"RTEMS parrot runs, now what ?" (3 lines) at http://nopaste.snit.ch/18530

Jonathan Leto jaleto at gmail.com
Wed Nov 4 04:14:11 UTC 2009


I am preparing a blog post about all this. Coming Soon.

But to summarize, RTEMS still needs some patches to our
configure/build system to make cross-compiling work properly. But
Chris has succeeded in porting Parrot to RTEMS with a custom Makefile
and some patches to Parrot, with a little help from the #parrot crew.

> Great job guys!  How much of Parrot does this prove works?

The benchmarks exercise a fair bit of the basic internals. So far I
have seen the Fibonacci benchmark.
If the other benchmarks all run without segfaulting, I would say that
most of Parrot is working. Of course
there are many other subsystems that are not being exercised, but
those are optional/additional.

For example, RTEMS has no random number generation support right now,
because RNG's are dynamically loadable ops and RTEMS does not support
that. But Parrot could just as easily be compiled with all of the
dynops statically linked in, so it is easily circumvented.

I would say most issues left for Parrot on RTEMS are of that nature:
small issues that are easily fixed/circumvented.

> What else needs to be done before the champagne is popped? :)

I think we can announce that we have a port to RTEMS mostly working,
but which needs to be polished
up for public consumption. We need to see what type of
images/binaries/packages are possible.

Blog post forthcoming.

Thanks to Chris again for all the amazing work!



Jonathan Leto
jonathan at leto.net

More information about the parrot-dev mailing list