Personally, I'd like to give my software away for free and still put bread on the table. I don't know if it's possible, but I like to think it is. And I'll be damned if I don't at least try.
Rock on my friend. I'm 18, gap year student, I get paid to write non free software. I'd rather work on Linux. I have some ideas on how we can make money, and write free software, in a way that scales to everybody. Email me about it if you're interested.
$300 a piece for the cheapest?? I don't see any mention of discounts. Even so, if your company can afford to give away that much stuff to its employees then tell us where you work!!
Where I work, I write software for all kinds of stuff, but get about 60p more per hour than when I was at the museum telling customers where the loos were. I have to work late each day next week to save up hours so we can have the afternoon off to go to a pub lunch. That is what we get for christmas. Nothing. Nada. No time off for the christmas lunch (which we are paying for). My dept isn't doing badly, but the others are finding it tough, hence the austerity measures.
yeah, that's possible i guess. On the other hand, I don't think using google has much to do with technical competence, all the people I know use it, and they range from can barely click a mouse to software engineers.
I find it hard to believe that 2% of the worlds computer using population could bias the google results so heavily.
Re:Am I out of touch...?
on
Web Zeitgeist
·
· Score: 2
There hasn't been any good pop music since Michael Jackson's Thriller.
Just how old are you Nine?:)
FYI Las Ketchup is a "craze" coming from Spain, 3 teenage girls who sing a silly song. Actually, I saw the video and it's not at all memorable, I'm not sure why it's got such attention. It's like that "I kiss you" thing a while back with the turkish guy. Just one of those wierd internet things you get now that everybody has instant communication across national boundaries.
Also to be noted in the google stats is Mac's broke 5% of the total searches for the first time since google started publishing stats. They omitted this from the year end results, but if you check the archives you can see this.
More news from the alt platform crowd too - Linux was the 4th top technology search, beaten only by MP3, SMS and.... winzip?
In contrast, xbox and Microsoft were at 7 and 9 respectively. Mac didn't feature (I wonder what position it was at).
I guess that means there's a lot of interest in Linux but it's not being turned into actual installed user base. I think we already knew that though...
simple, compile everything from source or then have several versions of the binary for different systems. But source for everything seems the best way. that way you don't even have to think of the binary format of the system.
Do you really think users are willing to compile all their software? Hell, I'm a developer and I'm not happy about it. Waiting a couple of hours for the latest and greatest Mozilla, or manually hunting down all the depended -devel packages etc, really, really sucks and gives a terrible impression to people who are new to Linux.
And of course for closed source software (ie games) it's not an option at all.
Binary packages are badly needed, for all those reasons and more.
Add a smart installer that can adapt the directory structure of different distros, plus an easy way to search for a certain package among all different package types.
Unfortunately, directory structure isn't actually the main problem with current packaging systems, there are a whole host of issues, several related to the actual oconstruction of the binaries themselves. Believe me, I should know about this, I've been building such a "smart installer" for the last 6 months, and we're still likely to be a year or two away from getting to version 1. Here's a rundown of the problems we face (the last one is the biggest so far):
Directory structures. Yes, this is an issue, for instance KDE apps expect to be installed into the same prefix as KDE itself, but on Redhat that's/usr, and on SuSE it's/opt/kde3 or/opt/kde2. It's easy to work around that however by having remapper modules that figure out where files should be placed.
Compiled in prefixes. This is related to point 1 - 99% of Linux software is built using autotools (which is where the name for my system comes from in fact:). Automake allows you to define macros that correspond to paths relative to the prefix, and a lot of apps use this facility to let them locate data files at runtime. Images, Glade user interface files and so on. Often programs, especially programs built with frameworks like GNOME or KDE, contain code like this:
result = open_file( DATA_DIR "/myfile" );
The C preprocessor expands and merges DATA_DIR with the next string, meaning that if you run configure with --prefix=/opt/kde3, DATA_DIR becomes "/opt/kde3/share/myapp" for instance. The app is then compiled, and the prefix is effectively burnt into the binary. This sort of thing happens a lot, and it means that resultant binary can then not be installed to a different prefix. There are two ways to solve that issue. The first is that on Linux it's possible to find out where the binary is located by reading/proc. You could then work from that path relative to the binary to locate the data files. Unfortunately proc is not particularly portable, and many "Linux" apps are in fact also UNIX apps. Libraries also can't use this trick. In autopackage we have a simple way around that problem - a small patch to a program allows it to use libprefixdb which lets the program discover various paths from a database (the installation database currently) at runtime.
