Oh right, so Microsoft should ship out Vista without a browser at all? They should sell that as an additional package?
Yes they should. Dell or Gateway or Sony or Lenovo are free to bundle a browser with Windows and a computer and a bunch of bananas if they want and sell them to you, but the decision legally must be theirs without any coercion or incentive to choose on anything other than quality and price. This benefits consumers and the market by letting each vendor choose and the market decide through competition (the reason capitalism works).
Or does that mean Ubuntu is 'illegally' bundling Firefox with its OS? Is it ok to bundle a browser as long as it is open source?
It is legal to bundle anything with anything so long as one of the items does not wield monopoly power in a different market. Ubuntu wields monopoly power in no market and neither does Firefox. Gateway wields no monopoly power and can bundle IE and Windows legally. Apple wields no monopoly power (probably) and can bundle Safari and OS X. For that matter, because they are open source it is unclear if Ubuntu or Firefox would ever be able to wield monopoly power unless a barrier to forking can be somehow created without violating the license.
If the US courts did not enforce these laws in the past, Microsoft would not exist because IBM would have crushed them before they even got started.
Because no company has ever drummed up support for its product. It is easier to say a feature did not make it then to try and proclaim at the last minute that x, y, and z cool widgets made it into your program.
Vista is not "the most secure OS ever" and IE 7 is not "standards compliant." You can call it marketing if you want, but it is deceptive and I'm certainly not going to trust the word of a company that makes deceptive statements like this. Anyone who does is a fool.
Say[sic] thing as IE.
Yeah, a concept that they apparently don't teach in Econ 101 anymore. How is it that som many people on Slashdot can be so oblivious the concepts of monopolies and tying? How is it they then insist on making assertions without taking five minutes and educating themselves anyway?
OMG! I installed Ubuntu last weekend and it came illegally bundled with FireFox and XMMS! How do I sue?!?!?
Please educate yourself. Bundling only breaks the capitalist system of economy when it is bundling something with a product that wields monopoly power in a different market. None of your examples do that and none of them are against the law. All of the examples I gave do that and are illegal.
For the longest time one of the big complaints against Microsoft has been their closed nature and their lack of interaction with developers outside of their own organization. Now they are opening up, or at least they are presenting the appearence of opening up.
Not really. They aren't moving to open standards, protocols, and formats. They aren't publishing their formats or protocols completely. They aren't opening up, they are just trying to make the most money while giving up the least amount of lock-in.
In my opinion, I think that Microsoft seriously does see the hand-writing on the wall and they do want to do more to ensure that their OS supports the programs that people want to use.
This has always been MS's policy. Make sure the popular stuff works, then gradually bundle a competitor or otherwise illegally motivate people to move to something that makes MS money.
On another level, Microsoft is trying to avoid what happened to Novell in the 1990s. Netware was a great operating system but it got to the point where they barely had any third party support. The same thing could happen to Microsoft if enough developers decide that using Microsoft dev tools is a PITA and if enough developers decide that coding to the Microsoft OS is a PITA.
There are only two ways for this to happen. One is for MS to lose a huge portion of the existing market. The other is for a cross-platform intermediate layer to exist. The former isn't going to happen without a huge revolution. MS has been mostly successful in killing attempts at the latter, like by intentionally breaking and not supporting Web standards by default and keeping the Web from being a viable intermediate layer until they can control it with something proprietary, like Active X or.net.
The one ray of hope is "standards" but as we've all seen, Microsoft will just ignore a standard until enough people want to use it.
Most users and purchasers don't want standards, they just want the benefits standards bring. MS does a great job of marketing things they claim will bring those same benefits but don't, or using a bait and switch to provide something they claim is a standard, but which turns out not to be (OpenXML).
You're seeing it now with IE7... Now enough web devs have complained loudly enough and they're finally getting what they want.
No, we're seeing them make lots of noise from their marketing department about standards in IE and then excuses like backwards compatibility when they are called on it by people with a clue. IE7 fails to implement huge portions of CSS and does not support XHTML despite the fact that numerous other companies and hobbyists had no trouble doing so with much fewer resources and time. And before you bring up the backwards compatibility excuse, you'd better have a good explanation why all the missing (not broken) CSS features and inclusion of XHTML would break backwards compatibility in any meaningful way.
The problem with using a standard to fight Microsoft is that standards are very rarely proprietary.
No, the problem is while MS fails to implement standards demanded by customers and courts and spends millions on marketing to obfuscate this fact, they are also breaking dozens of other standards that people are less cognizant of. MP3, MPEG, PDF, PNG, JPEG, etc. are all being quietly pushed out by bundled, proprietary alternatives while the pressure is on Web and Office formats.
The bottom line is MS does not play nice and deceives people about what they are doing. MS illegally bundles everything and if you use the platform without the utmost of care (think average user or even average corporate admin), you're screwing yourself over for the future.
Not to get too serious here, but this is a perfect example of a situation where MS can't win. Invite the folks up? "It's a trap! They'll steal your code, kill you, etc." Don't invite them up? "When is MS going to treat OSS developers like any one else, Firefox has many users, they should get the same respect as any other org."
Not to make a bad analogy here, but let me present a perfect example of a situation where J. Dahmer can't win. Find the body of a missing person in his apartment? "He's raping the dead." Don't find a missing body in his apartment? "He probably ate it and dissolved the bones in those acid vats."
The solution to this dilemma is don't spend more than a decade gaining notoriety by constantly screwing people over, breaking the law, and behaving unethically. People suspect the worst of MS, because MS delivers on a regular basis. That's not prejudice, it's experience.
Microsoft has taken some serious steps to clean itself up over the last year or so.
Really? You mean they are no longer illegally bundling IE? Oh yeah they still are. You mean they are no longer paying companies to spread FUD about security and performance? Oh, they're still doing that too. So they are not misleading people by overstating the security of Vista and the compliance of IE? Oh, they did that too. Well surely they aren't still illegally bundling their media player? Huh, they're doing that too. Have they stopped illegally tying their server and desktop to take over more of the server space with an inferior product? No, they are still doing that as well. How exactly have they "cleaned up?"
That said, it's really sad to see that 98% of the comments here are based on distrust, hatred, and bad jokes.
Trust is earned. After the fiftieth or sixtieth time someone punches me the kidneys when I'm looking the other way, it is not sad that I talk about how I suspect they might be trying again. If MS wants my trust they have to earn it and it will take years of ethical, trustworthy behavior before I'm willing to admit that this time they might not be maneuvering for another cheap, sucker punch. Not punching me when I'm looking right at them and a cop is paying attention does not earn them any trust.
This is one of many times you'll see Microsoft bent by the immense power and will of open source!
This is MS looking out for their bottom dollar, probably by trying to get new "Vista only" feature into Firefox to help motivate corporate upgrades either by selling that feature or by hoping it will hasten the demise of mainstream support for Firefox on old versions of Windows. Does this make me trust MS to any greater degree? Hell no, and nor should it.
There's a simple explanation: it's probably NOT working well, and they want to have a heads-up on what kinda complaint level they'll have. OR, they want to make sure to "break" certain firefox features so that IE looks better.
This is possible, but I don't think it is likely. I suspect the issue is slightly different. Vista's biggest competitor is going to be earlier versions of Windows. Many corporate customers are still using Win2K and many are also using Firefox. Why would they upgrade?
The Firefox crew is pretty sharp but they are techno-junkies. So MS invites the Firefox guys to see some of the whizbang new features of Vista that they can integrate with Firefox to make it better. Maybe they can even get these guys excited about the potential of something. The hope is that the Firefox people will add some feature that will motivate people to want to upgrade to Vista. Even if they just get a feature built into the core tree, maybe the older versions will become unsupported more quickly and for security reasons people will need to move to Vista to have a secure browser.
