The original story was a FUDsterpiece.
|
| Author | Content |
|---|---|
| caitlyn Jun 23, 2009 5:24 PM EDT |
The original story was a FUDsterpiece. This response isn't perfect but it makes the point well. For most normal users running a distro with a well stocked repository software installation in Linux is easier than Mac, Windows, or an iphone. It's point and click. |
| hkwint Jun 23, 2009 6:06 PM EDT |
Quoting:It's point and click. Well, at least if they use their package manager and don't want to download their apps manually. Lots of people still don't get this, sadly. Someone should hack the browser so that all searches for software start the package manager and close the search or something like that I guess. |
| caitlyn Jun 23, 2009 6:15 PM EDT |
Even where you download, for something like Firefox or OpenOffice or any number of others I could mention it's download, open file manager, point, and click. That's kind of like Windows, isn't it? |
| aronzak Jun 24, 2009 12:44 AM EDT |
Yes, this is probably getting a lot of page views, but this is a valid issue. There are a few specific points that stand out from the original article. One: "The first step, is to face the truth and admit that there is a problem. " This rings very true. Two: " Every Linux distribution at the moment confuses system software with end user software" They are treated in the same way. The issue of needing root permissions is valid, as it should be possible to easily install user specific applications (and this isn't an issue in systems with a single user) without root access. Why are system libraries and games installed in the same way? Why is downloading a package from a third party website so risky? Because the current system is set up only to allow root to install software system wide, automatically allowing the package system wide access. This is the main reason why third party packages, and wider contribution in the packaging of applications is shunned. Finally, as much as you may disagree with their business, there really does need to be an easier way for proprietary software developers to be able to make software for Linux. That would lower one of the barriers to companies releasing software for Linux, and therefore make more people consider the platform. Sure, it's great that there is work into developing free alternatives to popular software packages, but while they are not yet best of breed, we shouldn't be forcing users to stick with them by offering only the free versions. |
| caitlyn Jun 24, 2009 1:49 AM EDT |
@arzonak: Did you even read the response? Many distros designed for ease of use (i.e.: Ubuntu) DO NOT require root access to install software. You need your own user password, nothing more. There is nothing "shunning" third party packages and a lot of ISVs do develop successfully for Linux, particularly in enterprise space. Ubuntu and many other distros have a non-free section to their repositories to make installation of third party proprietary software particularly easy. Heck, even a relatively small distro like Vector Linux managed to automate the installation of Skype. The package is basically a script that pulls the Skype package from upstream and scripts the installation, making it all transparent to the end user without violating the license terms of the proprietary software. Sorry, no, it does not ring true. Installing software is generally easier in Linux than in Windows. At least in Linux in a distribution with a well stocked repository it's one stop shopping. |
| aronzak Jun 24, 2009 3:07 AM EDT |
The privileges issue is valid. "You need your own user password, nothing more. " This makes no difference. This is still giving superuser access, just with more convenience and less security. The underlying multiuser security architecture, with user level access and superuser level access is the same. It still doesn't make sense to be installing programs and device drivers in the same way. You've probably never been to a university, so you wouldn't understand, but there are such things as multiuser Linux setups that not everybody gets administrator access to. Manually compiling and setting libraries and binaries in their own user folders is a huge PITA (especially if there are a huge number of libraries needed.) The point is, it is still a pain for application developers to have to lease with package maintainers etc... and try and make sure their program would actually work on systems with different DEs, sound architectures, etc.... This is irritating ant it hurts small projects, and makes life difficult for everyone. I don't have the patience to explain in detail why you're wrong. Sooner or later though, developers are going to need to pull their heads out and start working on this. |
| Sander_Marechal Jun 24, 2009 4:01 AM EDT |
aronzak: You know what the strange thing is, every Windows user I moved to Linux (and there are quite a few) absolutely *love* how package management works. They find it fantastic that they can simply click "Add/Remove software" and search through the list. They love that they can add and remove programs without rebooting and that installation doesn't take hours. They love that automatic updates takes care of all their installed software and not just the OS. And they absolutely love it that their kids can't install random cr@p anymore because they don't know the root (or sudo) password :-) "No annoying virus scanner" and "package management" are the two praises I hear most from all my converts. |