C++ ABI breakage. This is currently causing pain, especially for games as they tend to be closed source and so you can't recompile them and they are often written in C++. GCC3.2 changed the c++ application binary interface significantly. If your distro was compiled with 3.2, so must the C++ apps you run. Because this change is recent however, most binaries are compiled for gcc2.95. By the time autopackage comes out, it's likely 3.2 systems will be in the majority and the network is planned to be able to cope with differences like this.
Link tree conflicts. This is one I was learning about last night, and is the real kicker. The problem is relatively simple: I have a game (any app really) that links against, say, libpng and libSDL. libpng comes in two flavours, 2 and 3, reflecting an ABI break. Normally we'd say, so what, and throw both onto a system, libtool versioning allows it to work. The game is linked against libpng3.
BUT! libSDL also links against libPNG for its own use, and the distro compiled it against libpng2. Now when the app loads, there will be 2 versions of libpng in the same address space and you will get symbol conflicts. Best case scenario - the app segfaults on startup. Worst case, it segfaults mysteriously during execution.
There is no good way to solve that problem except by recompiling the apps and hoping 2 and 3 are source compatible. Alternatively of course, recompile the whole distro against version 3.
I'm not making this up, it happened with Mandrake:(
Unfortunately it appears the only way to really solve this problem is to rewrite the linker logic (in ld.so) so it doesn't merge all the symbols into one large hash chain, instead it progressively maps, performs fixup and then drops the symbol tables. We're still figuring out how this impacts stuff like lazy binding, dlopen() and so on.
Finally, there are a whole host of other issues, like the different naming schemes used by distros, the different installation mechanisms, incompatible install-info scripts and the difficulty of resolving dependancies in a way that still integrates well with the system. Throw in the needs of corporate/enterprise deployment scenarios, and you see what a challenge it has become
As you can see, binary distribution of packages alone is nightmarishly complicated.
The rest of your points are kind of out of date, transparent Wine is a reality, as is seamless driver installation, as is partition management (actually some distros don't even ask that, they have "smart" allocators that guess how much space you'll want for linux and leave the rest to windows). Lindows can do direct Windows email/address book/bookmarks imports. Windows installer apps are dealt with by Wine and run just like they would in Windows.
Not to be a naysayer, but in 12 years Linux has managed to gain only a few percentage points worth of the desktop market. Users really don't care, don't know, and have no reason to be aware of the development model used to create their software.
And they won't be. I think you're missing a fairly crucial point which is that the desktop Linux effort only really started in '96 with the launch of KDE. In '98 KDE1 was released, and that's when the ball started rolling. That means the linux desktop has in effect been in existance for 4 years now.
In that time, KDE and GNOME have gone from ugly, unstable and primitive desktops into powerful, beautiful and yes, in the case of GNOME2 even usable desktops. Not only that, but a truckload of applications have been developed, installation of the OS has become childs play and an open standards effort has been started to unify the interfaces between desktop components.
That's a lot of progress.
And, it will not happen if too many Linux developers continue to imagine that their development model is what they're selling. It isn't.
Given that Linux has never been marketed as such, it's only ever grown through word of mouse, I think there is sufficient interest in not just the technology but also the development model.
Business in particular is keen on the idea of ridding themselves of vendor lockin, being in control and being able to easily maintain old software if the original vendor/maintainer no longer carries on.
Actually, I'd say most users are cheap AND lazy. If they can get something as good as what they've got, for less, they'll go for it.
I think there's one scenario in which Linux could become a viable desktop alternative.
And why exactly does a big company have to do this? KDE and GNOME are building all we need to make an easy to use desktop. They are being helped by companies, and of course people like the guys at Xandros bring it all together and make it a cohesive whole, but the idea that after all that's been achieved, of course only a big powerful corp can make Linux workable on the desktop is ludicrous.