Remember, MS does not sell IE. They sell a bundle of IE and Windows. Every Firefox user on Windows has already paid them for IE, so using Firefox does not really cost them anything other than a minor strategic bump right now. People not upgrading to Vista costs them hard cash, plus a number of strategic bumps when they don't adopt all the new lock-in anti-features in Vista.
I'm not particularly interested in sacrificing functionality so people who are afraid or unwilling to learn command line basics.
I agree with many of the points you made, but the above sentence strikes me as a little wrong headed. Why would you have to sacrifice functionality by providing a non-CLI interface? Ideally the interface is somewhat abstracted and programs should be fully accessible from both the CLI and the GUI. I think it is important to remember that adding a GUI does not remove any functionality from the CLI unless you do it very, very badly.
I haven't followed this story so correct me if I'm wrong but -- the "private data" here were just search terms with no user identification, right?
It was a bit more than just search, it was complete records of internet usage from the ISP.
If that's the case, who the hell cares?
In many cases it is simple to piece together who a user is from these records and some of the data mining potential is more than a little invasive. This is stuff like someone who routinely edits a myspace page with personally identifying information on it and the online stores and Websites they've visited. Gee someone edits John Smith's page and they have been looking at resources for dealing with HIV. Well they won't be winning the local county seat election now will they?
think about it another way. If you are a painter having just completed your masterpiece stretching across a huge canvas, would you be happy if someone just took a detail from it and refused to see the whole work?
No, but I don't believe it is right for the person who bought the work from me and is now reselling it to add a technological mechanism to force everyone viewing it to look at every part of it either, especially when my original vision was for for a 3x3 work, but they insisted I add another 3x3 section as well as "filler" so they could charge more. I especially don't think so if this happens long after I've died.
back to music how happy do you think beethoven would be to know that his epic works have been reduced to a mobile phone ringtone?
Amused, astounded, and gleeful that his work and fame is spreading so long after his death.
does it mean that they don't have the right to ask how they would like their music to be listened to? again hell no!
There is a difference between asking you to listen to the whole album in order, and refusing to sell the good song unless you also purchase the crappy ones at a greatly inflated price. Not that they don't have the right to do so, but the ones doing so out of a sense of artistic integrity instead of greed are few and far between.
For the same reason, I doubt the US would open up their F-16 software. Any bugs (remember that all software contains them) that could be exploited by another government simply by scrutinizing the source code create a tactical disadvantage.
Refusing to release the code used for control systems is one of the reasons why NATO agreements for a common platform have started to exclude the US. The US basically said, "hey it will be easier if we can share munitions and if you guys build your fighters on the same designs we do. Also, you can just buy the parts from American companies and it will make them cheaper for everyone. Then, they refused to share the code they use to run the hardware, making the whole thing unfeasible and making it cheaper for them to design their own systems, which most of Europe can share but we can't.
Which is exactly what 's being done right now.
Actually, countries are sharing, just not with the US or vice versus.
Open sourcing the F-16 software would give no advantage to any government, not even the one buying the F-16. They'll most likely just be more interested in the technical manual of the systems onboard and hand those to an engineer, than they would be in the source code itself.
This is certainly not true. As I understand, it was the deal breaker that prevented a common NATO fighter plane platform from being adopted by the US and Europe.
Funny thing, I bought this house in a nice area, but a short walk from a high crime area. It was built in the 50's out of concrete block (two blocks thick on the ground floor). All the windows on the ground floor were glass block and could not be opened. The upstairs was the real living area with lots of windows, but a full story (slightly higher than a normal house) up in the air. It had a back deck on the second story, with no stairs going down and with the deck overhanging the supports a good three feet or more (making climbing it really hard). It also has a monitored security system (I've since updated). I haven't had a break in yet.:) It is the OpenBSD of houses.
ISP's need to inform people they have bots and if they are infecting other computers they need their internet access dropped. Tough love.
And this gets the ISPs more money in what way? Many ISPs can pull up and print out a list of infected hosts by worm and by the amount of traffic they generate. They can automatically integrate this into their notification system and send e-mail to the host's account or shut down access. They don't because then they have to answer the phone calls explaining what is going on and that costs them money. It is easier to throttle that host's traffic and filter out the worm propagation traffic to their core than it is to deal with users' broken Windows installs. If it hits other hosts in the same router group, why should they care?
This is a good point and a lot of the IRC channels are running on rooted Linux boxes. What I find interesting is how the botherder community knowledge limits what they do. Linux desktops are not protected only by the fact that they are rare, but also by the fact that a lot of these people have no idea what they are doing beyond the tried and true tools. The community has the knowledge to root Linux servers and Windows servers, but aside from that they rely heavily upon metasploit type tools and existing code. If either platform undergoes a security transformation, expect the other to be the exclusive control channel for a while.
Be sure to check whether those are running OS X or Linux.
Perhaps my views are colored by the conferences I attend. Almost everyone I talk to ends up running OS X as their main OS, even while doing Linux development. A lot of people initially set them up to dual boot, but most that I know rarely end up booting Linux much and often end up nuking that partition after a while.
I don't need reliable hardware if I can have someone come replace everything every few months.
Yeah, I've done that with a machine room full of cheap Dell desktops where losing any given one was not a big deal. For my workstation, the time it takes for me to get a replacement and anything lost between backups is worth a lot more than the price difference between a Dell and an Apple (to my company who foots the bill). Not that I'm looking to upgrade to a new machine until Apple has cycled through a revision or two with Intel notebooks.
I doubt that that is going to be solved unless one distribution gains significant marketshare, unfortunately.
Actually, I see two ways for this to happen. One would be for a single distribution to gain a lot of marketshare. The other would be for one or more distributions to adopt a standard that they document and that also provides significant advances (like GNUStep can provide). In that case, I can see other distributions adopting it since it is not that much work (code exists) and it brings real benefits beyond just the standardization. I agree, however, this is pretty iffy.
That is quite interesting. It should be possible with some work to set up a system in Ubuntu that would act like that using deb packages and a modified file manager. It would also probably not be a bad idea to have themes set up that way, as opposed to the current, very inflexible theme manager. I will have to discuss it with some others. Unfortunately, it would only work with Ubuntu...
Yeah, the thing is these benefits can be combined with the above simply by adopting the existing GNUStep standard, with an already well tested code base.
do you mean registration as in network-based licensing, or registration as in registering with a server? A standard for the latter would be nice, but I doubt that developers requiring the former would be willing to use a standard system.
I was speaking of the latter, but I implied the former as well in other comments. The reason I think developers would be willing to adopt such a system is because of security. If you look at FreeBSD jails and the recent trend towards VMs as the direction of the industry you quickly see a few things. Restricting individual applications by resource can eliminate a lot of the security problems we face today, but only if the resources that are accessed can be standardized so users have less exceptions to deal with. This is especially important for end-user desktops. You download two new XML editing suites. When you run the first all runs smoothly, but when you run the second the OS informs you it is trying to connect to the internet, not using the service reserved for registration and updates and may be malware or spyware. On this aspect alone, you may be much more inclined to use the first suite and the latter loses market share.
I firmly believe the industry is moving this way for security and more granular control and I also believe it will bring huge benefits to the first group to bring it to the desktop user in a usable way. This necessitates an official service which will require a standard for widespread acceptance. If the reasons we were discussing are not enough to motivate developers, I think combined with this it will be.
