There is no such thing as an "Apple Tax". Darwin is Open Source and so it could never be a "Tax" that you have to pay against your will. IF you want to buy apple software too, like the Aqua UI, or.Mac, then you can-- but you don't have too.
uh huh. And how many people do you know that run pure Darwin, without the rest of OS X, on Intel hardware? Not many. Darwin is nice, but it's nothing amazing. We already have too many free unix kernels.
And the important point is that by charghing for their UI and their software, and by making hardware, Apple is able to fund desktop unix in a way that nobody has been able to for Linux. Rad Hat has done an admirable job-- and most of their work and apple work compliments each other.
That is bollocks. You are making a fundamental mistake here, which is assuming that the desktop is like Darwin. Red Hat and Apple are not at all compliments of each other, as the software RedHat fund is given away under the GPL - not so for Apple. They are competitors. GNOME and MacOS are not at all complimentary, as GNOME exists to be a good desktop (let's not argue about whether it is or not), and the MacOS desktop exists to make Apple money.
It is disingenuous to compare this to a company that charges people for its software whether they want it, use it or buy it, or not. Apple only charges people who freely choose to buy its software. Calling it an "apple tax" only undermines your creditability.
In theory, only people who want to pay for Windows. Doesn't really work that way does it? If Apple were at 90% market share tomorrow, you can bet it'd be an Apple tax. Just because they are small and largely insignificant NOW doesn't mean they should be supported - at the end of the day, they stand for the same thing as Microsoft: profit through closed platforms.
Therefore, if you really believe in choice and freedom, you MUST support Apple and Darwin. Doesn't mean you have to buy their software or hardware-- but opposing it is opposing freedom.
Bullshit. Most people couldn't care less about Darwin, the UNIX kernel has been done to death already. The thing that interests people is MacOS, you know, that thing that is proprietary and runs only on Mac hardware. What are you smoking? If you really believe in choice, then why the hell are you supporting a company that makes its money from not giving its users choices in the same of simplicity?
Well now. There is an interesting thread above about money and open source software. The core problem seems to be that you can't pay people to write code and then give it away for free, unless you make money on consulting/support instead. The CEO of BitMover appears to believe that you can only sell support for a broken product - this is hardly true, but he's right in that if your product is not one that is complex enough to need commercial support, you can't make money this way.
There's a simple solution to this dilemma, which is, don't make your products open source if you want to make money out of it. Free software is great for writing operating systems, but only Stallman has ever claimed it is the be all and end all of software development. Note that you can write open software without giving away the source, simply by documenting the file formats and protocols. I don't respect companies that don't do this anyway, as it implies that they feel they need artificial lockin to stay afloat rather than just producing quality software.
Having said all this, I have a problem with Tom Lord asking for donations, and ditto for Rob Levin with openprojects.net. There are countless open source projects in the world, many of which are very important. The Linux kernel, KDE and so on are all huge projects, yet I don't see them begging for cash. I also write open source software, but I do it in my spare time, and delegate work that I cannot handle, because my projects are by necessity non commercial. No project should be so dependant on one person that they have to work on it full time. This goes for writing source control systems, or running IRC networks. I think projects should either be non commercial, in which case you have a paying job during the day and work on it in your spare time, or you figure out a way of making money from it (ie by keeping the source closed).
I don't see any good way, or any good reason, for attempting to make money directly from donations for open source projects. BitMover has got the right idea, they are getting mindshare and free testing by giving away their product to free software developers, but charging for it for commercial operators. They've figured out a way to tread the line, but most don't.
I, for one, would not be willing to take a Mac user out to lunch. Why? Because I'm one of those irritating people who believes openness matters to platforms. When somebody invests in a platform, they are supporting it. Mac users I've found have a tendancy to try and sell all their friends Macs as well, even people who don't have shares in Apple. This is doubly the case for developers, as when they switch to a platform, they may well write software for it, so further supporting the platform.
Unfortunately, for all its good points, the Mac is a proprietary platform. I see a lot of confusion about this in these forums. People say things like, it's based on Darwin which is open source, so OSX must be an open platform. Wrong. There is a simple test to see if a platform is open or not, namely, can make my own version and then sell it. With a Windows PC, you can get partially there, as you can build your own PC and sell it, but you can't do that for the OS. A Windows PC is semi-open.
Could I make my own Mac? No, I could not, as the hardware is closed. Nobody but Apple may make Macs, or even computers that look like Macs, as the "brand" is one of their most valuable assets. OK, so let's say I tried to clone OS X. Could I? No, I could not. I would get as far as downloading and customising Darwin, and then.... oh, whoops. It seems the rest of OS X is not open. I couldn't even recreate the artwork without getting sued. Yes, I know you can compile Linux apps on MacOS, but you can do that on Windows too. I don't see people telling me to take Windows users out for lunch.
So the Mac is proprietary. It dismays me to see people supporting such a platform. We've all, every one of us, learnt the hard way what happens when a company gets control of an important platform. Are people really willing to throw away not only openness of the OS, but also the hardware, merely to get a colour co-ordinated desktop? And for those who believe the 3 platforms could exist in peaceful cooperation, I have bad news: proprietary platforms have a tendancy to get a tight grip, as people get locked into them. Mac apps cannot be easily ported to Linux for instance - the flow is one way only.
In short: has nobody learnt anything from Microsoft? Do we really have to learn the same painful lessons twice, once from Gates, and once from Jobs?
I got quite a long way with it - on that page are ideas for how such a musical marketplace could work, how quality would be ensured (ie how do you sort the wheat from the chaff without record companies A&R depts) and answers to questions musicians would commonly have.