Never forget we're in this mess in the first place because big and powerful corps don't throw money into the black whole that is OS development. Not even Apple - if they didn't sell hardware with huge margins, OS X wouldn't even exist, and they're finding it hard going as it is.
There's one big thing, I think, that will prevent this from happening: the GPL
How many times does this have to be thrashed out on slashdot? It's worse than the "X is slow" thing - look at Xandros: XFM is entirely proprietary, yet they do not violate the GPL, and they still give code back to the community. The idea that the GPL makes something uncommercialisable should have disappeared 5 years ago, but still it persists, despite the existence of companies who've been around for years and make money out of free software.
Of course, Apple already did exactly this. They just based their OS on FreeBSD instead of Linux. So the idea is sound; it's just that in GPL-land, there's no possibility of commercial motivation, which means no reason to invest the necessary time and work.
Please, this is just pure FUD. IBM, RedHat, SuSE, Xandros, Sun, TiVO, Sharp etc are not basing their products on Linux because their top execs are all high on crack.
This whole post is just a total troll. Every point has counter examples. It rolls the Linux Desktop up with the BSD vs GPL, along with a healthy dollop of FUD as well.
Exactly what the more rational and less rabid of us have said all along. Linux has its place, which for most people is not the desktop.
What, you mean, the less optimistic and more cynical of us have said?
Today people say, Linux has it's place, but it's not the desktop. Yesterday they said, Linux has it's place, and it's running servers, not on big iron or in embedded devices. The day before that, it was Linux is a cool toy for hackers, but the real world will never use it.
Look to the future my friend! Man, I'm glad the KDE and GNOME hackers don't take that approach, otherwise it'd be a self fulfilling prophecy.
believe me, it's not. In fact you CAN look at large pieces of M$ source, just go download a DDK
Or Wine. No, it's not the MS code, but it is a faithful reproduction, function for function, bug for bug. Wine also has a very similar internal architecture (modelled after NT). For the purely curious, looking at Wine is as good as the real thing in terms of educational value. Obviously it's just the API implementations rather than Explorer et al, but good enough for some uses.
Actually, it's known that the NSA have their own keys in ADVAPI.DLL, which provides crypto services to Windows, so you don't have to see the source, just go read some papers from the security community.
There are free IDEs (Eclipse, Forte), and there are free UML tools (ArgoUML, probably others)
Indeed, but what makes Delphi so unique and special isn't so much the IDE (though the IDE is excellent), it's the compiler.
Delphi is fast. I mean, really fast. No other compiler comes close. To put things in perspective, when Borland first added multiple error reporting to the compiler (ie one compile would report more than one error) I didn't understand why they did it. I hadn't seen javac at that point. I didn't understand, because it was actually faster for me to press Ctrl-f9 to trigger a recompile in order to move to the next error, than it was to move my hands from the keyboard to the mouse and back again.
Their compiler is that fast. It can do a project with over 100,000 lines of code I have sitting here in less than 8 seconds. The resultant binaries are tight. When I tried my first C++ program, I was astonished at how long it took to compile as it read in all the headers etc. I was sure I must have done something wrong.
Part of the reason it's so fast is just long history, Borland have had a lot of time to optimize it, but the other was the language design. Object Pascal is designed for fast compiles. For instance, it doesn't use headers, but each compiled unit (.dcu ->.o) included header information with it, meaning it's insanely fast to link them together. It also has excellent remake logic, if you only touched one file, only one file was recompiled. There is no [preprocessor, so the compiler can be single pass.
2) If Codeweavers was to reverse engineer it, Microsoft could claim that their implementation was circumventing an access control and take them to court under the DMCA. Moreover, since Codeweavers actually sells these products, they could actually be brought up on criminal charges.
So get somebody in Hungary to do it. Any non american in fact.
Would CodeWeavers get sued? I don't think so, because it's to do with interoperability, not circumvention. They are trying to faithfully recreate the Office XP product activation stuff for instance.
Essentially, Europe's insistence on one standard, which worked nicely for 2G, screwed the pooch raw with 3G, that, and the fact that Asia and the US didn't license out 3G, so European cell carriers had to take on debt for billions for 3G whereas no one else did.
