Dear Developers,
here is the result of the test suite in Octave 6.0.90, on the system given below (compiled without GUI):
Summary:
PASS 15891
FAIL 7
XFAIL (reported bug) 26
SKIP (missing feature) 197
SKIP (run-time condition) 14
Failed:
plot/appearance/camlookat.m .................................... pass 9/10
FAIL 1
sparse/ichol.m ................................................. pass 20/24
FAIL 4
sparse/pcg.m ................................................... pass 15/16
FAIL 1
specfun/cosint.m ............................................... pass 28/29
FAIL 1
System (Intel Core i3-7130U):
* Ubuntu 20.04 (Linux Kernel 5.4)
* Custom (non 20.04) libs:
+ mimalloc 1.6.4
+ openblas 0.3.10
+ flexiblas 3.0.2
+ qrupdate-ng 1.1.3
+ SuiteSparse 5.8.1
+ arpack-ng 3.7.0
+ sundials 5.3.0
I remember that "camlookat" failed previously (as in 5.1/5.2) for me, too.
Best
Christian
PS: Some preliminary testing shows ode23s is very slow compared to MATLAB 2020a, once I know more I will send some info.
PPS: I may have found a scoping bug which could result from the nested function functionality: A variable named "t" declared in a function is not visible in a three level nested anonymous function later in that function, where the second level anonymous function has a "t" as a parameter. I will report a bug once I can reproduce it in simpler code.
----- On Aug 27, 2020, at 3:38 PM, John W. Eaton jwe@octave.org wrote:
> The following files are now available for testing from alpha.gnu.org in
> the directory gnu/octave
>
> -rw-r--r-- 32508461 Aug 26 19:54 octave-6.0.90.tar.gz
> -rw-r--r-- 21424489 Aug 26 19:54 octave-6.0.90.tar.lz
> -rw-r--r-- 25050816 Aug 26 19:55 octave-6.0.90.tar.xz
>
> -rw-r--r-- 352766877 Aug 27 13:21 octave-6.0.90-w64-installer.exe
> -rw-r--r-- 341029732 Aug 27 13:25 octave-6.0.90-w64.7z
> -rw-r--r-- 586445094 Aug 27 13:31 octave-6.0.90-w64.zip
>
> Please report any new issues using the release tag 6.0.90 in the bug
> tracker.
>
> jwe
--
Dr. rer. nat. Christian Himpe
Computational Methods in Systems and Control Theory
Max Planck Institute for Dynamics of Complex Technical Systems
Sandtorstr. 1
39106 Magdeburg
Germany
https://himpe.science
Your openblas library is probably broken. Try the one with the omp interface or just the reference blas.
Dmitri.
--