I can so see this happening. Music is, in a way, a parallel of the software world. Music is effectively information and can be replicated and copied for zero cost. It's dominated by a rich and powerful entity (the RIAA vs MS), and the world is crying out for something better. It's ripe for the gift economy to be established here.
What Jabber needs is a single, absolutely solid Windows client. Every few months, I have a look for one, and always come up short. The most common "easy" client is probably JIM, made by Jabber Inc. Unfortunately, it's buggy, leaks resources (how?!? it's in delphi!) and has "enterprise" features that are of no use to my friends. It also hasn't been updated in a dog age.
The best right now is probably Rival, which is coming along nicely, but unfortunately isn't finished yet. It's also not open source, dunno why not as it's freeware. It's written in Visual Basic too unfortunately, occasionally when I've been talking with Dan who makes it about a feature, he's told me simply "VB can't do that" - and that's the end of it:(
If somebody was to make a really solid, easy to use client that can compete with MSN, Trillian and so forth, Jabber could take off. All of my friends are on MSN but there's no way I can "convert" them to Jabber unless the clients are solid.
After that - well, the network has been stable for a while now (as long as you don't want AIM, but that network doesn't have much of a presence in the UK). I run the theoretic.com server with Theo, a friend of mine, and it's got features like IM headlines (from slashdot:p) and soon an integrated weblog. The features are there, too bad the clients are not.
If said data was regulated by an open protocol, you could probably achieve much the same kind of thing; however, MSFT is a demonstrated monopoly, and as such can dictate a data-transfer protocol and make it a defacto standard.
Indeed, and that protocol is called.NET Remoting, and is a set of extensions to SOAP which allow for true object serialization a la CORBA. SOAP is useful now, but one thing I never understood until a few months ago is why they left out what's known as object-by-reference, which lets you pass objects around (pretty useful). Then I remembered that SOAP was partially designed by MS. I can just see the meetings:
Microsoft: "No no, we don't want objects-by-reference, we want to keep the spec clean and lean"
only to then turn around and add it as their own extension to SOAP. Now SOAP is just XML so it'd be pretty easy to reverse engineer said extensions, but who says it's not patented?
Well I didn't cite because I saw these figures buried in one of the KDE mailing lists and didn't have the link handy. Also, you're right about runtime speed but I was talking about link speed (to do with startup time) which it doesn't appear to improve at all. Most people don't have a beef with the runtime speed of Linux apps (which is generally good), they don't like the amount of time C++ apps take to start.
Maybe this was just me, but when I last tried Kylix 2 I found it incredibly slow and buggy. This is such a shame, as Delphi is excellent. I put this down to it being a winelib app, is this still the case? Can anyone who's used it for serious work tell me how it holds up?
The most important difference is that 3.1 does a much better job at c++, which makes a huge difference in KDE
In theory yes, in practice unfortunately it seems that it doesn't. I saw some statistics for 3.1, it takes TWICE as long to compile, and the resulting binaries are slightly slower at linking. The reason KDE seems slow is due to start up time, caused by inefficient linking, so 3.1 doesn't seem to improve KDE at all. Apparently it now requires an improved glibc, but I may be wrong here.
That's basically it, in a word. Remember at all times, that it's your project, not theirs. If somebody submits a patch you don't like, tell them so - but (and this is important) make sure you have good reasons for not liking it.
You say, what if a patch goes against my "mission". I'm not sure what you mean here, other than perhaps you have an instinctive feel that something is wrong. I say, don't worry about it. For starters, most projects debate patches before work starts on them, so you can usually veto an idea before it gets turned into code. Use this opportunity to steer things towards your vision.
One last tip: don't release too early. If what you're writing is complex, make sure the fundamental architecture is there, and make sure the rest is basically filling it out and hacking on it. If you just announce a vague idea, people are going to get involved who have very different ideas to you, and you'll either get pushed around or engaged in a flamewar. If you ensure you have the fundamental design there first, then people who don't like the design will push off, and those that do will (hopefully) stay to help.
Like said in other posts - the "friendlyness" and "easyness" should scale up to the advanced level - where everything is open. Claiming that spewing config files all over my home dir would be better than putting them in e.g. ~/config just because you can hide it is silly.
Well once again, putting all config files into ~/config (or rather ~/.etc) would be nice... but converting all the apps would be a lot of effort and all that would have changed is some file system locations.
Now that's a gem. You never use the filesystem layout on your systems? Would YOU want to use the system never seeing the filesystem? Or is this exlusively for the stupid newbies? Come on... in your vision of a newbie system the paperclips aren't that far off - and just admit that you would never run such a system without an "advanced" switch that let you disable all your nice UI gadgets.
Huh? I'm lost, where did paperclips come into it? I use the filing system a little yes, because I'm currently developing an installation system. Other than that, I rarely access the / filing system directly unless I need to edit an obscure config file for which there is no GUI. I never access the rest of the filing system from a graphical browser, because they are designed to hide all that from me, whereas the command line is not.
And I don't think of myself as a leet superwhizzkid. I think of myself as a Linux user.
it's hardly ever possible to guess all of the places unix installs go - that without even providing that service of adding things to your menus.
Hmm, a lot of these gripes seem to be do to with the package management system - I share these gripes by the way, which is why I'm writing a software manager that can deal with adding menu options etc. Even with RPM though, you never need to guess where things go, you can query the package. Or use something like Red Hat, which dumps all the software into/usr
That's a big exagguration. Config files are placed in/etc or your home directory. I believe KDE/GNOME apps have their own system which doesn't use etc, but I can't speak for them. Yes, occasionally a badly behaved program "invents" its own etc directory, but badly behaved programs aren't going to disappear anytime soon.
