EL/IX: Unifying APIs for Linux and Post-PC Computing

Where will EL/IX go from here?

EL/IX is based on the internationally accepted POSIX standard that is already fundamentally supported in Linux through glibc. EL/IX will grow as components of that POSIX interface are implemented in such a way that they can be easily configured in or out of a given embedded Linux implementation. EL/IX will also grow as other components of Linux are enhanced to support the configurability and scalability that EL/IX enables, and it will also cover a growing number of development platforms.

EL/IX will need to be tested, both in implementation and in concept. The short-term test will be the ability to build simple applications, such as a web server with dynamic page generation, that can use EL/IX to run on desktop and embedded Linux as well as eCos. The medium-term test will be the ability to build complex applications, such as GNOME and some of its utilities, that can use EL/IX to run on an embedded application such as a point-of-sale kiosk. The long-term test will be the ability to use a standard set of configuration tools to unpack any Linux application and port it without modifications to any system that supports the EL/IX APIs that it uses.

EL/IX will also need to take some sort of position on rationalizing the I/O problem. It is too soon to present a solution to this problem, but the scope is at least clear: for each kind of I/O device that we want to support, from an X-terminal to an LED, and from a PCI bus to a serial port, the use of the interfaces should be consistent not just for Linux and embedded Linux, but for deeply embedded systems as well. This means being able to configure I/O devices without necessarily configuring in a whole mess of functionality that is really only needed to support an abstract I/O interface. Similarly, in keeping with the philosophy of a unified development platform, the choices made here should also extend not only to real hardware devices, but to virtual hardware components as well, and the mechanisms to support these devices used should integrate not only with Linux but again also with deeply embedded operating systems as well. The eCos development team has some experience in both these areas which is likely to be valuable in defining an I/O annex for EL/IX.

It is important to be clear that EL/IX is not a Linux distribution. EL/IX is an API that is operating system neutral. We believe that due to the open source nature of both Linux and EL/IX, that EL/IX should be available in a current version of desktop Linux, but not all implementations will be supported by all embedded operating systems (Linux or otherwise). The value of EL/IX is that developers can create applications on the desktop and then use EL/IX to determine where and/or how to migrate them to which embedded platforms. Thus, even while hardware becomes more diverse, it will always be easy to know what software can be ported to what hardware with a simple recompile.

It is also important to be clear that EL/IX is not going to make Linux suitable for deeply embedded targets, nor will it enable very lightweight operating systems to replace Linux as a desktop operating system. In other words, EL/IX will not give operating systems properties or functionality they don't already have, but will instead give developers the ability to target a broader range of application platforms than they can in today's highly fragmented embedded systems market.

Who's behind EL/IX?
Cygnus Solutions, now Red Hat, is creating the first implementation of EL/IX, but we are not the only ones behind it. Developers from the Free Software Foundation and the GNOME project and industry analysts agree that EL/IX provides the necessary continuity between desktop and embedded Linux to ensure that the Linux market does not fragment as it connects with the Post PC revolution.

Cygnus, now Red Hat, is uniquely qualified to maintain EL/IX. Cygnus engineers were behind the original GNU configure script, the first script used to configure all GNU software in a consistent fashion. Cygnus engineers also help to create, and currently maintain autoconf, a next-generation generator of configure scripts. Cygnus pioneered the concept of source-level configurabilty for real-time kernels with eCos, implementing a graphical user interface that made it easy for users to use the power of this open source operating system without needing to fully understand the source-level details. And Cygnus employs a majority of the developers that maintain the GNU software development tools, tools that are fundamental to both Linux and embedded systems development alike.

When will EL/IX be available?
As with any new Open Source project, the answer is "Good Question!". We will post updated API drafts, schedule information, and other information to www.redhat.com/embedded/technologies/elix/. By visiting this site and registering your interest (in using and/or contributing to EL/IX), we can keep you up to date on the latest developments.