Hi Guys,
I am currently writing an Ubuntu-based distribution, and I was wondering what you guys would want more out of Ubuntu, so that I can incorporate it into my OS. Also, is there a way to further improve the Linux kernel's speed and efficiency?

Although I am not done yet, the iso image file is 1.1 GB as of right now (I'm thinking of uploading it onto a torrent). However, is there a way I could upload it to a mirror (in other words, is someone willing to host a mirror or has some extra space to host my OS?)? The reason why I need a mirror is so that I can host my own apt repository so that my OS can download its own OS version updates instead of Ubuntu's.

Also, how can I change the Ubuntu themes, such as the usplash and desktp themes, and the default GRUB entry which is installed after a fresh Ubuntu install?

Thanks,
Flouran

What would your distribution offer that is not in Ubuntu, Kubuntu, Xubuntu, Christian Edition, Muslin Edition, Satanic Edition, or Mint and could not be added as a program to any of these?

...why, what for, what would it achieve?

Well, I'd like it to be a bit less like Ubuntu, and, if you are going to make kde available, I'd like a better, more 'finished', kde, please.

But you'd probably be wrong to conclude that because I want it anyone else does. So this goes back to the point '...what for?', because until you define that, it is difficult to comment on what would help it achieve its objectives.

Probably, up to a point. If you are prepared to put work into optimising for a particular application, there are some small gains to be had. But the stuff you do to optimise a server probably de-optimises a desktop. And, maybe, either de-optimises things in an appliance context, where memory usage is very important. Or maybe you can contribute optimised algorithms to the kernel development group, if you have good ideas that are generally applicable.

But, again, what for...?

Also, be very careful to stick within Ubuntu's allowed approach to trademarks and naming.

I want to create my own Linux distribution (which would be a derivative of the Ubuntu Desktop Edition) from customizing an Ubuntu LiveCD. My intention is definitely not to simply make a copy of compiled binaries and call them my own. I want to create a distribution that has more software, packages, and more functionality than the current Ubuntu desktop systems.
Originally, I thought I would base my system off of Debian. I soon realized that I wanted to integrate Ubuntu's easy yet powerful desktop functionality into my own distribution. Thus, my OS would be an unofficial branch of the Ubuntu desktop edition (much like Kubuntu and Xubuntu, although they are official). want my OS to be so user-friendly that people will want to support/promote Ubuntu and the rest of the Open-Source world. What problems do you think might I encounter if I try to create an unofficial branch/customization of the Ubuntu Desktop Edition? Or do you guys have any suggestions on what I could do better in order to help achieve this goal?

The advantage of having more software is that everything comes pre-installed, and from the first install, the Linux distribution achieves greater functionality than a distribution which doesn't come with pre-packaged applications. You see, most Linux Gurus/Long-Time Users don't really care about their Linux operating system coming with many programs which can play mp3's, movies, and install Windows programs (the list goes on) because they themselves have the ability to easily find these software programs on their own. However, for users who are simply switching from commercialized, "easy to use" operating systems( although Windows is not at all easy to use, it's just plain painful ), and are used to having everything come out-of-the-box, it is somewhat of a challenge to add more applications in order to fit their needs. I remember when I first switched from Windows to Ubuntu; I couldn't play mp3's, flash movies, install Windows programs, and it was nightmarish because I had installed Linux over my Windows, which had all of those capabilities. My main goal is to create a Linux distribution (based off of Ubuntu, most likely should not be a derivative of Ubuntu, but its own distribution), which is not only easy to use and welcoming to average end users who are switching from proprietary operating systems, but enticing to computer programmers/Linux Gurus as well. Hopefully, then, the stereotype of Linux being "hard and unusable for the average end-user" could be abolished. That's my philosophy.

Also, I the main reason I want to make the Linux Kernel faster is to further improve the efficiency of my system. I found a software called, aio_opensource.tar.gz, from http://sourceforge.net/projects/linuxperf/. Has anyone tried this software?

Thus, although I plan to have my OS to be distributed via BitTorrent, the only way it can download OS version updates (like Ubuntu 7.10 can download Ubuntu 8.04 via its Update Manager) is if someone can host an APT repository/mirror to store my OS version updates. Also, I am open to any suggestions from anyone who are willing to help in improving the functionality of the Ubuntu system, and I will be especially grateful if someone could host an APT repository for my OS (I've looked into PPA, Personal Package Archives, but they only allow up to 1 GB of free space, whereas my OS is 1.1 GB in size as of this moment).