I can't really think how you'd change the filing system layout to make it massively better without, well, changing Linux as a whole./var isn't particularly descriptive, but it's used for a whole range of stuff. What would a better name be?/etc is an odd name too, but once you know etc is where config files are stored, you know it. Renaming it to/config would be a lot of work, for purely aesthetic gain.
Linux would do much better from a redesign of some of its key parts all at once, for instance creating a decent object model, and fixing the mess that is autoconf/automake. Which will happen some day, I have no doubt, but for now it's not a major barrier to adoption.
Linux is staggering under the incipid design of the X windowing environment. X has created more problems than it has solved and it has forced developers into all sorts of application kludges and likewise forced users to expect trouble.
The incipid design? That's a new one. What exactly is wrong with X? Actually, it's pretty similar to most other platforms graphics engines, except most other platforms don't have network transparency. Or what, do you think GDI+ is a model of good design?
My dream? All those hard working XFree86 developers design a new system from scratch. They're very talented and hard working. They could easily produce something MUCH better than X if they even gave it half a try.
A few people thought that same thing, and started Berlin. And it's pretty cool. But nobody uses it, and almost nobody develops for it. Why? Because X works, and it's here, today. Berlin, for all its futuristic architecture, has its own problems. Nothing is perfect - I don't get why people are so hung up about the graphics API. X is not the problem of Desktop Linux.
Sorry for the depressive info, but if you wanna make anything even remotely "friendly" or "easy" to Joe User using linux, you need to make major changes to the filesystem layout.
No no no no NO! First off, Windows is not more organised than Linux. Far from it. Apps frequently install themselves god knows where, usually in Program Files, but sometimes also in directories off the root, sometimes people install apps into "My Documents" (I have seen this!) often if you have multiple partitions there is more than one Program Files folder etc. It's a mess.
And what's this about apps being spread out everywhere? Windows apps usually install stuff into their directory, the windows or windows/system directories, the registry, sometimes they add stuff to your Profile and they add shortcuts. I don't hear anybody complaining about that.
Joe is going to have a lot of questions rather quickly. For instance, why isn't there user stuff in/usr? Who is/usr/share shared with? What's optional with/opt and why isn't the rest optional? And why is my home directory full of config files if config files go into/etc? And why are there at least two */bin dirs (containing not only binaries but other runnable files btw)?
Joe user will never need to know the answers, or even ask the question in an easy to use distro, for the same reason that nobody asks "Why does configuration data go into HKEY_LOCAL_MACHINE but also HKEY_CURRENT_USER?" - the registry is never accessed by users, and ditto for the Linux file system.
Say what you want about the Windows registry, but at least it's not laying around in plain view in Joe's home directory.
Why should the root file system be? At the moment, when I start Konqueror I don't even see the root system unless I ask for it. It's not very polished, but that can be fixed. MacOS has all these directories too, but the user never sees them. There's no need to rename them, it's just a case of good UI.
If you're truly going for an easy-to-use idiot-friendly linux, you're going to have to take some tough decitions. Toss the old layout out the window, pick something like/apps,/config,/system,/documentation, or whatever - and spend a long time compiling stuff from scratch to make it work.
That would be nice, but it would bring only one clear advantage - nicer names. I can't think of any compelling reason to do this at all, considering that these are paths normally only used by software. With proper software management, the user need never touch these directories, so why bother changing them?
But the world outside of us is that people do NOT want or need or know how to change the program's behavior, all they want is that the program does what they want - whatever they want. That's why we have NORTON UTILITIES for Windows, and there's none of Linux.
Norton Utilities is mainly made up of programs that either a) undelete things, b) fix problems with windows or c) optimise things.
All of these are features of the OS, not a utility pack! Modern Linux filing systems optimise themselves, and it doesn't state on their website what sort of probems Norton solves exactly with windows. Undelete? Isn't that what the recycle bin is for?
We have Kylix from Borland (FREE !) and how many of us are using Kylix to develop USEFUL UTILITIES for the users ?
None of us, because Kylix sucks. This is such a shame, because Delphi is excellent, a truly wonderful program that I've used to develop utilities before. Kylix is primarily meant for GUI apps though, and I think you'll find most Linux utilities are written as command line apps.
Mod me down as a troll if you must, and it's sort of off-topic because the article here is talking about Linux on x86, but I've thought of "Linux on the Desktop" as total ass myself for a couple of years now. Now we get two articles in two weeks saying as much. Which is exactly why I've been working hard (and finally succeeded) to get OS X running on my old Power Mac instead of putting Yellow Dog or Debian on it.
LOL! Like you'd get modded down for praising the Mac around here.
First, XFree was a pain in the ass to get set up. I haven't tried it since 4.x, but 3.x sucked because all the setup programs wanted to compute "optimum" modelines for your monitor and display card, which inevitably never worked for me. This instead of what I wanted: resolution and refresh, from the list of VESA standard modes.
Then there's getting the desktop environments running themselves. I didn't get very far on them, but in my experience, if you didn't pick the window manager favored by the distro, the others simply weren't configured to do anything useful. The only way to get menus to contain anything useful seemed to be by editing config files, and by this time I wasn't in any mood to search for more damn config files to edit.
a) You're wrong. I don't know which distros you tried, but I've never had to edit config files to edit menus (this is SuSE 7.3)
b) You're right. Desktops don't share applinks currently, so if you install a KDE app, it may not play nice with the GNOME menu and vice-versa. There is currently a standardisation effort underway to fix this (and it'll provide a far more advanced menu system than any other OS to boot). Check out freedesktop.org for more info.
