On 11/6/07, Detlef Schmicker <d.schmicker at physik.de> wrote:
> sparc:
> http://buildd.debian.org/fetch.cgi?&pkg=maxima&ver=5.13.0-2&arch=sparc&stamp=1193345769&file=log
The failed tests are tests of the random number generator.
First the rng state is assigned, then many numbers are generated.
It appears that at first the expected numbers are generated, but at
some point the generated sequence diverges from the expected,
and after that most or maybe all further rng tests fail.
If someone wants to compare the actual output of Problem 5
(1000 numbers) to the expected, I'll generate the expected numbers
and send them.
> arm:
> http://buildd.debian.org/fetch.cgi?&pkg=maxima&ver=5.13.0-2&arch=arm&stamp=1193185352&file=log
>
> with a gcl compilation failuer:
> ; - Loading binary file "binary-gcl/nregex.o"
> Loading binary-gcl/nregex.o
> Error in FUNCALL [or a callee]: Unknown bfd format
I don't know what this is about. I found one other mention of this
exact error message "Unknown bfd format" in the Debian bug tracker.
The author speculates it is a toolchain bug.
http://groups.google.com/group/linux.debian.bugs.rc/msg/07f9742484c8f803
For both the sparc and arm errors, it might be possible to work around
them by using some Lisp other than GCL, e.g. Clisp. Anecdotal
evidence suggests GCL is not the most reliable Lisp.
> As I said, both were compiled with the 5.12 version, so maybe you
> can find the problem.
I'd like to see these problem be resolved, but unfortunately, it's unlikely
to happen unless the process is driven by someone who is running Debian.
best
Robert Dodier