Any help is greatly appreciated,

Aran Nayebi

This would be at least slightly difficult (but don't let me put you off if you are determined).

In essence, you are trying to do the same thing as Ubuntu, but you are trying to do it better without the teams of people that Ubuntu have (in a sense then it would have been easier had you said that you were trying to focus on something slightly different from Ubuntu themselves, like, say, multimedia or CAD/CAE, as then you could easily see why you might do the area in which you wanted to specialise than the generalist Ubuntu distro...or, you had said that you wanted to make a media-centric distro with a different GUI or something).

I see this as a making your life more difficult.

...and there are reasons for each one of these issues, and in making your own distro you'll be facing the same problems as, say, Ubuntu

Many people would like their system to be faster. Unless you have some actual idea how to do this (if we adopted this revised algorithm for memory management, if we adopted that revised algorithm for task scheduling...) then I don't see that just wishing it will help much.

So far, the only thing that I've seen that you will really be doing differently ("is not only easy to use and welcoming to average end users who are switching from proprietary operating systems, but enticing to computer programmers/Linux Gurus as well" doesn't seem that different from Ubuntu itself) is that you might have a better selection of packages available by default on the DVD (if there is one, that is) - in this respect, you might argue that Ubuntu is non-ideal, or in the default install but if you have access to the net, the selection of packages available is hard to criticise, except maybe to say that sometimes there are b more alternatives than are strictly necessary (but, how big a problem is that in reality?).

And, I'm afraid, that while you can say that the default Ubuntu install isn't ideal for people without good net access, your approach doesn't sound to solve this problem (and its unclear how much of an issue this is). And while it is true that the method of installing new/extra apps isn't as obvious as it should be to new users, the solution to that probably involves making the method of installing apps more obvious to new users, which is an area on which you haven't commented.

You see, my OS uses Ubuntu as a base system, thus integrating the code made by the teams of Ubuntu developers into my own distribution. As a result, my OS will download similar (but not the same) updates as Ubuntu. However, it has a hundred more apps by default (this is a literal amount), so that the end-user doesn't necessarily have to know what program opens which file, but instead they can download a file without any hassle and not have to worry about finding the program which best suits their needs (since it all comes "in the box"). For example, if a user is not Linux-savvy and thinks that they can install Windows apps on Linux, they will be able to, without having to figure out what Wine is or have to install it. As a result, it is much easier and less of a hassle than Ubuntu. Or, say they want to install a program, and it is a tar-ball and they need to compile it from source but don't know how; guess what, there is a program which automates this process as well. Or, say they want to mount a Mac OS X dmg file which contains a Mac program or all of their documents (if they are switching over from Mac or use both OSes), they can do that as well, without searching the internet. (The program which does this, in case anyone is wondering, is a Nautilus script I made that has GUI based and point-and-click prompts, which you can download from: http://linux.softpedia.com/get/Deskt...er-39815.shtml).

But, perhaps you are correct in that I should make the user more aware of the apps they are using. Thus, hopefully, I can re-program Ubiquity to list all of the programs to be installed by default, and the user can then select what programs he/she will use to best fit their needs, much similar to the Fedora installer (and if anyone knows how I can re-program Ubiquity, that would be great).

Also, although I am not sure of this yet, I plan on patching the Linux kernel using a C program in order to make it a little more reliable (I haven't looked into it fully, but I will in the near future).

Hi,
What problems are you having with the Linux kernel? What version?
You could go to 'The Linux Kernel Archives' (kernel.org).

I have no problems with stability issues for the 2.6.24 on my hardware. That's not saying others don't have a problem though.

How are you going to make the kernel more reliable? What are the issues that you seem to think that need addressing with the kernel?

This link and others are available from 'Slackware-Links'. More than just Slackware® links!

What you are wanting to do has been done in various ways by various people with mixed success. You might want to look at what has been done with Linux Mint, MEPIS, Kanotix, Sidux, or any of the other half a million "easy-to-use newbie desktop" debian/ubuntu derivatives have done and figure out how you are going to solve the usability, stability, and performance problems differently then they have. Then just do it.

According to the aio software mission, namely the program I will use to "improve" the Linux kernel for my OS, it makes "Linux better by improving Linux kernel performance, with special emphasis on SMP scalability. We measure, analyze, and improve kernel performance, focusing on platform-independent issues, by using open source workloads" (http://sourceforge.net/projects/linuxperf/).

Regarding the design of my OS, I really am leaving it up to you guys. I know exactly what I want from a Linux distribution that I don't see in most distros, and have incorporated those aspects into my OS. Now, I leave it up to you guys. What do you mostly want out of a Linux distribution (programs, features, etc.) to best cater to your needs as an end-user, so that I can incorporate it into my OS?