It just works, without a bunch of tweaking, partly because Macs have nowhere near the hardware nightmare that exists in the x86 world.
Yes yes, we know. "It just works" - what a good slogan. People seem to miss the point about Linux: IT IS FUNDAMENTALLY DIFFERENT. It relies on people creating things themselves, them on users liking it and adopting it, then on people forming standards etc. There is no "Steve says we do this, so we do this" policy. That has the advantage of the fact that it scales reliably as seen in other industries, however there is no overall guiding vision other than what people agree on collectively.
I don't see why people have trouble understanding this. Everything else works in this way, for instance the car market. You have choice, standards, competition. Why do people bitch when governments don't protect them from monopolies in real life, but run straight into the arms of proprietary systems in computing?
And it's full of that unix-y goodness which let me kill a frozen AOL client on her machine remotely.
Hmm, but from what I remember, you can't start Mac apps from the command line (because they are directories). So you can kill AOL from SSH, but not start it. Great. Please, stop with the Mac shills.
Finally, I can't believe people still stay things like "Linux lost, Windows won". Or something. I must have missed the point at which Linux started competing on level ground with Windows. Because right now, as far as I'm concerned, it's not there yet in terms of ease of use (software management etc, but it's being worked on). Linux hasn't lost - au contraire, it hasn't started the race yet. But it will soon - once awareness of Linux is high, and normal (read non geeks) are trying it on the desktop THAT is when the race will have begun. Not before.
ReiserFS v4 makes some BIG changes to the kernel - for instance, it requires that files can be also directories at the same time. It's currently slated for December 31st, so I doubt it'll be in the next kernel. It's a big bit of work. Anyway, I think most distros have had ReiserFS v3 (which is also journalled) in their for ages, I know SuSE has, or you can recompile your own. Having it in the base kernel is nice, but hardly necessary.
Not especially any particular rule, it's just a viewpoint held by me (and many others). Look at the fuss caused by the W3C looking at RAND for one example.
Open standards are meant to be agreements between people on a common way of doing things, that is available to all. That is their purpose and nature. If all those people now have to pay in order to use that agreement, then by definition it is not open to anybody, as some people may not be able to afford the licensing fees. Is there a license for the internet? No, and that's why it could take off so fast.
What is there about Ogg that I don't know yet that would make me say, ``Yeah, that's way better than MP3?'' Is it technically better, somehow? Can I squeeze that 15 GB music collection into 1 GB with no noticable loss of sound quality, or something?
You'll get lots of technical information about why Ogg is better. Here's one non-technical reason:
Unless you paid up for your encoder (and maybe you did, but many do not) you broke the law.
In theory, whenever you encode an MP3, you're supposed to pay for it. Ditto for streaming radio. If you use a free encoder, you violated Thomson Multimedias patents, and you're in violation of the law.
If you did pay for your encoder (and for all I know Apple did a bulk deal so if you use iThing it'll be ok), this still sucks because you just paid to create something in what is supposed to be an open standard. Open standards are not supposed to have intellectual property licenses. Period.
Re:Propriety formats are Apple's enemy. Or should
on
Sorenson Countersues Apple
·
· Score: 3, Insightful
And yet, Apple continues to develop and promote decidedly unopen formats like Quicktime, which are definately not friendly to alternative platforms.
And unfortunately those platforms are not very friendly back. I'm beginning to think QuickTime is Apples own worse enemy here - given how long they (or Sorensen) have been jerking open source video makers around, I wasn't surprised to read that an old version of Sorenson had been cracked. How long can it be until the latest versions are too? And what will that mean for the lawsuits?
I, by the way, don't know what to think. Surely Sorenson and Apple have rights to their own creations, but on the other Natalie Portman was available only in their proprietary format. That is like a big neon sign to the geek community saying "Come on, crack me!".
If I was going to create a closed codec, I'd make damn sure there were players for pretty much every platform out there. I'd make high quality players for Windows, Linux and maybe the Mac, and then a library for everybody else so people can write their own players if they need to. Otherwise, the moment good content gets encoded using it, by by secrects.
Linux not just for geeks anymore
on
Forbes on Linux
·
· Score: 2
For example, Linux will never rival Windows as a mainstream desktop PC operating system, and Linux still has a long way to go before being widely used in "back-end" applications like databases and enterprise resource planning.
But that's my whole point! Packages should not need to be given loving care and attention to make sure they are integrated. They should just work too. About compiling from the source, many people (ie end users) aren't willing to compile from the source, and would prefer binaries whenever possible. But at the moment, there is no good binary distribution method - which is why distributions have to hand-integrate all their software.
You will loose Classic support though, and everything will have to be recompiled, but MacOS X provides a mechanism for distributing multiarchitecture binaries.
That's something that I haven't seen raised yet. People say "porting OS X to Intel" in a rather loose fashion, apparently without realising that if Apple released OSX/i686 tomorrow, it would be useless. If I went out and spent a lot of money on it (and it would be a lot of money) I would get the standard Apple apps and.... that's it.
I don't think vendors would recompile their apps. Why do I think this? Past experience. Look at Linux, it runs on almost any architecture under the sun, including PPC. The only reason this is viable at all is because virtually all the software has the source available - if you look at commercial vendors like Real, Yahoo! and so on, they invariably only support Linux/i386. Extracting builds for other architectures is like getting blood from a stone, they just don't want to know .