Actually the European carriers screwed themselves. The airspace was sold off in auction, and they bid each other up to stupid amounts. Having signed the deals during the height of the dotcom bubble, they're now stuck with it. Kind of a shame, but equally they did sign a contract with the government - I don't think they should be let off because they were stupid.
The estimates for increases in Linux desktop share are, I would imagine, far lower than the number of people expected to move to 3G mobile system in the next 3 years.
Well, that's what the telcos would have us believe, but from what I've gathered most 3rd party analysts aren't so confident. So far, 3G providers (in europe anyway) are horrifically loaded with debt and are desperately trying to find customers who are willing to pay huge amounts for what are essentially gimmicks.
There are loads of adverts here telling us about all the funky things you'll be able to do with 3G phones - maybe, oneday. Everybody takes the piss out of them. 3G is by no means a certain bet.
Compared to Linux which has a huge amount of interest, and a lot of people saying "Well, I would use it but there's still 1 problem left......" I think it's pretty likely that Linux will grow faster than 3G in the next few years, simply because it has a lot more people interested in it.
Don't get me wrong, 3G will happen, but not after some abortive starts and it'll take a lot longer than we think.
Well, depends on how you define ease of use. Installing IE6 into WineHQ for instance is not just hard, unless you have good knowledge of the Windows registry and are happy manually moving keys around etc, it's simply not possible (there is no reboot logic in WineHQ).
I've found I get much better results with CrossOver than WineHQ, even though they are largely the same. Hopefully soon WineHQ will have caught up, but they are currently focussed on technical targets.
I thought the Windows discount actually worked by massively increasing the price of Windows over market value if machines were sold with other operating systems? The "discount" brings it down to a little bit below the retail price.
I mean, I can't see Microsoft selling Windows for anything anywhere near $20.
Slashdot Mirror
Rock on my friend. I'm 18, gap year student, I get paid to write non free software. I'd rather work on Linux. I have some ideas on how we can make money, and write free software, in a way that scales to everybody. Email me about it if you're interested.
A RealPlayer you say? Are you sure about that? :)
A rather fine piece of software. I should use it :)
$300 a piece for the cheapest?? I don't see any mention of discounts. Even so, if your company can afford to give away that much stuff to its employees then tell us where you work!!
Where I work, I write software for all kinds of stuff, but get about 60p more per hour than when I was at the museum telling customers where the loos were. I have to work late each day next week to save up hours so we can have the afternoon off to go to a pub lunch. That is what we get for christmas. Nothing. Nada. No time off for the christmas lunch (which we are paying for). My dept isn't doing badly, but the others are finding it tough, hence the austerity measures.
I find it hard to believe that 2% of the worlds computer using population could bias the google results so heavily.
Just how old are you Nine? :)
FYI Las Ketchup is a "craze" coming from Spain, 3 teenage girls who sing a silly song. Actually, I saw the video and it's not at all memorable, I'm not sure why it's got such attention. It's like that "I kiss you" thing a while back with the turkish guy. Just one of those wierd internet things you get now that everybody has instant communication across national boundaries.
More news from the alt platform crowd too - Linux was the 4th top technology search, beaten only by MP3, SMS and .... winzip?
In contrast, xbox and Microsoft were at 7 and 9 respectively. Mac didn't feature (I wonder what position it was at).
I guess that means there's a lot of interest in Linux but it's not being turned into actual installed user base. I think we already knew that though...
Do you really think users are willing to compile all their software? Hell, I'm a developer and I'm not happy about it. Waiting a couple of hours for the latest and greatest Mozilla, or manually hunting down all the depended -devel packages etc, really, really sucks and gives a terrible impression to people who are new to Linux.
And of course for closed source software (ie games) it's not an option at all.
Binary packages are badly needed, for all those reasons and more.