Hmm... that sounds reasonable for nearly all desktop applications. What about an MTA or database, which will use/var?
I guess I should have been a little more clear. Preferences are dependent upon the user running the program, not the location it is installed. If a user is running their own copy o
Yes, there needs to be a common way for packages to install, and it really should include an option to install within your $HOME if your not root (if you really dont want users doing that, you can mount/home noexec). It should be useable from commandline and gui, so you can click on a package within whatever program, and have it launch the installer, which should work out if your root, try to use sudo or let you install to your homedir if your not.
This is solved by GNUStep as implemented by OS X. You install software by putting the package somewhere. If it is in your home directory it is accessible to just that user and if it is somewhere globally available, anyone can run it. You can install with the GUI by dragging it or via the CLI using "mv" or "cp."
It should be integrate with the existing package manager database, so you can update the packages from a single place,
This only works for Apple branded programs on OS X and is just the update portion, not the discovery and download. There should be a standard protocol for this so the tools can access multiple repositories for software downloads and updates, preferably cross-platform.
and it can recognise if necessary dependencies are installed. It should be able to call the standard system package manager to install dependencies, or install lowest-common-denominator prebuilt libraries from the package supplier if necessary.
GNUStep, as implemented by OS X, removes the need for most of this, by bundling the necessary libraries with the package. It takes less disk space than you'd think. Other people here have suggested that this breaks dynamic linking, but actually it merely requires dynamic linking to be modified to take it into account, as OS X does. Basically, users should never have to worry about dependencies.
It should support "online" and "offline" packages, wit the former being little more than a stub listing where to download the actual data from
I'm not sure this is a good idea. What if a novice user grabs an "offline" package from a disk and tries to run it without internet access? I think it needs to be clear that these packages are the programs and always have what they need. What you refer to as an "offline package" should just be a link to the repository and should be distinctly different from real packages.
And finally it should have advanced options available to users who want them, including ability to rebuild the package from source etc.
If you look in a GNUStep package you'll see the binary, plug-ins, resources, and system services each in their own dir. What should be added is a directory for source code and build instructions. Then any application can be built from source either automatically by the OS, by the package manager tool, by the user via a service supplied by the OS (right-click, build custom binary) or manually from the CLI.
I'm glad we can agree for the most part on functionality that would be useful. The other thing I was trying to convey was that almost all of the work necessary to get there is simply updating GNUStep to support a few more options and integrate with existing package managers, defining a standard for those package managers, and getting the major players to support it in their OS's. We have almost all the pieces, we just need to put them together and agree on a standard. Sadly, that is likely the hardest part of this.
Doctor, it hurts when I do this! Don't do it then.
You think just not doing the tasks users want is an adequate solution. Yeah, just keep thinking that. I'm sure users will be switching to Linux in droves any day now to get in on the wonderful lack of functionality.
The whole point of a distribution is to make sure ~10,000 disparate pieces of software work well together. By downloading some random blob that some clown has packaged without really understanding the distribution you're asking for trouble.
No distribution can include every piece of software I might want, including games, niche applications for a particular field, and unstable software that I still need. You're asking me to just stop using more than half of all the software out there, software I need. That is not a workable solution.
Distributions have package repositories for a reason.
And that reason is to provide the base software and the freely available software a user might need. It is not to obviate the need to run any programs not in the repository, because, like it or not, commercial binaries are not going away anytime soon. Ignoring them and hoping they go away is not going to work.
Your model works for very small installations. It doesn't grow well.
Really? Have you seen how many sys admins it takes to administer a thousand OS X boxes? It is a dream come true if you can get the work.
But, understand that Redhat, who is driving Gnome (along with SUN and others) is solving the problem from the upper end. In other words, the solutions that will be generated will be alien to you.
Redhat has a motivated interest in keeping Linux difficult to administer and deploy because it is how they get paid.
I guess the term "average desktop user" is partly to blame here. I don't consider the "home/casual" user to be "average". My "average" is corporate user.
There are more home and small business computers than corporate owned workstations. Further, corporate users have a tendency to fight against using different systems at home and work. Thus, a centrally administered system would require that we move to service based computing for the home or face serious challenges. We'll see which way the market jumps.
None of your arguments address the fact that OS X works today and very well. Please actually try running it for a while as a normal system, and learning how to do things their way (not trying to recreate Linux workflows exactly). The truth of the matter is, Linux can learn a lot from OS X, and benefit from the way they have re-architected things if Linux developers are willing to consider that maybe some things can be done more cleanly. I use both systems and trying to use Linux as a desktop now feels like stepping into a time machine to the past.
You can argue that RedHat will soon be inventing a better solution, and if they do, great. That doesn't help me today though. If they come up with something better, I hope everyone adopts it, but I don't think it is reasonable to assume that RedHat is going to come up with some unknown system that will somehow be better. We have the solutions we need today, if we just piece them together in a way that works. The simple fact is, I can do a lot of tasks on OS X, simply and easily that I can't do at all on Linux. This is a deficiency in Linux and is enough to keep me from using it as my main workstation. A lot of other people have come to the same conclusion.
An application being 'self contained' means really just distributing it with its own set of dynamic libraries which is functionally the same as just using static libraries. Fifteen copies of libxyz on your system. Do you trust that they're all safe versions?
Sigh. If you're going to argue that an architecture is inferior, maybe you should actually read something about said architecture. OS X applications contain a copy of the needed libraries with each package, but does not use that version if it has a more recent minor update from another application, the same as any other use of dynamic binaries.
Your ideal 'solution' is basically ignoring shared objects and thus is like stepping back in time twenty years.
Before making claims such as this, actually investigate what you are talking about in future, please.
There is no dependency resolution, no automatic updates for apps, no upgrade system.
They have removed the need for dependency resolution. They are missing the other two parts. What they do have are applications that are self contained packages and are completely portable. These packages have all the resources easily available, handle fat binaries easily, and don't need to be installed or uninstalled. This is far superior to most Linux distributions, especially for use as a desktop/workstation. They need updates and repositories and services for licensing. Linux needs self-contained, portable packages, one format, and a standard protocol for updates and registration.
On OS X and OpenStep and in my proposed solutions programs a directories ending in ".app" and containing a particular structure. if you double click on the directory it launches the program. If you use the CLI or right-click you can navigate into the application. You can place the application on your desktop, or anywhere else you like. You can also create "shortcuts," "symlinks," or aliases to the application in a number of ways.
How would OS X know if a link is removed? Short of maintaining a database of file reference counts (and if its softlinks, it would be a database of file references, paralleling the directory structure). It doesn't.
Actually, it does. Go ahead an try it. If you move an application or file the links are modified. If you delete an application or file that is referenced, it asks if you want to remove the links as well.
And the desktop icon is simply representative of the application. It isn't the application.
On OS X the desktop is simply a directory in your home dir that is mirrored on the desktop. Applications can be in that directory and work just fine.
Different approach for other systems. As an example, assume that the file system itself is "virtual" in a sense -- things are NOT present unless called for by name (automount - remember I mentioned that?).
How things are arranged and linked is unimportant to most desktop users. They care about the interface you give them and whether it works as they expect. Saying it does not work as they expect does not mean you shouldn't fix other things.
That is,/opt/firefox/bin/firefox would not EXIST in your reference space unless you actually referenced it. Now, how is the icon to exist? There is a disconnect between the "desktop metaphor" and a physical file implementation.
