Raymond Toy wrote:
> pip wrote:
> >>> I could not load(lapack) within maxima-5.18.0.
> >>> The last message I have is
> >>> ..................
> >>> ; compiling (IN-PACKAGE #:CL-USER)
> >>>
> >>> ; /home/va/.maxima/binary/binary-sbcl/share/lapack/binary-
> >>> sbcl/lapack/dlaed6.fasl written
> >>> ; compilation finished in 0:00:00.027
> >>> ;; loading #P"/home/va/.maxima/binary/binary-sbcl/share/lapack/binary-
> >>> sbcl/lapack/dlaed6.fasl"
> >
> > Error is exactly here, in this file dlaed6.lisp. Unfortunately the sbcl
> > debugger is not very informative to me. However, when I replaced this
> > file with old version from maxima 5.17.1, the compilation goes further up
> > to
>
> Aargh. There's a bug introduced between those versions. The variable
> first$ is declared to be a single-float, but then it assigns T to it.
>
> On a side note, the compilation messages are much better if you use
>
> :lisp (mk:oos "lapack" :compile). That's not something you're expected
>
> to know, though.
Your guess is correct. After :lisp (mk:oos "lapack" :compile);
I've got
.....
; loading #P"/home/va/.maxima/binary/binary-sbcl/share/lapack/binary-
sbcl/lapack/dlaed6.fasl"
Maxima encountered a Lisp error:
The value T is not of type SINGLE-FLOAT.
Automatically continuing.
To reenable the Lisp debugger set *debugger-hook* to nil.
;
; compilation unit aborted
..............
>
> > dgesvd where I met another problem with "heap exhausted during garbage
> > collection" (This is on the linux mashine with dual core celeron 1.86 and
> > 2Gb RAM)
>
> Do you know if the previous version of dgesvd worked better?
Right now I think this issue come from the altlinux kernel specific. I will
try to investigate situation further. The problem goes away if I work under 64
bit kernel.
best regards
Valery