I'm not sure why this is the case. After all, producing a PPC build is only a matter of going out and buying a Mac, installing Linux on it and recompiling. Nothing to it. Perhaps it's because they think, well if we support i386 and PPC, suddenly users of all sorts of other wierd architectures will start demanding builds, and we can't do all of them. I don't know. But I do know that it would take a long time for the number of MacOS/intel programs to reach the number of MacOS/PPC programs. It'd have the chicken and egg situation that killed BeOS, prevented so many other platforms from even taking off, and Linux had to work around by making all the software open source. Without customers, they wouldn't support Intel builds.
Oh and also, when you mention that there is a mechanism for distributing multiarchitecture binaries, I assume you mean you can put different builds into the same appfolder. That strikes me as an extremely dumb idea. Why exactly would a Mac user want their harddisk cluttered up with unusable Intel binaries and vice-versa? Better to do what is already done, and provide separate distributions for each architecture and you choose which one you want at the website/shop.
This isn't what Ian was talking about by the way, but really the big problem with apt is that it doesn't scale.
Having a huge centralised repository worked OK in the early days, when there wasn't much Linux software and most of it didn't have a homepage anyway. These days, the situation is different. There are MANY distros, and Debian is far from being the most popular. Having huge collections of distro-specific packages (which is what apt is) requires huge amounts of manpower - I think Debian has over 1000 people involved.
What's needed is something like apt, but decentralised so when a software developer releases some software, they don't have to wait for an approved packager to build it for them and add it to apt. It should also be able to work on any Linux distro, so there's none of this Suse-RPM, RedHat-RPM, Mandrake-RPM, Debian, Slackware packaging rubbish.
I'm not sure about the FireWire issue, but I know for a fact that Apple have some kind of patent on font hinting, which means FreeType can't ship the sources with it enabled. Apple have never sued FreeType, however if you want decent anti-aliased fonts in theory you must have a license from Apple.
In order to get around this patent, the FreeType people implemented an auto-hinter that attempts to automatically guess what the information would have been. The fact that this is necessary at all is seriously lame in my books, I'm not sure what Apple would say if FreeType started using the real hinter again, but at the moment the FT people seem to think Apple are holding them up.
Slashdot Mirror
Really? Last time I checked, they'd sold a whopping 400 of them.
uh huh. And how many people do you know that run pure Darwin, without the rest of OS X, on Intel hardware? Not many. Darwin is nice, but it's nothing amazing. We already have too many free unix kernels.
And the important point is that by charghing for their UI and their software, and by making hardware, Apple is able to fund desktop unix in a way that nobody has been able to for Linux. Rad Hat has done an admirable job-- and most of their work and apple work compliments each other.
That is bollocks. You are making a fundamental mistake here, which is assuming that the desktop is like Darwin. Red Hat and Apple are not at all compliments of each other, as the software RedHat fund is given away under the GPL - not so for Apple. They are competitors. GNOME and MacOS are not at all complimentary, as GNOME exists to be a good desktop (let's not argue about whether it is or not), and the MacOS desktop exists to make Apple money.
It is disingenuous to compare this to a company that charges people for its software whether they want it, use it or buy it, or not. Apple only charges people who freely choose to buy its software. Calling it an "apple tax" only undermines your creditability.
In theory, only people who want to pay for Windows. Doesn't really work that way does it? If Apple were at 90% market share tomorrow, you can bet it'd be an Apple tax. Just because they are small and largely insignificant NOW doesn't mean they should be supported - at the end of the day, they stand for the same thing as Microsoft: profit through closed platforms. Therefore, if you really believe in choice and freedom, you MUST support Apple and Darwin. Doesn't mean you have to buy their software or hardware-- but opposing it is opposing freedom.
Bullshit. Most people couldn't care less about Darwin, the UNIX kernel has been done to death already. The thing that interests people is MacOS, you know, that thing that is proprietary and runs only on Mac hardware. What are you smoking? If you really believe in choice, then why the hell are you supporting a company that makes its money from not giving its users choices in the same of simplicity?
There's a simple solution to this dilemma, which is, don't make your products open source if you want to make money out of it. Free software is great for writing operating systems, but only Stallman has ever claimed it is the be all and end all of software development. Note that you can write open software without giving away the source, simply by documenting the file formats and protocols. I don't respect companies that don't do this anyway, as it implies that they feel they need artificial lockin to stay afloat rather than just producing quality software.
Having said all this, I have a problem with Tom Lord asking for donations, and ditto for Rob Levin with openprojects.net. There are countless open source projects in the world, many of which are very important. The Linux kernel, KDE and so on are all huge projects, yet I don't see them begging for cash. I also write open source software, but I do it in my spare time, and delegate work that I cannot handle, because my projects are by necessity non commercial. No project should be so dependant on one person that they have to work on it full time. This goes for writing source control systems, or running IRC networks. I think projects should either be non commercial, in which case you have a paying job during the day and work on it in your spare time, or you figure out a way of making money from it (ie by keeping the source closed).
I don't see any good way, or any good reason, for attempting to make money directly from donations for open source projects. BitMover has got the right idea, they are getting mindshare and free testing by giving away their product to free software developers, but charging for it for commercial operators. They've figured out a way to tread the line, but most don't.
Unfortunately, for all its good points, the Mac is a proprietary platform. I see a lot of confusion about this in these forums. People say things like, it's based on Darwin which is open source, so OSX must be an open platform. Wrong. There is a simple test to see if a platform is open or not, namely, can make my own version and then sell it. With a Windows PC, you can get partially there, as you can build your own PC and sell it, but you can't do that for the OS. A Windows PC is semi-open.