Which is part of what confuses the hell out of most users. You lie to them by showing them an application when in reality you're showing them a link to an application. From the end user perspective, the icon that says "Firefox" instead of "Firefox Shortcut" with a modified icon, is Firefox and deleting it should delete firefox. Copying it onto a remote volume should copy a working version of firefox onto that remote volume. Not doing this is failing to provide a good UI to the average desktop user.
Even at home, I have a terrabyte online. I don't want to drill down hundreds of thousands of files
If you have to, your directory structure is overly complex and should be fixed. What is wrong with/Applications/Firefox.app? It works just fine and does not require anyone to drill down through hundreds of thousands of files. Note, we're talking GUI applications for end users here, not CLI commands, which have no place showing up in the GUI.
Or, lets say your new OS X application can handle files of type "video/mp4v-es" (just choosing a real mime type at random). You want your web browser to create a window, and pass that window handle and the file stream to this application when the data type appears in a web page. Ok, how does THAT work? PS. No pre-launching of the application allowed here (just to make my point as to what an INSTALLER should be doing). What browser is it, anyway? You want to "launch" an existing file (which is why you just copied the application to your computer, after all) and have it come up in "edit" mode.
Sigh, you're making this way harder than it has to be, because you're couching everything in terms of how Linux would do it, not how a cleanly architected system should do it. If I want Firefox to open a certain video type in a certain player, I either implement that player as a plug-in for the browser or the browser looks at the mime type and asks the OS what application I have assigned to it and if it is allowed to auto-launch it or if it is to just save the file and wait for me to do so manually. If it is allowed to auto-launch (
Eh?? Linux is booming in the film, animation and scientific industries.
Linux is doing fine as a server OS. It is doing poorly as a workstation in most fields. It does well in the film industry because of filmgimp, but even there a lot of people have moved to OS X for the workstation. As for scientific "industry" the number of macs as workstations and laptops in most of the sciences has skyrocketed in the last few years, much more so than Linux.
The problem with saving "preferences" in the application directory is that it makes the application not shareable. Big no-no there.
Globally available applications have a global configuration setting configurable by the admin. These settings are inherited by the users, but unless specifically configured otherwise are overridden if a given setting conflicts with a local user's setting. Thus, users get good defaults when they first run an application. Any changes they make effect only their use of the program. The system actually looks at both files when it runs the application, merging the two. Changes a individual user makes are not viewable by other users.
However, a statically linked Linux application, x86, built for a 2.4 (2.x) kernel will run on Linux, OpenBSD and Solaris. No rebuild or relink needed. It's not the issue.
That's pretty narrow and it means one program cannot benefit if another has a more up to date version of a library. A fat package with multiple binaries for different platforms and architectures and source for building a custom binary, all sharing the same resources within the package is ideal, IMHO.
And, to boot, package management is not an issue here EITHER. Package management is only a problem if shared objects are in use. Not an issue with the class of shareable/copyable applications we are talking about here.
Regardless of if a package is sharable I still want to be able to have it operate in a jail and I still want to keep it up to date with a centralized service rather than individually. I still want it to use a standard registration and licensing service so I can regulate, schedule, and monitor it.
What is the issue? A well written x86 Linux application is about as portable as you can get.
There are multiple kinds of portability. One kind is the ability to run the same package on multiple platforms. The other is the ability to move functional copy to other locations and systems and still have it work. I've never successfully IM'd a real Linux app to someone and had it work for them on the other side. I have done so with.app packages from OS X.
As an example, how are your "desktop icons" and/or "program launcher" going to be done? How is this updated if the program is removed? How are links to document types created? How are extension programs for your browser installed? These are the issues that an installer should address.
A "desktop icon" can be the functioning application or it can be a link to it. If you remove a file and it has links, the OS warns you, just like OS X now. Program launching is simply double clicking or single clicking or opening a file with that program as the default application. Everything you mention is a solved issue on OS X already. Get with the times.
The "low-level" issues have been addressed. Shouldn't be a problem -- if they ARE a problem, the software itself can be considered defective.
why continue to paint yourself into the proprietary, lock everyon eelse out world then? people don't like malware, viruses, reinstalling their os every year, etc, either.
This isn't an either/or choice. I never re-install, or get viruses or malware and I install most applications to my desktop by dragging them into my/Applications folder.
...but if you use th edistro's installer, all that is taken care of. iow, debian's synaptic doesn't go out looking for redhat's rpms. the package manager takes care of it. you can always compile from source, if you have to have the latest and greatest prior to a package being developed.
If you use the distro's installer you get a subset of software. What if you need to use any niche software, you know the stuff someone wrote on one distro and never even tried anywhere else. If you're compiling from source, you can run into all sorts of issues with dependencies, the config script, etc. Anyone who uses Linux for real work runs into the software they want to use, but just doesn't work on their distro.
i don't see your issues as being real issue. i've been on linux 1 year and i haven't had that much trouble transitioning.
I run Linux every day and have for many years. It is a great server, but it is weak for a general purpose desktop, especially for non-expert users. People complain about not having a standard package format all the time as well as about the lack of functionality of all these package managers. Blowing this off as "theoretical" is just another way of ignoring the problem. What ever happened to the open source community rallying around open standards? Linux is quite simply lacking a lot of the functionality I want for managing applications. They are not portable in that the same package won't run on all distros and they are not portable in that I can't just copy an installed application from one machine to another. They don't have a standard protocol for repositories, registration, or updating. If all the players would agree to a standard, then we could all pick the best tools and stop duplicating tons of effort and instead concentrate on improvement. The cynical side of me says this has not happened because the major players like using their repositories as a way to make it harder for people to migrate and as a way to collect support fees for something that should not be hard in the first place.
You presume that the Windows Installer actually works.
Get this message through your head. If Linux installation and package management is as good as Windows or even slightly better, Linux still loses. Linux needs to have good package management because Linux does not have a monopoly and is subject to pressure from users.
If followed, the binary will be "portable" across multiple distributions (at least on the same processor and major OS version). The "installer" can the be trivial -- and can simply be a self-run shell script.
Or we could get rid of installers in the first place and have truly portable applications that are packages. All a user has to do is run OS X for a month and run Linux for a month. You'll notice both platforms are deficient in a number of ways. Combine the two and add a little extra. Call it a standard and get everyone onboard and we'll all enjoy security and functionality no one does today.
Basically, "Linux Installers" are trying to solve problems that don't really exist.
I have an audio program installed on my machine. It is freeware, but no longer distributed. My buddy wants to use it, but doesn't know where to get a copy. So, I drag it from my Applications folder into my IM chat with him. It transfers and he runs it from his desktop. He likes it a lot and drags it onto his thumb drive and erases it from the work computer he was using. He takes it home and drags it off of the thumb drive onto his home machine and uses it there. Later, back at work, he needs it again, so he plugs the thumb drive back in to his work computer and runs it off of the thumb drive. It even still has his preferences saved. Did I mention one computer was 32 bit and one was 64 bit? One container ending in ".app" that you double click to run, drag where you want to transfer, never worry about "installing" and drag to the trash to uninstall.
This little anecdote is an example of things I can do on OS X, but can't on Linux
There are also things I can do on Linux, but can't easily on OS X, like schedule auto updating of all applications at 1AM every Wednesday.
I want to be able to do all of these things on Linux, OS X, Solaris, OpenBSD, etc. I want to be able to use the same applications packages between them and I want to be able to use the same or different tools to manage them because they all conform to a standard. Anyone here claiming that we don't need such a standard had better have some bloody good explanations of how I can do all of the above on Linux without reading any manuals, writing any scripts, or going out of my way, because those are things regular desktop users want to do.
Slashdot Mirror
Oh right, so Microsoft should ship out Vista without a browser at all? They should sell that as an additional package?