| krisum Jun 24, 2009 4:04 AM EDT |
Quoting: It still doesn't make sense to be installing programs and device drivers in the same way.You could always use something like autopackage/klik, if installing from source is not an option. Quoting: The point is, it is still a pain for application developers to have to lease with package maintainers etc... and try and make sure their program would actually work on systems with different DEs, sound architectures, etc.... This is irritating ant it hurts small projects, and makes life difficult for everyone.So which project have you been part of and faced these issues? Probably you should try and get help for those issues on help forums instead of complaining on a news site. With freedesktop and LSB, most of these should be no longer any issues. |
| jacog Jun 24, 2009 5:24 AM EDT |
I saw a forum/blog/news post not too long ago that complained that "Linux" (I think the name is misused) does not have a centralised way to add and remove software. It was a baffling comment. |
| r_a_trip Jun 24, 2009 9:03 AM EDT |
@ Aronzak No, the first step should be to determine if there is a problem. Not admitting you have one just because someone tells you to. The package manager/repository vs install shield/click next-->next--> next methods has been thoroughly chewed through in the past 5 years and it yielded the insight that GNU/Linux has a good, usable implementation in the package manager/repository method. That Windows refugees have a hard time adapting to another method, shouldn't force implementations on GNU/Linux to mimic the Windows way. What you propose is an extra local sandbox (/home/< username >/UserApps) for the user to play in. This already exists in AutoPackage and still closed source software vendors are moaning about difficulties. There shouldn't be any difficulties at all. Just compile everything statically and dump the package either in /opt (if rights elevation is an option) or dump it in /home/< username > and place a shortcut on the desktop. Besides, Codeweavers, Sun, Transgaming, Google, Skype Technologies, Corel, etc. don't seem to have any problems with making proprietary software working well on GNU/Linux and they do it in a user friendly way to boot. The problem is not rights elevation. On systems with a single user, that rights elevation takes place easily, because the single user is also the administrator. It isn't even a security issue, because malware can also be successfully run from a user account. Software form untrusted sources are always risky, irrespective of the OS. The distinction between root and user is to keep the system and the user accounts safe from each other. Your argument about having to compile libraries under your user account is something else completely. That is you trying to circumvent the policies of the owner of the machine, because clearly you don't have elevation rights and you didn't get permission to have that software software installed for you. So basically you are going rogue and install the software anyway. It's a good thing you can only dump it in your user account, because the rightful owner didn't want it system wide. Did you know that the situation is the same under Windows? If it is set up properly, you are in a limited account and need administrator privileges to install software. The only problem with Windows... most user machines run as administrator by default. Once you get in a corporate setting, enforced policies and limited user accounts will give you the same freedom to install software you encountered on the university *Nix machines. None whatsoever, unless you can drop it in your user account under the radar. |
| jdixon Jun 24, 2009 9:23 AM EDT |
> It still doesn't make sense to be installing programs and device drivers in the same way. It does if there is any possibility of your system having multiple users. If it's a solely a single user system, then you might have a point. However, how many home users actually run single user systems? I'd say that most home systems are multiple user systems. > ...but there are such things as multiuser Linux setups that not everybody gets administrator access to... Well, duh. There's a reason for that, it's called system security. If you want software installed on those systems, you contact the administrator. It's not YOUR system, you don't have the right to be installing software on it. > The point is, it is still a pain for application developers to have to lease with package maintainers etc... and try and make sure their program would actually work on systems with different DEs, sound architectures, etc.... This is irritating ant it hurts small projects, and makes life difficult for everyone. You know, I'm getting really tired of this "commercial support for Linux is so hard because there are too many options" garbage. Yeah, and developing for Windows 2000, Windows XP, Windows Vista, and soon Windows 7 is soooo much easier. Pick your supported distros, specify which ones they are, and leave it at that. |