Unfortunately, directory structure isn't actually the main problem with current packaging systems, there are a whole host of issues, several related to the actual oconstruction of the binaries themselves. Believe me, I should know about this, I've been building such a "smart installer" for the last 6 months, and we're still likely to be a year or two away from getting to version 1. Here's a rundown of the problems we face (the last one is the biggest so far):
result = open_file( DATA_DIR "/myfile" );
The C preprocessor expands and merges DATA_DIR with the next string, meaning that if you run configure with --prefix=/opt/kde3, DATA_DIR becomes "/opt/kde3/share/myapp" for instance. The app is then compiled, and the prefix is effectively burnt into the binary. This sort of thing happens a lot, and it means that resultant binary can then not be installed to a different prefix. There are two ways to solve that issue. The first is that on Linux it's possible to find out where the binary is located by reading /proc. You could then work from that path relative to the binary to locate the data files. Unfortunately proc is not particularly portable, and many "Linux" apps are in fact also UNIX apps. Libraries also can't use this trick. In autopackage we have a simple way around that problem - a small patch to a program allows it to use libprefixdb which lets the program discover various paths from a database (the installation database currently) at runtime.
BUT! libSDL also links against libPNG for its own use, and the distro compiled it against libpng2. Now when the app loads, there will be 2 versions of libpng in the same address space and you will get symbol conflicts. Best case scenario - the app segfaults on startup. Worst case, it segfaults mysteriously during execution.
There is no good way to solve that problem except by recompiling the apps and hoping 2 and 3 are source compatible. Alternatively of course, recompile the whole distro against version 3.
I'm not making this up, it happened with Mandrake :(
Unfortunately it appears the only way to really solve this problem is to rewrite the linker logic (in ld.so) so it doesn't merge all the symbols into one large hash chain, instead it progressively maps, performs fixup and then drops the symbol tables. We're still figuring out how this impacts stuff like lazy binding, dlopen() and so on.
As you can see, binary distribution of packages alone is nightmarishly complicated.
The rest of your points are kind of out of date, transparent Wine is a reality, as is seamless driver installation, as is partition management (actually some distros don't even ask that, they have "smart" allocators that guess how much space you'll want for linux and leave the rest to windows). Lindows can do direct Windows email/address book/bookmarks imports. Windows installer apps are dealt with by Wine and run just like they would in Windows.
And they won't be. I think you're missing a fairly crucial point which is that the desktop Linux effort only really started in '96 with the launch of KDE. In '98 KDE1 was released, and that's when the ball started rolling. That means the linux desktop has in effect been in existance for 4 years now.
In that time, KDE and GNOME have gone from ugly, unstable and primitive desktops into powerful, beautiful and yes, in the case of GNOME2 even usable desktops. Not only that, but a truckload of applications have been developed, installation of the OS has become childs play and an open standards effort has been started to unify the interfaces between desktop components.
That's a lot of progress.
And, it will not happen if too many Linux developers continue to imagine that their development model is what they're selling. It isn't.
Given that Linux has never been marketed as such, it's only ever grown through word of mouse, I think there is sufficient interest in not just the technology but also the development model.
Business in particular is keen on the idea of ridding themselves of vendor lockin, being in control and being able to easily maintain old software if the original vendor/maintainer no longer carries on.
Answering only one throwaway sentance is out of character for you Twirlip - where are the responses to my actual points?
Net based movie piracy only became practical once DVD encryption was cracked, as they could then be recoded using DivX which made transfer feasable.
Remember you could pirate CDs before MP3, just that nobody did.
I think there's one scenario in which Linux could become a viable desktop alternative.
And why exactly does a big company have to do this? KDE and GNOME are building all we need to make an easy to use desktop. They are being helped by companies, and of course people like the guys at Xandros bring it all together and make it a cohesive whole, but the idea that after all that's been achieved, of course only a big powerful corp can make Linux workable on the desktop is ludicrous.
Never forget we're in this mess in the first place because big and powerful corps don't throw money into the black whole that is OS development. Not even Apple - if they didn't sell hardware with huge margins, OS X wouldn't even exist, and they're finding it hard going as it is.
There's one big thing, I think, that will prevent this from happening: the GPL
How many times does this have to be thrashed out on slashdot? It's worse than the "X is slow" thing - look at Xandros: XFM is entirely proprietary, yet they do not violate the GPL, and they still give code back to the community. The idea that the GPL makes something uncommercialisable should have disappeared 5 years ago, but still it persists, despite the existence of companies who've been around for years and make money out of free software.