Could I make my own Mac? No, I could not, as the hardware is closed. Nobody but Apple may make Macs, or even computers that look like Macs, as the "brand" is one of their most valuable assets. OK, so let's say I tried to clone OS X. Could I? No, I could not. I would get as far as downloading and customising Darwin, and then .... oh, whoops. It seems the rest of OS X is not open. I couldn't even recreate the artwork without getting sued. Yes, I know you can compile Linux apps on MacOS, but you can do that on Windows too. I don't see people telling me to take Windows users out for lunch.
So the Mac is proprietary. It dismays me to see people supporting such a platform. We've all, every one of us, learnt the hard way what happens when a company gets control of an important platform. Are people really willing to throw away not only openness of the OS, but also the hardware, merely to get a colour co-ordinated desktop? And for those who believe the 3 platforms could exist in peaceful cooperation, I have bad news: proprietary platforms have a tendancy to get a tight grip, as people get locked into them. Mac apps cannot be easily ported to Linux for instance - the flow is one way only.
In short: has nobody learnt anything from Microsoft? Do we really have to learn the same painful lessons twice, once from Gates, and once from Jobs?
Indeed, I can imagine that. These were my ideas
I got quite a long way with it - on that page are ideas for how such a musical marketplace could work, how quality would be ensured (ie how do you sort the wheat from the chaff without record companies A&R depts) and answers to questions musicians would commonly have.
I can so see this happening. Music is, in a way, a parallel of the software world. Music is effectively information and can be replicated and copied for zero cost. It's dominated by a rich and powerful entity (the RIAA vs MS), and the world is crying out for something better. It's ripe for the gift economy to be established here.
The best right now is probably Rival, which is coming along nicely, but unfortunately isn't finished yet. It's also not open source, dunno why not as it's freeware. It's written in Visual Basic too unfortunately, occasionally when I've been talking with Dan who makes it about a feature, he's told me simply "VB can't do that" - and that's the end of it :(
If somebody was to make a really solid, easy to use client that can compete with MSN, Trillian and so forth, Jabber could take off. All of my friends are on MSN but there's no way I can "convert" them to Jabber unless the clients are solid.
After that - well, the network has been stable for a while now (as long as you don't want AIM, but that network doesn't have much of a presence in the UK). I run the theoretic.com server with Theo, a friend of mine, and it's got features like IM headlines (from slashdot :p) and soon an integrated weblog. The features are there, too bad the clients are not.
Indeed, and that protocol is called .NET Remoting, and is a set of extensions to SOAP which allow for true object serialization a la CORBA. SOAP is useful now, but one thing I never understood until a few months ago is why they left out what's known as object-by-reference, which lets you pass objects around (pretty useful). Then I remembered that SOAP was partially designed by MS. I can just see the meetings:
Microsoft: "No no, we don't want objects-by-reference, we want to keep the spec clean and lean"
only to then turn around and add it as their own extension to SOAP. Now SOAP is just XML so it'd be pretty easy to reverse engineer said extensions, but who says it's not patented?
Well I didn't cite because I saw these figures buried in one of the KDE mailing lists and didn't have the link handy. Also, you're right about runtime speed but I was talking about link speed (to do with startup time) which it doesn't appear to improve at all. Most people don't have a beef with the runtime speed of Linux apps (which is generally good), they don't like the amount of time C++ apps take to start.
thanks -mike
In theory yes, in practice unfortunately it seems that it doesn't. I saw some statistics for 3.1, it takes TWICE as long to compile, and the resulting binaries are slightly slower at linking. The reason KDE seems slow is due to start up time, caused by inefficient linking, so 3.1 doesn't seem to improve KDE at all. Apparently it now requires an improved glibc, but I may be wrong here.
Anybody care to correct me on this?
You say, what if a patch goes against my "mission". I'm not sure what you mean here, other than perhaps you have an instinctive feel that something is wrong. I say, don't worry about it. For starters, most projects debate patches before work starts on them, so you can usually veto an idea before it gets turned into code. Use this opportunity to steer things towards your vision.
One last tip: don't release too early. If what you're writing is complex, make sure the fundamental architecture is there, and make sure the rest is basically filling it out and hacking on it. If you just announce a vague idea, people are going to get involved who have very different ideas to you, and you'll either get pushed around or engaged in a flamewar. If you ensure you have the fundamental design there first, then people who don't like the design will push off, and those that do will (hopefully) stay to help.
Well once again, putting all config files into ~/config (or rather ~/.etc) would be nice... but converting all the apps would be a lot of effort and all that would have changed is some file system locations.
Now that's a gem. You never use the filesystem layout on your systems? Would YOU want to use the system never seeing the filesystem? Or is this exlusively for the stupid newbies? Come on... in your vision of a newbie system the paperclips aren't that far off - and just admit that you would never run such a system without an "advanced" switch that let you disable all your nice UI gadgets.
Huh? I'm lost, where did paperclips come into it? I use the filing system a little yes, because I'm currently developing an installation system. Other than that, I rarely access the / filing system directly unless I need to edit an obscure config file for which there is no GUI. I never access the rest of the filing system from a graphical browser, because they are designed to hide all that from me, whereas the command line is not.
And I don't think of myself as a leet superwhizzkid. I think of myself as a Linux user.
it's hardly ever possible to guess all of the places unix installs go - that without even providing that service of adding things to your menus.
Hmm, a lot of these gripes seem to be do to with the package management system - I share these gripes by the way, which is why I'm writing a software manager that can deal with adding menu options etc. Even with RPM though, you never need to guess where things go, you can query the package. Or use something like Red Hat, which dumps all the software into /usr
instead of /etc, /usr/shared/something/etc|config|picksomething, /usr/local/something, /opt/*/config|etc, ~/.mycrypticnamerc
That's a big exagguration. Config files are placed in /etc or your home directory. I believe KDE/GNOME apps have their own system which doesn't use etc, but I can't speak for them. Yes, occasionally a badly behaved program "invents" its own etc directory, but badly behaved programs aren't going to disappear anytime soon.