Yes they should. Dell or Gateway or Sony or Lenovo are free to bundle a browser with Windows and a computer and a bunch of bananas if they want and sell them to you, but the decision legally must be theirs without any coercion or incentive to choose on anything other than quality and price. This benefits consumers and the market by letting each vendor choose and the market decide through competition (the reason capitalism works).
Or does that mean Ubuntu is 'illegally' bundling Firefox with its OS? Is it ok to bundle a browser as long as it is open source?
It is legal to bundle anything with anything so long as one of the items does not wield monopoly power in a different market. Ubuntu wields monopoly power in no market and neither does Firefox. Gateway wields no monopoly power and can bundle IE and Windows legally. Apple wields no monopoly power (probably) and can bundle Safari and OS X. For that matter, because they are open source it is unclear if Ubuntu or Firefox would ever be able to wield monopoly power unless a barrier to forking can be somehow created without violating the license.
If the US courts did not enforce these laws in the past, Microsoft would not exist because IBM would have crushed them before they even got started.
Because no company has ever drummed up support for its product. It is easier to say a feature did not make it then to try and proclaim at the last minute that x, y, and z cool widgets made it into your program.
Vista is not "the most secure OS ever" and IE 7 is not "standards compliant." You can call it marketing if you want, but it is deceptive and I'm certainly not going to trust the word of a company that makes deceptive statements like this. Anyone who does is a fool.
Say[sic] thing as IE.
Yeah, a concept that they apparently don't teach in Econ 101 anymore. How is it that som many people on Slashdot can be so oblivious the concepts of monopolies and tying? How is it they then insist on making assertions without taking five minutes and educating themselves anyway?
OMG! I installed Ubuntu last weekend and it came illegally bundled with FireFox and XMMS! How do I sue?!?!?
Please educate yourself. Bundling only breaks the capitalist system of economy when it is bundling something with a product that wields monopoly power in a different market. None of your examples do that and none of them are against the law. All of the examples I gave do that and are illegal.
For the longest time one of the big complaints against Microsoft has been their closed nature and their lack of interaction with developers outside of their own organization. Now they are opening up, or at least they are presenting the appearence of opening up.
Not really. They aren't moving to open standards, protocols, and formats. They aren't publishing their formats or protocols completely. They aren't opening up, they are just trying to make the most money while giving up the least amount of lock-in.
In my opinion, I think that Microsoft seriously does see the hand-writing on the wall and they do want to do more to ensure that their OS supports the programs that people want to use.
This has always been MS's policy. Make sure the popular stuff works, then gradually bundle a competitor or otherwise illegally motivate people to move to something that makes MS money.
On another level, Microsoft is trying to avoid what happened to Novell in the 1990s. Netware was a great operating system but it got to the point where they barely had any third party support. The same thing could happen to Microsoft if enough developers decide that using Microsoft dev tools is a PITA and if enough developers decide that coding to the Microsoft OS is a PITA.
There are only two ways for this to happen. One is for MS to lose a huge portion of the existing market. The other is for a cross-platform intermediate layer to exist. The former isn't going to happen without a huge revolution. MS has been mostly successful in killing attempts at the latter, like by intentionally breaking and not supporting Web standards by default and keeping the Web from being a viable intermediate layer until they can control it with something proprietary, like Active X or .net.
The one ray of hope is "standards" but as we've all seen, Microsoft will just ignore a standard until enough people want to use it.
Most users and purchasers don't want standards, they just want the benefits standards bring. MS does a great job of marketing things they claim will bring those same benefits but don't, or using a bait and switch to provide something they claim is a standard, but which turns out not to be (OpenXML).
You're seeing it now with IE7... Now enough web devs have complained loudly enough and they're finally getting what they want.
No, we're seeing them make lots of noise from their marketing department about standards in IE and then excuses like backwards compatibility when they are called on it by people with a clue. IE7 fails to implement huge portions of CSS and does not support XHTML despite the fact that numerous other companies and hobbyists had no trouble doing so with much fewer resources and time. And before you bring up the backwards compatibility excuse, you'd better have a good explanation why all the missing (not broken) CSS features and inclusion of XHTML would break backwards compatibility in any meaningful way.
The problem with using a standard to fight Microsoft is that standards are very rarely proprietary.
No, the problem is while MS fails to implement standards demanded by customers and courts and spends millions on marketing to obfuscate this fact, they are also breaking dozens of other standards that people are less cognizant of. MP3, MPEG, PDF, PNG, JPEG, etc. are all being quietly pushed out by bundled, proprietary alternatives while the pressure is on Web and Office formats.
The bottom line is MS does not play nice and deceives people about what they are doing. MS illegally bundles everything and if you use the platform without the utmost of care (think average user or even average corporate admin), you're screwing yourself over for the future.
Not to get too serious here, but this is a perfect example of a situation where MS can't win. Invite the folks up? "It's a trap! They'll steal your code, kill you, etc." Don't invite them up? "When is MS going to treat OSS developers like any one else, Firefox has many users, they should get the same respect as any other org."
Not to make a bad analogy here, but let me present a perfect example of a situation where J. Dahmer can't win. Find the body of a missing person in his apartment? "He's raping the dead." Don't find a missing body in his apartment? "He probably ate it and dissolved the bones in those acid vats."
The solution to this dilemma is don't spend more than a decade gaining notoriety by constantly screwing people over, breaking the law, and behaving unethically. People suspect the worst of MS, because MS delivers on a regular basis. That's not prejudice, it's experience.
Microsoft has taken some serious steps to clean itself up over the last year or so.
Really? You mean they are no longer illegally bundling IE? Oh yeah they still are. You mean they are no longer paying companies to spread FUD about security and performance? Oh, they're still doing that too. So they are not misleading people by overstating the security of Vista and the compliance of IE? Oh, they did that too. Well surely they aren't still illegally bundling their media player? Huh, they're doing that too. Have they stopped illegally tying their server and desktop to take over more of the server space with an inferior product? No, they are still doing that as well. How exactly have they "cleaned up?"
That said, it's really sad to see that 98% of the comments here are based on distrust, hatred, and bad jokes.
Trust is earned. After the fiftieth or sixtieth time someone punches me the kidneys when I'm looking the other way, it is not sad that I talk about how I suspect they might be trying again. If MS wants my trust they have to earn it and it will take years of ethical, trustworthy behavior before I'm willing to admit that this time they might not be maneuvering for another cheap, sucker punch. Not punching me when I'm looking right at them and a cop is paying attention does not earn them any trust.
This is one of many times you'll see Microsoft bent by the immense power and will of open source!
This is MS looking out for their bottom dollar, probably by trying to get new "Vista only" feature into Firefox to help motivate corporate upgrades either by selling that feature or by hoping it will hasten the demise of mainstream support for Firefox on old versions of Windows. Does this make me trust MS to any greater degree? Hell no, and nor should it.
There's a simple explanation: it's probably NOT working well, and they want to have a heads-up on what kinda complaint level they'll have. OR, they want to make sure to "break" certain firefox features so that IE looks better.
This is possible, but I don't think it is likely. I suspect the issue is slightly different. Vista's biggest competitor is going to be earlier versions of Windows. Many corporate customers are still using Win2K and many are also using Firefox. Why would they upgrade?
The Firefox crew is pretty sharp but they are techno-junkies. So MS invites the Firefox guys to see some of the whizbang new features of Vista that they can integrate with Firefox to make it better. Maybe they can even get these guys excited about the potential of something. The hope is that the Firefox people will add some feature that will motivate people to want to upgrade to Vista. Even if they just get a feature built into the core tree, maybe the older versions will become unsupported more quickly and for security reasons people will need to move to Vista to have a secure browser.