Of course, Apple already did exactly this. They just based their OS on FreeBSD instead of Linux. So the idea is sound; it's just that in GPL-land, there's no possibility of commercial motivation, which means no reason to invest the necessary time and work.
Please, this is just pure FUD. IBM, RedHat, SuSE, Xandros, Sun, TiVO, Sharp etc are not basing their products on Linux because their top execs are all high on crack.
This whole post is just a total troll. Every point has counter examples. It rolls the Linux Desktop up with the BSD vs GPL, along with a healthy dollop of FUD as well.
What, you mean, the less optimistic and more cynical of us have said?
Today people say, Linux has it's place, but it's not the desktop. Yesterday they said, Linux has it's place, and it's running servers, not on big iron or in embedded devices. The day before that, it was Linux is a cool toy for hackers, but the real world will never use it.
Look to the future my friend! Man, I'm glad the KDE and GNOME hackers don't take that approach, otherwise it'd be a self fulfilling prophecy.
Or Wine. No, it's not the MS code, but it is a faithful reproduction, function for function, bug for bug. Wine also has a very similar internal architecture (modelled after NT). For the purely curious, looking at Wine is as good as the real thing in terms of educational value. Obviously it's just the API implementations rather than Explorer et al, but good enough for some uses.
Actually, it's known that the NSA have their own keys in ADVAPI.DLL, which provides crypto services to Windows, so you don't have to see the source, just go read some papers from the security community.
Indeed, but what makes Delphi so unique and special isn't so much the IDE (though the IDE is excellent), it's the compiler.
Delphi is fast. I mean, really fast. No other compiler comes close. To put things in perspective, when Borland first added multiple error reporting to the compiler (ie one compile would report more than one error) I didn't understand why they did it. I hadn't seen javac at that point. I didn't understand, because it was actually faster for me to press Ctrl-f9 to trigger a recompile in order to move to the next error, than it was to move my hands from the keyboard to the mouse and back again.
Their compiler is that fast. It can do a project with over 100,000 lines of code I have sitting here in less than 8 seconds. The resultant binaries are tight. When I tried my first C++ program, I was astonished at how long it took to compile as it read in all the headers etc. I was sure I must have done something wrong.
Part of the reason it's so fast is just long history, Borland have had a lot of time to optimize it, but the other was the language design. Object Pascal is designed for fast compiles. For instance, it doesn't use headers, but each compiled unit (.dcu -> .o) included header information with it, meaning it's insanely fast to link them together. It also has excellent remake logic, if you only touched one file, only one file was recompiled. There is no [preprocessor, so the compiler can be single pass.
Already???? Well, no offence to Tim, but it's got some way to go before it's SourceForge scale ;)
I can't see the interview either - maybe something is up with my connection...
So get somebody in Hungary to do it. Any non american in fact.
Would CodeWeavers get sued? I don't think so, because it's to do with interoperability, not circumvention. They are trying to faithfully recreate the Office XP product activation stuff for instance.
Actually the European carriers screwed themselves. The airspace was sold off in auction, and they bid each other up to stupid amounts. Having signed the deals during the height of the dotcom bubble, they're now stuck with it. Kind of a shame, but equally they did sign a contract with the government - I don't think they should be let off because they were stupid.
Well, that's what the telcos would have us believe, but from what I've gathered most 3rd party analysts aren't so confident. So far, 3G providers (in europe anyway) are horrifically loaded with debt and are desperately trying to find customers who are willing to pay huge amounts for what are essentially gimmicks.
There are loads of adverts here telling us about all the funky things you'll be able to do with 3G phones - maybe, oneday. Everybody takes the piss out of them. 3G is by no means a certain bet.
Compared to Linux which has a huge amount of interest, and a lot of people saying "Well, I would use it but there's still 1 problem left......" I think it's pretty likely that Linux will grow faster than 3G in the next few years, simply because it has a lot more people interested in it.
Don't get me wrong, 3G will happen, but not after some abortive starts and it'll take a lot longer than we think.
I've found I get much better results with CrossOver than WineHQ, even though they are largely the same. Hopefully soon WineHQ will have caught up, but they are currently focussed on technical targets.
Or rather, OS/2 is dead.
Rest in peace.
I mean, I can't see Microsoft selling Windows for anything anywhere near $20.