I want stability and reliability, i.e. stuff that doesn't break all the time and comes with a steady stream of updates and security fixes. A strong community is also important. That would certainly exclude an Ubuntu derivative maintained by one person. No offense meant but have you thought about the amount of work and expertise that this will require, not to mention which niche the distro (distro, not OS!) will fill?

/CZ

I think that is exactly the wrong direction to go. The amount of software available is increasing faster than download speeds or media sizes. Your ability, when constructing a distribution, to guess which packages people will want won't be high to begin with and will go down as more packages are created.

I agree there is a big problem there. A moderate effort by someone who cares and has some time to spend could go a long way. But the focus should be on making it easier to add the applications you need, not on guessing the applications and having them already there.

Using Synaptic it is very easy to install applications in any Debian based distribution, if you know which applications you want to install. But ...

1) Linux applications, including synaptic itself, are amazingly badly named. Given the name of an application, can you guess what it does: good luck. Given an idea what functionality you want, can you guess the name of the application: forget it.

1a) Whatever is the best new user tool for package install in a distribution should be clearly identified and easy to reach in the initial menu system. Before Mepis, I used a few Debian based distributions in which Synaptic was well hidden and not mentioned in new user doc's, so I was trying to use apt-get (not good for a new user with a preference for GUI). Synaptic was available, but how could I guess that was the program I needed.

1b) Once you're in Synaptic, you're way better off than with apt-get. But it still is pretty hard. Someone should add a "suggested applications" tab to Synaptic, and/or improve its existing categories breakdown to have more than one layer and some help text. There are too many categories and too many choices in each category and too little chance that the person who decided on the category for an application thinks the way a new user thinks. So guessing a keyword and searching in All ends up working better than going to the category you want (but it doesn't work very well).

If you think you can select the right 1GB worth of applications for a new user, then you ought to be able to select a slightly larger set and arrange them by purpose, and for each purpose give a real description plus the names of the reasonable choices in Synaptic for that purpose and maybe even the pros and cons of each choice where there are alternatives. That ould be what I call "suggested applications". The key is that a new user needs help selecting them. Synaptic already makes installing them easy enough.

2) Often you don't even know the functionality you want. You just know the file extension of the downloaded file that you want to be able to open. Windows seems to address that better than Linux (but still not very well). Maybe there is some great tool in Linux and I just don't know what it is (so a typical new user wouldn't either).
I don't think you're going to accurately guess what files your users will want to open and which applications to preinstall to cover them. Instead you should provide obvious access, in some form, to a big list of extensions and the suggested application(s) for opening each. I think that should be another tab in Synaptic.

3) Some beginners (me for example) immediately start trying to do things that are too hard for their level of Linux expertise. Usually that gets you very quickly to an error message naming a file that isn't on your computer and needs to be. What package is the file in? That was a frequent horrible problem until I found the two ways to get that answer. Why doesn't anyone tell newbies how to get that? (Ubuntu has a web page that will answer that and there is a command line utility you can install on your own system). Why make that a separate operation? If you use Synaptic to install packages, you should be able to use Synaptic to answer the question "which uninstalled package contains this file?". Synaptic works as a front end for other "apt" operations. Why can't it also front end that "apt" operation?

One way I plan to make it more obvious to end-users what programs they want to install, which best fit their needs, is by using Click N' Run, also called CNR. It makes it much easier to download and find programs than in Synaptic.

Regarding customer service, I plan to have a website that has a help section which displays an email account that helps fixes customer problems, a forums page, a wiki, and a Linux hardware compatibility list (from: http://www.linuxquestions.org/hcl/index.php).

Hopefully that is enough customer support (for the beginning years of my OS).

I have used CNR and found it lacking compared to every other method except Slackware which has nothing by default.
Actually CNR is okay for updates. It sucks for adding software. IMO

I had never heard of CNR before and I wanted to install a couple things into my Mepis system, so I tried to find them in CNR. The category structure in CNR is much clearer than in Synaptic, but the packages I wanted weren't in CNR. So I got them via Synaptic as usual. I didn't try installing anything else via CNR. I was curious but nervous (I don't know how things installed by CNR interact with Synaptic for update or removal, etc.).

I don't think Debian based distributions really need a better way to install software. They need a better way to figure out what software you want to install. CNR apparently tries to address both, and their focus on install may be diluting the potential for improvement in figure out what to install.