Remember, MS does not sell IE. They sell a bundle of IE and Windows. Every Firefox user on Windows has already paid them for IE, so using Firefox does not really cost them anything other than a minor strategic bump right now. People not upgrading to Vista costs them hard cash, plus a number of strategic bumps when they don't adopt all the new lock-in anti-features in Vista.
I'm not particularly interested in sacrificing functionality so people who are afraid or unwilling to learn command line basics.
I agree with many of the points you made, but the above sentence strikes me as a little wrong headed. Why would you have to sacrifice functionality by providing a non-CLI interface? Ideally the interface is somewhat abstracted and programs should be fully accessible from both the CLI and the GUI. I think it is important to remember that adding a GUI does not remove any functionality from the CLI unless you do it very, very badly.
I haven't followed this story so correct me if I'm wrong but -- the "private data" here were just search terms with no user identification, right?
It was a bit more than just search, it was complete records of internet usage from the ISP.
If that's the case, who the hell cares?
In many cases it is simple to piece together who a user is from these records and some of the data mining potential is more than a little invasive. This is stuff like someone who routinely edits a myspace page with personally identifying information on it and the online stores and Websites they've visited. Gee someone edits John Smith's page and they have been looking at resources for dealing with HIV. Well they won't be winning the local county seat election now will they?
think about it another way. If you are a painter having just completed your masterpiece stretching across a huge canvas, would you be happy if someone just took a detail from it and refused to see the whole work?
No, but I don't believe it is right for the person who bought the work from me and is now reselling it to add a technological mechanism to force everyone viewing it to look at every part of it either, especially when my original vision was for for a 3x3 work, but they insisted I add another 3x3 section as well as "filler" so they could charge more. I especially don't think so if this happens long after I've died.
back to music how happy do you think beethoven would be to know that his epic works have been reduced to a mobile phone ringtone?
Amused, astounded, and gleeful that his work and fame is spreading so long after his death.
does it mean that they don't have the right to ask how they would like their music to be listened to? again hell no!
There is a difference between asking you to listen to the whole album in order, and refusing to sell the good song unless you also purchase the crappy ones at a greatly inflated price. Not that they don't have the right to do so, but the ones doing so out of a sense of artistic integrity instead of greed are few and far between.
For the same reason, I doubt the US would open up their F-16 software. Any bugs (remember that all software contains them) that could be exploited by another government simply by scrutinizing the source code create a tactical disadvantage.
Refusing to release the code used for control systems is one of the reasons why NATO agreements for a common platform have started to exclude the US. The US basically said, "hey it will be easier if we can share munitions and if you guys build your fighters on the same designs we do. Also, you can just buy the parts from American companies and it will make them cheaper for everyone. Then, they refused to share the code they use to run the hardware, making the whole thing unfeasible and making it cheaper for them to design their own systems, which most of Europe can share but we can't.
Which is exactly what 's being done right now.
Actually, countries are sharing, just not with the US or vice versus.
Open sourcing the F-16 software would give no advantage to any government, not even the one buying the F-16. They'll most likely just be more interested in the technical manual of the systems onboard and hand those to an engineer, than they would be in the source code itself.
This is certainly not true. As I understand, it was the deal breaker that prevented a common NATO fighter plane platform from being adopted by the US and Europe.
You probably had Windows...
Funny thing, I bought this house in a nice area, but a short walk from a high crime area. It was built in the 50's out of concrete block (two blocks thick on the ground floor). All the windows on the ground floor were glass block and could not be opened. The upstairs was the real living area with lots of windows, but a full story (slightly higher than a normal house) up in the air. It had a back deck on the second story, with no stairs going down and with the deck overhanging the supports a good three feet or more (making climbing it really hard). It also has a monitored security system (I've since updated). I haven't had a break in yet. :) It is the OpenBSD of houses.
ISP's need to inform people they have bots and if they are infecting other computers they need their internet access dropped. Tough love.
And this gets the ISPs more money in what way? Many ISPs can pull up and print out a list of infected hosts by worm and by the amount of traffic they generate. They can automatically integrate this into their notification system and send e-mail to the host's account or shut down access. They don't because then they have to answer the phone calls explaining what is going on and that costs them money. It is easier to throttle that host's traffic and filter out the worm propagation traffic to their core than it is to deal with users' broken Windows installs. If it hits other hosts in the same router group, why should they care?
What do you think the C&C machines are running?
This is a good point and a lot of the IRC channels are running on rooted Linux boxes. What I find interesting is how the botherder community knowledge limits what they do. Linux desktops are not protected only by the fact that they are rare, but also by the fact that a lot of these people have no idea what they are doing beyond the tried and true tools. The community has the knowledge to root Linux servers and Windows servers, but aside from that they rely heavily upon metasploit type tools and existing code. If either platform undergoes a security transformation, expect the other to be the exclusive control channel for a while.
Be sure to check whether those are running OS X or Linux.
Perhaps my views are colored by the conferences I attend. Almost everyone I talk to ends up running OS X as their main OS, even while doing Linux development. A lot of people initially set them up to dual boot, but most that I know rarely end up booting Linux much and often end up nuking that partition after a while.
I don't need reliable hardware if I can have someone come replace everything every few months.
Yeah, I've done that with a machine room full of cheap Dell desktops where losing any given one was not a big deal. For my workstation, the time it takes for me to get a replacement and anything lost between backups is worth a lot more than the price difference between a Dell and an Apple (to my company who foots the bill). Not that I'm looking to upgrade to a new machine until Apple has cycled through a revision or two with Intel notebooks.
I doubt that that is going to be solved unless one distribution gains significant marketshare, unfortunately.
Actually, I see two ways for this to happen. One would be for a single distribution to gain a lot of marketshare. The other would be for one or more distributions to adopt a standard that they document and that also provides significant advances (like GNUStep can provide). In that case, I can see other distributions adopting it since it is not that much work (code exists) and it brings real benefits beyond just the standardization. I agree, however, this is pretty iffy.
That is quite interesting. It should be possible with some work to set up a system in Ubuntu that would act like that using deb packages and a modified file manager. It would also probably not be a bad idea to have themes set up that way, as opposed to the current, very inflexible theme manager. I will have to discuss it with some others. Unfortunately, it would only work with Ubuntu...
Yeah, the thing is these benefits can be combined with the above simply by adopting the existing GNUStep standard, with an already well tested code base.
do you mean registration as in network-based licensing, or registration as in registering with a server? A standard for the latter would be nice, but I doubt that developers requiring the former would be willing to use a standard system.
I was speaking of the latter, but I implied the former as well in other comments. The reason I think developers would be willing to adopt such a system is because of security. If you look at FreeBSD jails and the recent trend towards VMs as the direction of the industry you quickly see a few things. Restricting individual applications by resource can eliminate a lot of the security problems we face today, but only if the resources that are accessed can be standardized so users have less exceptions to deal with. This is especially important for end-user desktops. You download two new XML editing suites. When you run the first all runs smoothly, but when you run the second the OS informs you it is trying to connect to the internet, not using the service reserved for registration and updates and may be malware or spyware. On this aspect alone, you may be much more inclined to use the first suite and the latter loses market share.
I firmly believe the industry is moving this way for security and more granular control and I also believe it will bring huge benefits to the first group to bring it to the desktop user in a usable way. This necessitates an official service which will require a standard for widespread acceptance. If the reasons we were discussing are not enough to motivate developers, I think combined with this it will be.