| caitlyn Jun 24, 2009 9:29 AM EDT |
Quoting:Pick your supported distros, specify which ones they are, and leave it at that. Which is why many enterprise ISVs support Red Hat and SUSE and not much else. Those are the two leaders in enterprise space, accounting for >95% of the market in North America and Europe. Both use rpms and for many apps just one will do for both. I have a Lightscribe enabled CD/DVD burner. The software for it, whether from LaCie or HP, is proprietary. They provide a single generic rpm. I've installed it not only on rpm based systems, but also on Slackware, VectorLinux, Ubuntu, etc... I used alien on Ubuntu. Slackware and Vector both offer rpm as part of the distro. It seems to work on everything because, when you come down to it, Linux is Linux is Linux. If you compile with static libraries it's going to work on most anything. ISVs only have to provide multiple packages when they use dynamic libraries and fail to provide the distributors with what they need to build their own packages. If they are that tight about their code then they have to build the packages themselves. Even on the desktop you are talking about just a handful of major distros in any case. The whole "it's hard for ISVs" is a red herring used by Microsoft shills and astroturfers. Unfortunately it also gets repeated by some Linux users who just don't know enough to know it isn't true. |
| bigg Jun 24, 2009 9:35 AM EDT |
I don't know enough to give a reasonable answer, but I'm curious why Mozilla and OOo can provide packages that work with almost any distro. For example, I can download Firefox, unpack it in a directory of my choice, and it works fine. Similar for OOo: Robby Workman just repackages the official binary, and there are no dependency issues at all. Firefox and OOo are not small projects; if they can do it, why can't others? I will reiterate that I have no expertise in this area, so there may be technical reasons. |
| hkwint Jun 24, 2009 11:10 AM EDT |
Klik and Autopackage are great, in my opinion. If somebody made a package for one of those two, that package would run on any Linux distro supporting Klik/Autopackage and LSB. Problem is, it seems the development of both have come a bit to a halt. I read somewhere the Autopackage developer stopped because he was constantly being flamed at. I don't know if it's true, but that's a shame, in my opinion. Especially because packagers for all the distro's we have are duplicating efforts. Lots of people working on distro X (at least half of them I guess) are doing package management, the same package management that distro Y does. If the folks from different distro's were all working on the same autopackages instead of some of them on ebuilds, some on .tgz's, some on .debs and others on .rpms, the quality of the packages would probably be much better, and bug fixing would be faster. Also, when installing autopackages or klik packages, one doesn't need to type _any_ password. For klik, one package is one file, so uninstalling is as easy as deleting that file. Klik2 would also feature an 'ro' part in the image and an 'rw' part, so you could still contain whole Firefox + themes + your bookmarks in one single file. One should also look from the point of the developer. When one distributes ones package, it's patched differently and packaged differently for different distro's, and installed to different dirs. Also, different distro's have different versions. It would be great if the developer could make an autopackage, because than he could point everyone to the newest version and would not have to wait for the distro's to pick up on his newest release, and apart from that it would be much easier helping the users if they have problems or report bugs. This - of course - also goes for ISV's, and that's why they decide to only provide RPM's for - let's say - Redhat and SuSE. However, if more people were to adopt autopackage or klik, if they provided autopackages for RedHat it would almost automatically work for all distro's. Then there's the issue of 'distributing' packages: It would be nice if you could buy a DVD with autopackages that work on any Linux distro (adhering to some version of FSB I guess). You see, telling "Start the package manager and install X" is easy. However, that only works if you don't have to pay €1 for every 100MB you download, or worse, when you have a limit. The latter goes for large parts of the world. So I agree Linux software installation could be lots better, and autopackage and Klik(2) adddress most of these issues. Improvement is always welcome I guess. However, these two methods didn't gain much momentum. There can only be one reason why. Sure, Linux-users and package-maintainers react slow - and sometimes even unwilling - to change, but that's not the real reason. The real reason would probably be: Because current fragmented package distribution works well enough for most people, and most people feel it's not flawed. And package maintainers don't see a problem in duplicating the efforts others did before. |
Posting in this forum is limited to members of the group: [ForumMods, SITEADMINS, MEMBERS.]
Becoming a member of LXer is easy and free. Join Us!