Status update for AltimatOS - April 29, 2012

Short and quick update:

Most of the circular dependencies for the new build root are done, only a couple remain before I can start rebuilding various low level libraries to add support for LDAP and Kerberos. Happily, GCC 4.7.0 seems far less broken than the 4.5.1 and before releases (read that as less patching to core libraries to make build....) Soon we'll start reviewing the init options, and such for "suitability".

Progress on AltimatOS

Some know I've been working off and on in my time on a new Linux distribution, and today I decided to start writing up the process to allow better feedback as I work on the software and packages involved, and better show what goes into building a new flavour of distribution. To start, I guess I should cover what my goals with this are:

1. Home and Business User-centric: This means that I'll be forgoing a number of UNIX-isms in favour of user needs and wants. While the power of the GNU project's tools, KDE, and other free/open source tools are the building blocks of AltimatOS, the intent is to allow the average accountant, secretary, high school student, or end-user to easily use their computer without getting in their way, or exposing them to archaic command-line prompts to do normal tasks.
2. Desktop oriented: AltimatOS is not, and never will be a server platform. While the underside of the operating system is based on similar software as other GNU/Linux distributions, AltimatOS' design and filesystem layout are not targeted at the duties of acting as a high performance server.
3. Not another "Linux": AltimatOS is not another GNU/Linux distribution in that the normal GNU/Linux distributions mostly adhere to filesystem layouts and design philosophies that run counter to good end-user desktop and corporate desktop operating system design. AltimatOS aims to avoid many of these design pitfalls that have kept "Linux on the Desktop" from becoming a reality.

With these focuses in mind, here are the immediate goals:

1. The entire operating system must hide away the command line as much as is possible - This means that by default (unless in debug mode) the system should not expose the command-line heritage of GNU/Linux to the end-user. (Turn off the virtual consoles, disallow switching from X11 to the other consoles, etc.)
2. No normal-case configuration settings should require the command line - Printing, file-sharing, hardware configuration, security configuration, and network settings must be done graphically and should not require jumping to the command line.
3. Job scheduling (cron and others) must be graphically configured - self explanatory.
4. Filesystem layout must be simple to navigate - While the FHS and LSB standards are intended on "standardizing" GNU/Linux, they've been largely designed by UNIX engineers, which drive them from being useful to the typical end user. To this end I'm taking a page out of Apple's play book; AltimatOS will radically restructure the filesystem hierarchy to simplify the user experience. While the traditional filesystem hierarchy will be preserved with symlinks for some level of compatibility, software explicitly designed for AltimatOS should not rely on these compatibility symlinks.
5. The compatibility symlinks will be hidden from end-user view. The method by which we do so will be through patches to KDE's KIO and Dolphin to honour our distribution specific named extended file attribute: user.visibility.
6. Authentication must be done through the local LDAP and Kerberos services running on the system. The flat file method of authentication inspired from UNIX, while simple is dated and inflexible to extend.
7. In a central authentication environment, we must support LDAP+Kerberos - Simplifying enterprise integration with standard OpenLDAP and MIT Kerberos on a GNU/Linux server.
8. System configuration is tightly coupled to the Puppet configuration engine - Software packages and configuration tools must work through the custom Puppet registration on the local system, or a centrally manage puppet-master.

Today, we've our cross tools tree built, and are currently working on building the initial run-time. This will allow us to completely deviate from the SuSE installation that is acting as the cross build host inside the VM that is running the builds.

The rationale for using VMware for this is simple, While working on this, I can still have a usable system to use for my day to day until AltimatOS is to a reasonable state to use as my main desktop.

As the weeks roll on, I'll chronicle the work being done and at times may ask questions of my readers to see if you've ideas to make this even better. Thanks for joining me on this ride, comments always welcome.