Hmm... that sounds reasonable for nearly all desktop applications. What about an MTA or database, which will use /var?
I guess I should have been a little more clear. Preferences are dependent upon the user running the program, not the location it is installed. If a user is running their own copy o
Yes, there needs to be a common way for packages to install, and it really should include an option to install within your $HOME if your not root (if you really dont want users doing that, you can mount /home noexec). It should be useable from commandline and gui, so you can click on a package within whatever program, and have it launch the installer, which should work out if your root, try to use sudo or let you install to your homedir if your not.
This is solved by GNUStep as implemented by OS X. You install software by putting the package somewhere. If it is in your home directory it is accessible to just that user and if it is somewhere globally available, anyone can run it. You can install with the GUI by dragging it or via the CLI using "mv" or "cp."
It should be integrate with the existing package manager database, so you can update the packages from a single place,
This only works for Apple branded programs on OS X and is just the update portion, not the discovery and download. There should be a standard protocol for this so the tools can access multiple repositories for software downloads and updates, preferably cross-platform.
and it can recognise if necessary dependencies are installed. It should be able to call the standard system package manager to install dependencies, or install lowest-common-denominator prebuilt libraries from the package supplier if necessary.
GNUStep, as implemented by OS X, removes the need for most of this, by bundling the necessary libraries with the package. It takes less disk space than you'd think. Other people here have suggested that this breaks dynamic linking, but actually it merely requires dynamic linking to be modified to take it into account, as OS X does. Basically, users should never have to worry about dependencies.
It should support "online" and "offline" packages, wit the former being little more than a stub listing where to download the actual data from
I'm not sure this is a good idea. What if a novice user grabs an "offline" package from a disk and tries to run it without internet access? I think it needs to be clear that these packages are the programs and always have what they need. What you refer to as an "offline package" should just be a link to the repository and should be distinctly different from real packages.
And finally it should have advanced options available to users who want them, including ability to rebuild the package from source etc.
If you look in a GNUStep package you'll see the binary, plug-ins, resources, and system services each in their own dir. What should be added is a directory for source code and build instructions. Then any application can be built from source either automatically by the OS, by the package manager tool, by the user via a service supplied by the OS (right-click, build custom binary) or manually from the CLI.
I'm glad we can agree for the most part on functionality that would be useful. The other thing I was trying to convey was that almost all of the work necessary to get there is simply updating GNUStep to support a few more options and integrate with existing package managers, defining a standard for those package managers, and getting the major players to support it in their OS's. We have almost all the pieces, we just need to put them together and agree on a standard. Sadly, that is likely the hardest part of this.
Doctor, it hurts when I do this! Don't do it then.
You think just not doing the tasks users want is an adequate solution. Yeah, just keep thinking that. I'm sure users will be switching to Linux in droves any day now to get in on the wonderful lack of functionality.
The whole point of a distribution is to make sure ~10,000 disparate pieces of software work well together. By downloading some random blob that some clown has packaged without really understanding the distribution you're asking for trouble.
No distribution can include every piece of software I might want, including games, niche applications for a particular field, and unstable software that I still need. You're asking me to just stop using more than half of all the software out there, software I need. That is not a workable solution.
Distributions have package repositories for a reason.
And that reason is to provide the base software and the freely available software a user might need. It is not to obviate the need to run any programs not in the repository, because, like it or not, commercial binaries are not going away anytime soon. Ignoring them and hoping they go away is not going to work.
Aaaactually that's untrue.
Really? Last I saw Linux had a much smaller share of installations in non-server roles than OS X.
but most desktops in Pixar are GNOME (not that the desktop's really important there. It's about the apps.). Dreamworks are all Linux.
How is this representative of the desktop market?
Your model works for very small installations. It doesn't grow well.
Really? Have you seen how many sys admins it takes to administer a thousand OS X boxes? It is a dream come true if you can get the work.
But, understand that Redhat, who is driving Gnome (along with SUN and others) is solving the problem from the upper end. In other words, the solutions that will be generated will be alien to you.
Redhat has a motivated interest in keeping Linux difficult to administer and deploy because it is how they get paid.
I guess the term "average desktop user" is partly to blame here. I don't consider the "home/casual" user to be "average". My "average" is corporate user.
There are more home and small business computers than corporate owned workstations. Further, corporate users have a tendency to fight against using different systems at home and work. Thus, a centrally administered system would require that we move to service based computing for the home or face serious challenges. We'll see which way the market jumps.
None of your arguments address the fact that OS X works today and very well. Please actually try running it for a while as a normal system, and learning how to do things their way (not trying to recreate Linux workflows exactly). The truth of the matter is, Linux can learn a lot from OS X, and benefit from the way they have re-architected things if Linux developers are willing to consider that maybe some things can be done more cleanly. I use both systems and trying to use Linux as a desktop now feels like stepping into a time machine to the past.
You can argue that RedHat will soon be inventing a better solution, and if they do, great. That doesn't help me today though. If they come up with something better, I hope everyone adopts it, but I don't think it is reasonable to assume that RedHat is going to come up with some unknown system that will somehow be better. We have the solutions we need today, if we just piece them together in a way that works. The simple fact is, I can do a lot of tasks on OS X, simply and easily that I can't do at all on Linux. This is a deficiency in Linux and is enough to keep me from using it as my main workstation. A lot of other people have come to the same conclusion.
An application being 'self contained' means really just distributing it with its own set of dynamic libraries which is functionally the same as just using static libraries. Fifteen copies of libxyz on your system. Do you trust that they're all safe versions?
Sigh. If you're going to argue that an architecture is inferior, maybe you should actually read something about said architecture. OS X applications contain a copy of the needed libraries with each package, but does not use that version if it has a more recent minor update from another application, the same as any other use of dynamic binaries.
Your ideal 'solution' is basically ignoring shared objects and thus is like stepping back in time twenty years.
Before making claims such as this, actually investigate what you are talking about in future, please.
There is no dependency resolution, no automatic updates for apps, no upgrade system.
They have removed the need for dependency resolution. They are missing the other two parts. What they do have are applications that are self contained packages and are completely portable. These packages have all the resources easily available, handle fat binaries easily, and don't need to be installed or uninstalled. This is far superior to most Linux distributions, especially for use as a desktop/workstation. They need updates and repositories and services for licensing. Linux needs self-contained, portable packages, one format, and a standard protocol for updates and registration.
The application is the desktop icon? Or a link?
On OS X and OpenStep and in my proposed solutions programs a directories ending in ".app" and containing a particular structure. if you double click on the directory it launches the program. If you use the CLI or right-click you can navigate into the application. You can place the application on your desktop, or anywhere else you like. You can also create "shortcuts," "symlinks," or aliases to the application in a number of ways.
How would OS X know if a link is removed? Short of maintaining a database of file reference counts (and if its softlinks, it would be a database of file references, paralleling the directory structure). It doesn't.
Actually, it does. Go ahead an try it. If you move an application or file the links are modified. If you delete an application or file that is referenced, it asks if you want to remove the links as well.
And the desktop icon is simply representative of the application. It isn't the application.
On OS X the desktop is simply a directory in your home dir that is mirrored on the desktop. Applications can be in that directory and work just fine.
Different approach for other systems. As an example, assume that the file system itself is "virtual" in a sense -- things are NOT present unless called for by name (automount - remember I mentioned that?).
How things are arranged and linked is unimportant to most desktop users. They care about the interface you give them and whether it works as they expect. Saying it does not work as they expect does not mean you shouldn't fix other things.
