On Wed, Dec 22, 1999 at 09:41:37AM +0000, Paolo Mantegazza wrote:
> Hi David (and all of you),
>
> > I think I was the one that volunteered. In order to get everyone annoyed
> > at me as quickly as possible, I decided to release a version that is just a
> > collection of RTAI, RTL v1, and RTL v2. We can proceed from there, eh? Since
> > I can't find a copy of RTAI 0.9, I'm using 0.8 instead.
>
> RTAI 0.9 will be out by to-day, to-morrow at the latest.
>
> Definitly it is David that volunteered and I decided to take the chance,
> at least I will learn something new about makes and Makefiles. I've
> nothing against a pagkage containing different things. I'd just like
> that it should be possible for me to keep developping as I do now, i.e.
> fully outside the main Linux tree, in whatever place I like and without
> being burdened by other packages. Naturaly while keeping all David's
> Makefile so that he will not troubled by new RTAI releases.
I agree completely with the majority of development staying outside the
kernel tree. As we (collectively) learn more about how things should be
orgainized, perhaps it would be appropriate for the kernel patch to get
larger, but the goal is to make the patch to be as stable as possible.
My feeling is that the kernel patch is stuff that we think should go
to Linus and be included in the standard kernel.
The immediate goals, I think, should be:
- header files that will compile a module for either RTAI or RTLinux
simply by changing a define.
- separation of "real-time kernel" functions and the external interface,
and subsequent definition of the external interface. Currently, these
are mixed together in rtai.h.
- make the Makefiles sane.
- regression tests
- separate architecture dependent code
I've already started on some of these. I'd _really_ like someone to take
a shot at a regression test program -- it doesn't have to be complicated
at first, just something that tests a few things out, and doesn't crash
the computer (ideally).
After that, I want to see three levels of code -- the RTAI core (and RTLinux
core); drivers and modules that are independent of RTAI or RTLinux, either
in binary or source; and examples/applications/testing. At that point,
code from the RTAI/RTLinux core should gradually be moved to the independent
layer, until the only thing in the core is architecture-dependent code.
If they use the same interface, people should notice/care if the PowerPC
uses an RTLinux model and i386 uses an RTAI model. But basically, it's
to allow people with good ideas to write their own core, while still being
able to use the uppper-level code. Presumably, some day, someone will
write a core that Linus likes, and it will be included in the kernel.
dave...
--- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
----
For more information on Real-Time Linux see:
http://www.rtlinux.org/~rtlinux/