I can't really think how you'd change the filing system layout to make it massively better without, well, changing Linux as a whole. /var isn't particularly descriptive, but it's used for a whole range of stuff. What would a better name be? /etc is an odd name too, but once you know etc is where config files are stored, you know it. Renaming it to /config would be a lot of work, for purely aesthetic gain.
Linux would do much better from a redesign of some of its key parts all at once, for instance creating a decent object model, and fixing the mess that is autoconf/automake. Which will happen some day, I have no doubt, but for now it's not a major barrier to adoption.
The incipid design? That's a new one. What exactly is wrong with X? Actually, it's pretty similar to most other platforms graphics engines, except most other platforms don't have network transparency. Or what, do you think GDI+ is a model of good design?
My dream? All those hard working XFree86 developers design a new system from scratch. They're very talented and hard working. They could easily produce something MUCH better than X if they even gave it half a try.
A few people thought that same thing, and started Berlin. And it's pretty cool. But nobody uses it, and almost nobody develops for it. Why? Because X works, and it's here, today. Berlin, for all its futuristic architecture, has its own problems. Nothing is perfect - I don't get why people are so hung up about the graphics API. X is not the problem of Desktop Linux.
No no no no NO! First off, Windows is not more organised than Linux. Far from it. Apps frequently install themselves god knows where, usually in Program Files, but sometimes also in directories off the root, sometimes people install apps into "My Documents" (I have seen this!) often if you have multiple partitions there is more than one Program Files folder etc. It's a mess.
And what's this about apps being spread out everywhere? Windows apps usually install stuff into their directory, the windows or windows/system directories, the registry, sometimes they add stuff to your Profile and they add shortcuts. I don't hear anybody complaining about that.
Joe is going to have a lot of questions rather quickly. For instance, why isn't there user stuff in /usr? Who is /usr/share shared with? What's optional with /opt and why isn't the rest optional? And why is my home directory full of config files if config files go into /etc? And why are there at least two */bin dirs (containing not only binaries but other runnable files btw)?
Joe user will never need to know the answers, or even ask the question in an easy to use distro, for the same reason that nobody asks "Why does configuration data go into HKEY_LOCAL_MACHINE but also HKEY_CURRENT_USER?" - the registry is never accessed by users, and ditto for the Linux file system.
Say what you want about the Windows registry, but at least it's not laying around in plain view in Joe's home directory.
Why should the root file system be? At the moment, when I start Konqueror I don't even see the root system unless I ask for it. It's not very polished, but that can be fixed. MacOS has all these directories too, but the user never sees them. There's no need to rename them, it's just a case of good UI.
If you're truly going for an easy-to-use idiot-friendly linux, you're going to have to take some tough decitions. Toss the old layout out the window, pick something like /apps, /config, /system, /documentation, or whatever - and spend a long time compiling stuff from scratch to make it work.
That would be nice, but it would bring only one clear advantage - nicer names. I can't think of any compelling reason to do this at all, considering that these are paths normally only used by software. With proper software management, the user need never touch these directories, so why bother changing them?
Norton Utilities is mainly made up of programs that either a) undelete things, b) fix problems with windows or c) optimise things.
All of these are features of the OS, not a utility pack! Modern Linux filing systems optimise themselves, and it doesn't state on their website what sort of probems Norton solves exactly with windows. Undelete? Isn't that what the recycle bin is for?
We have Kylix from Borland (FREE !) and how many of us are using Kylix to develop USEFUL UTILITIES for the users ?
None of us, because Kylix sucks. This is such a shame, because Delphi is excellent, a truly wonderful program that I've used to develop utilities before. Kylix is primarily meant for GUI apps though, and I think you'll find most Linux utilities are written as command line apps.
LOL! Like you'd get modded down for praising the Mac around here.
First, XFree was a pain in the ass to get set up. I haven't tried it since 4.x, but 3.x sucked because all the setup programs wanted to compute "optimum" modelines for your monitor and display card, which inevitably never worked for me. This instead of what I wanted: resolution and refresh, from the list of VESA standard modes.
What, you mean like this?
Then there's getting the desktop environments running themselves. I didn't get very far on them, but in my experience, if you didn't pick the window manager favored by the distro, the others simply weren't configured to do anything useful. The only way to get menus to contain anything useful seemed to be by editing config files, and by this time I wasn't in any mood to search for more damn config files to edit.
a) You're wrong. I don't know which distros you tried, but I've never had to edit config files to edit menus (this is SuSE 7.3)
b) You're right. Desktops don't share applinks currently, so if you install a KDE app, it may not play nice with the GNOME menu and vice-versa. There is currently a standardisation effort underway to fix this (and it'll provide a far more advanced menu system than any other OS to boot). Check out freedesktop.org for more info.
It just works, without a bunch of tweaking, partly because Macs have nowhere near the hardware nightmare that exists in the x86 world.
Yes yes, we know. "It just works" - what a good slogan. People seem to miss the point about Linux: IT IS FUNDAMENTALLY DIFFERENT. It relies on people creating things themselves, them on users liking it and adopting it, then on people forming standards etc. There is no "Steve says we do this, so we do this" policy. That has the advantage of the fact that it scales reliably as seen in other industries, however there is no overall guiding vision other than what people agree on collectively.
I don't see why people have trouble understanding this. Everything else works in this way, for instance the car market. You have choice, standards, competition. Why do people bitch when governments don't protect them from monopolies in real life, but run straight into the arms of proprietary systems in computing?
