There are several problems around software installation and software management in Linux. Also, there are several different point of views, opinions and different “golden ways”, depending on whom you ask. This article is the first in a small series dedicated to that problem field, and it will show my position.
It will be important to know from which position I write these articles, which point of view I have and which background. It makes it easier to understand why I care more about specific aspects of a problem than you or others might care.
I myself do not really need package management or software management: I can build software by myself, and I can also write my own rpm packages for my distribution – in fact I am a maintainer for some fedora packages.
But since I know my way around with computers I’m asked once in a while by other people if they should switch to Linux – and my answer is always no. Because these people will not be able to install software besides the tiny bits provided at the distributions repository. Sure, Debian for example has a large repository (I think it is actually the biggest) but compare that just to the amount of software provided at gnomefiles.org and kde-apps.org. Even if you just take the cool and good software out of these databases you quickly realize that Debian integrates only a fraction of these apps. And the situation is worse for all other distributions.
So, when my friends ask me I cannot suggest installing Linux because they will be lost when they want to try it out, want to play with it: searching and installing new software is part of that game, and they would be bound to the repositories. They will not be able to search around the internet for cool apps or tools and install them. The reason is obvious: there is no widespread universal installer on Linux, not even an equivalent to msi packages. And of course compiling source code is out of question.
For me that is the main reason why Linux has not found adoption among main stream users (read: people who will never know what “alternative package management” or “compile the source code” means). And I’m afraid that Linux will not be adopted by these users anytime soon, until that problem is solved.
Keeping this in your mind my position comes clear: I will take the point of view of a user. Of a normal computer user who, in the end, cares about monopolies and vendor lock in, maybe even about the ideas of FOSS, but not about technical details. The user just needs a working computer where he can perform all usual tasks – and installing software is part of that task.
Some last words about me: I know my way around with package management and alternative package managers – other articles in this series will deal with such things. So, before you suggest me now something like klick or autopackage or whatever – don’t! I know these. And I will write about them. Until then, keep in mind that I wrote widespread. That is the important bit.
Also, if you have personal or philosophical or religious feelings against a widespread installer for Linux because it is the root of all evil – I will deal with that argument as well. But you might guess that I am not your opinion at all.
/home/liquidat 
I look forward to reading these articles, but I’d like to ask. Have you considered that the harder it becomes for a user to escape their distro’s repository, the less likely they are to need to. I know several non-techie users, of witch only one ever installs *any* software. The others just use a small set of applications provided by their company. And all easily replaceible useing only official debian pacakges. This may go for the one who installs his own software too. He dosn’t install much.
BTW I assume you know that the LSB has a packageing workgroup. I’m not sure what happend but the mailing list has gone silent 😦
Thanks for the comment – and yes, I know about the packaging group and the posts made by Ian to announce that list. And to me it seems as well that this list has gone silent.
But of course you are right – the more packages provided by the distro the more unlikely is it that a user wants to install other software, but I will address different user types and different user needs in other articles. Some catchwords are by the way “niches”, “young” and “time to react”. 🙂
Actually you misinterprited my first point.
“Have you considered that the harder it becomes for a user to escape their distro’s repository, the less likely they are to need to.”
Did not mean that bigger repositories makes it more unlikely that users want to leave (allthough thats true)
It actually ment that users with _low technical skill_ are less likely to want to leave the distro’s repositories.
I won’t respond to you’re catchwords untill they appear in an article 🙂
Ah, ok, you’re right, I misunderstood you. But now its clear.
But in this regard you are right too, but to digg deeper in that discussion it is better to have some stereotype users pre-defined. That’s one of the next articles task 🙂
The only standardized part in a Linux or GNU/Linux system is the kernel (the X server is almost standardized also). Everything else (desktop environments, browsers, package managers) is not.
While this might be advantageous for some users, it is not for general consumers.
I agree with you, in my opinion not only the package manager should be standardized in Linux, but also other parts of the system. Althought this may seem difficult for developers (who are already in their holy wars kde vs gnome vs xfce, or openoffice vs koffice ..), it is the only way for Linux (or GNU/Linux) to become a mainstream choice for regular users.
@hanisaf:
The question of standardization is also a question for definition: the kernel differs from distribution to distribution quite a lot since different distribution include different patches.
Also, the shell can be different, there are even distributions which do not use the gcc as the main compiler.
Still, the way to focus on one solution only as you recommend is not the important part: the important bit is the integration! If you come to a point where every Gnome application behaves and feels like a KDE application in KDE and the other way around the problem about different desktops is solved.
Therefore it is a question of common and shared standards, standardized interfaces and overall of integration.
And a last word about koffice and openoffice: these are simply just applications, nothing more. And it doesn’t make sense to focus on one application or the other. Especially in the world of applications you need more than one solution to have real competition – because this competition will lead to innovation.
@liquidat
In your post you touched the real problem in Linux distribution. Competition.
Because KDE and Gnome are competitive, each group thrives to add more unique feature to its system in order to be more appealing and useful to the user.
Overtime, the two systems are diverging and the hope of a standard is diminishing.
In fact, if -as you said- “every Gnome application behaves and feels like a KDE application in KDE and the other way around”, the choice between KDE and Gnome will be trivial, and meaningless. A user chose a system for the features that make it unique.
As a new bee, I started with KDE, then I switched to Gnome, because at that time KDE was not stable. Then I discovered that the application that runs natively on KDE, have problems with Gnome, and the way around. I also discovered that my personal info, like contacts, calendars and emails are not shared between the two systems.
So I went back to KDE and lived with it.
For me I was determined to use Linux, but for other users they will go back to Windows once they discover these inconveniences.
Concerning koffice and openoffice, yes they are applications, but the integration of an office suite with the desktop environment (the ability to preview, create, or send docs from file browsers) is very important in nowadays systems.
Can you integrated koffice with Gnome?
To conclude, choices are good in life, but not for all people. And sometimes they can be an overhead over the person (More is Less). I really feel that a standard should emerge, and without it the Linux systems will not be able to adapt to the regular users.
“Overtime, the two systems are diverging and the hope of a standard is diminishing.”
You are right, that would happen – if no one would care. But in these days we are far away from that, the opposite is the case. For example, with the release of KDE 4 icons and – even more important – mimetypes (aka start ogg with application xyz) will be shared between KDE and Gnome.
That was never the case before!
But of course, there is still *lots* to do, and it must be done to be appealing to end users – especially the field of personal information you mentioned is chaotic!
But there we might see the rise of Akanodi – it is planned as a sharing point of exactly these information, and between both desktop environments.
On this note. I *think* KDE will be the standard, probably not the only option definitely not, but it is already the most widespread and as of KDE 4 it’s going to look better, react more sleek and even work faster on older hardware.
This comes from a Gnome user. Have a look at: http://polishlinux.org/kde/kde-4-rev-790000-better-stability-and-performance/ and see just how professional it actually looks.
I agree that there must be competition merely because that will encourage innovation, as mentioned. Linux has been making huge headways on the desktop market over the last couple of years, but it is still mainly a SERVER OS for the big guys like Linus himself (the main kernel maintainer). To him everything is about stability and absolute perfection and I’ve read numerous reports that he rejects many new implementations simply because he doesn’t see the need/use of it.
This might be because of his own perspective of Linux (as a server os perhaps?) over others who are trying to make it a better desktop OS.
Just an examply anyway…
We need standards and a universal installer would be one of the best things for linux.