a filesystem can confuse applications.
A thing which I found interesting is that certain buses stay powered on, while others not (USB on, Firewire off). For instance, the buit-in USB port stay powered on, which make sense on a desktop as it is the mechanism to wake up the machine. On the other hand, I have a PCI card with additional USB ports, and this one gets powered off, so if you connect you keyboard mouse on this board, you cannot wake up the machine (it is logical in a way, still took me five minutes to figure it out).
Actually, I fully agree that certain software tools will evolve radically, but I don't think word-processors are among them. If you think about it, word processor were designed to prepare texts, letters and the nice feature the Mac Write paradigm brought us was WISWYG, but as paper become less the main media, this task will drift away. Now if you want to call 'word processor' a very advanced XML editor that is used as a front-end to a back-end that produces some printed outputs, we agree, except on the definition of 'word processor', e.g. I would not call Latex a word processor, but I agree this is open to discussion.
I should emphasise that the good things (drag-drop, contextual menus) are more related to changes in the OS than core changes in the design of Word. Actually I now realise that publish-subscribe also dissapeared.
Actually, Macwrite II was a very lightweight program, Microsoft Word for Macintosh had more features. Here is a list of improvement I have seen in Word since version 4.
Others features have disappeared since, like the latex-like equation code, the mailing subsystem and the integrated screen saver .
Some people are probably doing word-processing using really radical tools (vi?) but the overall design of the mainstream application is quite stable and will not change, the current model is good enough and what people are used to.
Maybe for some non-roman scripts, something completely different will be designed, and maybe back-ported for roman systems, but until then, word processors will look like Mac Write, spreadsheets will look like Visicalc, etc.
I suspect the system used to select adds is quite primitive, and I'm simply too strange a case to get tailored adds.
For instance could you build 'exclusive pool funds' which basically give control / ownership of the whole pool to one of fund owners based on some random condition? If this condition were random, you would get a lottery.
Somehow, if you consider a given stock as a random variable, you can 'code' some logic using basically less than and greater than operations with futures, does somebody know if can code more complex logic? I would think that you would need something that does mutual exclusion (only a few people win) and random (for instance by betting on the least significant 'bits' of a stock value).
If you take this analogy, the gaming industry can come out with a new iteration of the same game every five years. So while the gaming industry would probably never be wiped out, it would have to cool down a bit: there are not so many different game types, and the current upgrade frequency might not be sustainable once the novelty wears off.
If you think about it there is a huge variety of different movie types, video-game diversity is not that big. I don't think that video-gaming will go the way of the dodo, but there will be some adaptation from the current model. I hope this will bring out new interesting games.
You can run OS X without Quartz Extreme, in fact this is the case on older Macs who don't have graphics cards that are powerful enough. This simply taxes the CPU more and disables some fancy effects. Of course, if the graphic card supports the right OpenGL operations, Quartz Extreme will run.
With such algorithms people who only need to go down a few levels would use the stairs and not bother to call the lift and people who are really in a hurry would get the lift. Of course, the 'price' of the lift would increase at rush hours and go down during the night.
PHB would ask their secretaries to press the button multiple times for them. I suppose this would make a new job, lift button presser...
- More expensive than the Windows version.
- Less complete.
It is one thing to claim that if game XXX is ported to Linux, I will buy it and another to buy it when it comes out and is more expensive that the Windows version and doesn't have all the features.I recently bought Dungeon Siege for my OS X machine. It was more expensive than the Windows version (which was not so hot anymore and thus discounted), and some things are missing (the level editor). Still I paid for it, and I'm happily playing it.
I'm don't think that Linux users are cheap warez grabbing people and never implied this, but those I know are very cost-conscious and will resent buying over-priced, down-graded versions (especially when they can boot windows a play the first rate game).
Again this is anecdotal and based on the Linux users I know, but given the prominence of posts of Linux users on slashdot describing Mac users as idiots that buy overpriced stuff, I think this stereotype is quite common...
Porting a Win x86 game to Linux x86 is certainly technically easier that porting it to OS X - although I think you minimise the issues associated with Linux- technical issues are just that, technical issues that have to be solved, what is important is the money.The question is what makes better business sense, to do a port nobody will buy, or a slightly more difficult port that will be sold. TransGaming seems to think the later does.
- OS X users are used to pay for software, there is piracy of course, and open-source and free software, but companies, including Microsoft, are making money selling OS X software. Linux software by default is free.
- Most linux users are using a x86 processor, and have thus a machine that can boot Windows. I have seen a lot of people using Linux for desktop use, but most of them have still some version of windows lying around, typically as a dual boot option or for running VM-ware. The only option for Mac users to run windows software is virtual PC and it is not suitable for gaming - not recent games at least.
If you consider this, making games for OS X makes sense, the people are more willing to pay, and you can sell them a more expensive version of the game (By the time a Mac version comes out, the PC version of said game is discounted). This would be a hard sell for Linux people, as most of them have the option of rebooting to play the cheap windows version. So you basically have a low volume version that has to be as cheap as the high volume version. Tough.I know more people who use Linux than OS X (I work in a academic setting), yet I only know one Linux user who bought a Linux game, it was Heroes of Might a Magic by Loki, and he bought it because it was discounted (basically at the time Loki was going out of buisnes). Most OS X users I know have bought a few games.
I think there might also be technical issues (variety of sound system in different Linux distributions) or legal (lawyers of gaming companies being nervous about the GPL), but for me, those are secondary, the core issue is the target market.