Re: orocos core documentation

On Wednesday 08 October 2003 00:55, you wrote:
> Peter Soetens wrote:
> >LXRT is explained on my homepage
> > http://people.mech.kuleuven.ac.be/~psoetens/portingtolxrt.html
>
> Having looked at "/Introducing the Real Time Application Interface
> (RTAI) for Linux/" paper (Mantegazza, Bianchi, Dozio), I'm a bit
> confused with their suggestions of use with LXRT and what I interpreted
> yours to be. Correct me if I'm wrong, but are you working on creating an
> 'end product' which will attempt to be 'hard real-time' operating in
> 'user space'?
> The above-mentioned paper describes this as 'firm real-time', and a
> /method/ in which to develop code...
>
> /"when the developer is satisfied with a task's performance within LXRT,
> the task is simply recompiled as a module and inserted into the kernel
> (along with the associated modules which provide RTAI's real-time
> features) to transition _from firm to hard real-time_."/
>
>  From this, my understanding is that during development LXRT can be used
> to develop and debug, but ultimately the final product should be
> compiled into a kernel module.

This is only true if your application is not using _any_ userspace library, 
not even libc ...

>
> Has the /RTAI-wind/ made this document redundant, or is your suggested
> use of LXRT just another approach?

As you guessed, that paper is a bit outdated. The general consensus is now 
that most realtime applications can and will run under LXRT, and never go to 
kernel space. This is simply explained by the fact that no complex 
application can run in the kernel because of the absence of libraries. Look 
at what is happening with the original Linux itself : first someone writes a 
new piece of kernel functionality (like a http server, a WLAN Access 
Point, ....), because of the limited possibilities, the application moves to 
userspace only leaving a small kernel module behind to communictate with the 
hardware or serialize user requests. This is exactly what is happening with 
RTAI on a whole. Only RT-device drivers will reside in the kernel, the 
RT-programs will run under LXRT. It incurs some overhead, but opens losts of 
possibilities.

Peter

-- 
------------------------------------------------------------------------
Peter Soetens                                      http://www.orocos.org
Katholieke Universiteit Leuven
Division Production Engineering,                      tel. +32 16 322773
Machine Design and Automation                         fax. +32 16 322987
Celestijnenlaan 300B                   [email protected]
B-3001 Leuven Belgium                 http://www.mech.kuleuven.ac.be/pma
------------------------------------------------------------------------