That is, /opt/firefox/bin/firefox would not EXIST in your reference space unless you actually referenced it. Now, how is the icon to exist? There is a disconnect between the "desktop metaphor" and a physical file implementation.
Which is part of what confuses the hell out of most users. You lie to them by showing them an application when in reality you're showing them a link to an application. From the end user perspective, the icon that says "Firefox" instead of "Firefox Shortcut" with a modified icon, is Firefox and deleting it should delete firefox. Copying it onto a remote volume should copy a working version of firefox onto that remote volume. Not doing this is failing to provide a good UI to the average desktop user.
Even at home, I have a terrabyte online. I don't want to drill down hundreds of thousands of files
If you have to, your directory structure is overly complex and should be fixed. What is wrong with /Applications/Firefox.app? It works just fine and does not require anyone to drill down through hundreds of thousands of files. Note, we're talking GUI applications for end users here, not CLI commands, which have no place showing up in the GUI.
Or, lets say your new OS X application can handle files of type "video/mp4v-es" (just choosing a real mime type at random). You want your web browser to create a window, and pass that window handle and the file stream to this application when the data type appears in a web page. Ok, how does THAT work? PS. No pre-launching of the application allowed here (just to make my point as to what an INSTALLER should be doing). What browser is it, anyway? You want to "launch" an existing file (which is why you just copied the application to your computer, after all) and have it come up in "edit" mode.
Sigh, you're making this way harder than it has to be, because you're couching everything in terms of how Linux would do it, not how a cleanly architected system should do it. If I want Firefox to open a certain video type in a certain player, I either implement that player as a plug-in for the browser or the browser looks at the mime type and asks the OS what application I have assigned to it and if it is allowed to auto-launch it or if it is to just save the file and wait for me to do so manually. If it is allowed to auto-launch (
Eh?? Linux is booming in the film, animation and scientific industries.
Linux is doing fine as a server OS. It is doing poorly as a workstation in most fields. It does well in the film industry because of filmgimp, but even there a lot of people have moved to OS X for the workstation. As for scientific "industry" the number of macs as workstations and laptops in most of the sciences has skyrocketed in the last few years, much more so than Linux.
The problem with saving "preferences" in the application directory is that it makes the application not shareable. Big no-no there.
Globally available applications have a global configuration setting configurable by the admin. These settings are inherited by the users, but unless specifically configured otherwise are overridden if a given setting conflicts with a local user's setting. Thus, users get good defaults when they first run an application. Any changes they make effect only their use of the program. The system actually looks at both files when it runs the application, merging the two. Changes a individual user makes are not viewable by other users.
However, a statically linked Linux application, x86, built for a 2.4 (2.x) kernel will run on Linux, OpenBSD and Solaris. No rebuild or relink needed. It's not the issue.
That's pretty narrow and it means one program cannot benefit if another has a more up to date version of a library. A fat package with multiple binaries for different platforms and architectures and source for building a custom binary, all sharing the same resources within the package is ideal, IMHO.
And, to boot, package management is not an issue here EITHER. Package management is only a problem if shared objects are in use. Not an issue with the class of shareable/copyable applications we are talking about here.
Regardless of if a package is sharable I still want to be able to have it operate in a jail and I still want to keep it up to date with a centralized service rather than individually. I still want it to use a standard registration and licensing service so I can regulate, schedule, and monitor it.
What is the issue? A well written x86 Linux application is about as portable as you can get.
There are multiple kinds of portability. One kind is the ability to run the same package on multiple platforms. The other is the ability to move functional copy to other locations and systems and still have it work. I've never successfully IM'd a real Linux app to someone and had it work for them on the other side. I have done so with .app packages from OS X.
As an example, how are your "desktop icons" and/or "program launcher" going to be done? How is this updated if the program is removed? How are links to document types created? How are extension programs for your browser installed? These are the issues that an installer should address.
A "desktop icon" can be the functioning application or it can be a link to it. If you remove a file and it has links, the OS warns you, just like OS X now. Program launching is simply double clicking or single clicking or opening a file with that program as the default application. Everything you mention is a solved issue on OS X already. Get with the times.
The "low-level" issues have been addressed. Shouldn't be a problem -- if they ARE a problem, the software itself can be considered defective.
I think you completely missed the point.
why continue to paint yourself into the proprietary, lock everyon eelse out world then? people don't like malware, viruses, reinstalling their os every year, etc, either.
This isn't an either/or choice. I never re-install, or get viruses or malware and I install most applications to my desktop by dragging them into my /Applications folder.
If you use the distro's installer you get a subset of software. What if you need to use any niche software, you know the stuff someone wrote on one distro and never even tried anywhere else. If you're compiling from source, you can run into all sorts of issues with dependencies, the config script, etc. Anyone who uses Linux for real work runs into the software they want to use, but just doesn't work on their distro.
i don't see your issues as being real issue. i've been on linux 1 year and i haven't had that much trouble transitioning.
I run Linux every day and have for many years. It is a great server, but it is weak for a general purpose desktop, especially for non-expert users. People complain about not having a standard package format all the time as well as about the lack of functionality of all these package managers. Blowing this off as "theoretical" is just another way of ignoring the problem. What ever happened to the open source community rallying around open standards? Linux is quite simply lacking a lot of the functionality I want for managing applications. They are not portable in that the same package won't run on all distros and they are not portable in that I can't just copy an installed application from one machine to another. They don't have a standard protocol for repositories, registration, or updating. If all the players would agree to a standard, then we could all pick the best tools and stop duplicating tons of effort and instead concentrate on improvement. The cynical side of me says this has not happened because the major players like using their repositories as a way to make it harder for people to migrate and as a way to collect support fees for something that should not be hard in the first place.
You presume that the Windows Installer actually works.
Get this message through your head. If Linux installation and package management is as good as Windows or even slightly better, Linux still loses. Linux needs to have good package management because Linux does not have a monopoly and is subject to pressure from users.
If followed, the binary will be "portable" across multiple distributions (at least on the same processor and major OS version). The "installer" can the be trivial -- and can simply be a self-run shell script.
Or we could get rid of installers in the first place and have truly portable applications that are packages. All a user has to do is run OS X for a month and run Linux for a month. You'll notice both platforms are deficient in a number of ways. Combine the two and add a little extra. Call it a standard and get everyone onboard and we'll all enjoy security and functionality no one does today.
Basically, "Linux Installers" are trying to solve problems that don't really exist.
I have an audio program installed on my machine. It is freeware, but no longer distributed. My buddy wants to use it, but doesn't know where to get a copy. So, I drag it from my Applications folder into my IM chat with him. It transfers and he runs it from his desktop. He likes it a lot and drags it onto his thumb drive and erases it from the work computer he was using. He takes it home and drags it off of the thumb drive onto his home machine and uses it there. Later, back at work, he needs it again, so he plugs the thumb drive back in to his work computer and runs it off of the thumb drive. It even still has his preferences saved. Did I mention one computer was 32 bit and one was 64 bit? One container ending in ".app" that you double click to run, drag where you want to transfer, never worry about "installing" and drag to the trash to uninstall.
This little anecdote is an example of things I can do on OS X, but can't on Linux
There are also things I can do on Linux, but can't easily on OS X, like schedule auto updating of all applications at 1AM every Wednesday.
I want to be able to do all of these things on Linux, OS X, Solaris, OpenBSD, etc. I want to be able to use the same applications packages between them and I want to be able to use the same or different tools to manage them because they all conform to a standard. Anyone here claiming that we don't need such a standard had better have some bloody good explanations of how I can do all of the above on Linux without reading any manuals, writing any scripts, or going out of my way, because those are things regular desktop users want to do.