And it's full of that unix-y goodness which let me kill a frozen AOL client on her machine remotely.
Hmm, but from what I remember, you can't start Mac apps from the command line (because they are directories). So you can kill AOL from SSH, but not start it. Great. Please, stop with the Mac shills.
Finally, I can't believe people still stay things like "Linux lost, Windows won". Or something. I must have missed the point at which Linux started competing on level ground with Windows. Because right now, as far as I'm concerned, it's not there yet in terms of ease of use (software management etc, but it's being worked on). Linux hasn't lost - au contraire, it hasn't started the race yet. But it will soon - once awareness of Linux is high, and normal (read non geeks) are trying it on the desktop THAT is when the race will have begun. Not before.
ReiserFS v4 makes some BIG changes to the kernel - for instance, it requires that files can be also directories at the same time. It's currently slated for December 31st, so I doubt it'll be in the next kernel. It's a big bit of work. Anyway, I think most distros have had ReiserFS v3 (which is also journalled) in their for ages, I know SuSE has, or you can recompile your own. Having it in the base kernel is nice, but hardly necessary.
Open standards are meant to be agreements between people on a common way of doing things, that is available to all. That is their purpose and nature. If all those people now have to pay in order to use that agreement, then by definition it is not open to anybody, as some people may not be able to afford the licensing fees. Is there a license for the internet? No, and that's why it could take off so fast.
You'll get lots of technical information about why Ogg is better. Here's one non-technical reason:
Unless you paid up for your encoder (and maybe you did, but many do not) you broke the law.
In theory, whenever you encode an MP3, you're supposed to pay for it. Ditto for streaming radio. If you use a free encoder, you violated Thomson Multimedias patents, and you're in violation of the law.
If you did pay for your encoder (and for all I know Apple did a bulk deal so if you use iThing it'll be ok), this still sucks because you just paid to create something in what is supposed to be an open standard. Open standards are not supposed to have intellectual property licenses. Period.
And unfortunately those platforms are not very friendly back. I'm beginning to think QuickTime is Apples own worse enemy here - given how long they (or Sorensen) have been jerking open source video makers around, I wasn't surprised to read that an old version of Sorenson had been cracked. How long can it be until the latest versions are too? And what will that mean for the lawsuits?
I, by the way, don't know what to think. Surely Sorenson and Apple have rights to their own creations, but on the other Natalie Portman was available only in their proprietary format. That is like a big neon sign to the geek community saying "Come on, crack me!".
If I was going to create a closed codec, I'd make damn sure there were players for pretty much every platform out there. I'd make high quality players for Windows, Linux and maybe the Mac, and then a library for everybody else so people can write their own players if they need to. Otherwise, the moment good content gets encoded using it, by by secrects.
That's what she thinks.
cue evil laugh.. bwahahahahahaah
But that's my whole point! Packages should not need to be given loving care and attention to make sure they are integrated. They should just work too. About compiling from the source, many people (ie end users) aren't willing to compile from the source, and would prefer binaries whenever possible. But at the moment, there is no good binary distribution method - which is why distributions have to hand-integrate all their software.
That's something that I haven't seen raised yet. People say "porting OS X to Intel" in a rather loose fashion, apparently without realising that if Apple released OSX/i686 tomorrow, it would be useless. If I went out and spent a lot of money on it (and it would be a lot of money) I would get the standard Apple apps and .... that's it.
I don't think vendors would recompile their apps. Why do I think this? Past experience. Look at Linux, it runs on almost any architecture under the sun, including PPC. The only reason this is viable at all is because virtually all the software has the source available - if you look at commercial vendors like Real, Yahoo! and so on, they invariably only support Linux/i386. Extracting builds for other architectures is like getting blood from a stone, they just don't want to know .
I'm not sure why this is the case. After all, producing a PPC build is only a matter of going out and buying a Mac, installing Linux on it and recompiling. Nothing to it. Perhaps it's because they think, well if we support i386 and PPC, suddenly users of all sorts of other wierd architectures will start demanding builds, and we can't do all of them. I don't know. But I do know that it would take a long time for the number of MacOS/intel programs to reach the number of MacOS/PPC programs. It'd have the chicken and egg situation that killed BeOS, prevented so many other platforms from even taking off, and Linux had to work around by making all the software open source. Without customers, they wouldn't support Intel builds.
Oh and also, when you mention that there is a mechanism for distributing multiarchitecture binaries, I assume you mean you can put different builds into the same appfolder. That strikes me as an extremely dumb idea. Why exactly would a Mac user want their harddisk cluttered up with unusable Intel binaries and vice-versa? Better to do what is already done, and provide separate distributions for each architecture and you choose which one you want at the website/shop.
Having a huge centralised repository worked OK in the early days, when there wasn't much Linux software and most of it didn't have a homepage anyway. These days, the situation is different. There are MANY distros, and Debian is far from being the most popular. Having huge collections of distro-specific packages (which is what apt is) requires huge amounts of manpower - I think Debian has over 1000 people involved.
What's needed is something like apt, but decentralised so when a software developer releases some software, they don't have to wait for an approved packager to build it for them and add it to apt. It should also be able to work on any Linux distro, so there's none of this Suse-RPM, RedHat-RPM, Mandrake-RPM, Debian, Slackware packaging rubbish.
In order to get around this patent, the FreeType people implemented an auto-hinter that attempts to automatically guess what the information would have been. The fact that this is necessary at all is seriously lame in my books, I'm not sure what Apple would say if FreeType started using the real hinter again, but at the moment the FT people seem to think Apple are holding them up.
Software patents suck. Period.