Search Criteria
Package Details: python-slycot 0.5.0.0-1
Git Clone URL: | https://aur.archlinux.org/python-slycot.git (read-only, click to copy) |
---|---|
Package Base: | python-slycot |
Description: | Python wrapper for selected SLICOT routines, notably including solvers for Riccati, Lyapunov and Sylvester equations. |
Upstream URL: | http://github.com/python-control/Slycot |
Licenses: | GPL2 |
Conflicts: | python-slycot-git |
Provides: | python-slycot |
Submitter: | bnavigator |
Maintainer: | bnavigator |
Last Packager: | bnavigator |
Votes: | 1 |
Popularity: | 0.000000 |
First Submitted: | 2019-09-27 15:16 (UTC) |
Last Updated: | 2022-07-09 15:59 (UTC) |
Dependencies (12)
- python-numpy (python-numpy1.22AUR, python-numpy-flameAUR, python-numpy-mkl-binAUR, python-numpy-openblasAUR, python-numpy-mklAUR, python-numpy-gitAUR)
- blas (armplAUR, blis-gitAUR, blisAUR, blis-cblas-openmpAUR, blis-cblasAUR, aocl-blis-aoccAUR, aocl-blisAUR, blas-aocl-gccAUR, blas-aocl-aoccAUR, blas-mklAUR, blas-gitAUR, atlas-lapackAUR, openblas-lapackAUR, blas-openblas) (make)
- cmake (cmake-gitAUR) (make)
- gcc-fortran (gccrs-fortran-gitAUR, gcc-fortran-gitAUR, gcc11-fortranAUR) (make)
- lapack (armplAUR, aocl-libflame-aoccAUR, aocl-libflameAUR, blas-aocl-gccAUR, blas-aocl-aoccAUR, blas-mklAUR, lapack-gitAUR, atlas-lapackAUR, openblas-lapackAUR, blas-openblas) (make)
- python-build (make)
- python-installer (python-installer-gitAUR) (make)
- python-numpy (python-numpy1.22AUR, python-numpy-flameAUR, python-numpy-mkl-binAUR, python-numpy-openblasAUR, python-numpy-mklAUR, python-numpy-gitAUR) (make)
- python-scikit-build (make)
- python-setuptools (make)
- python-pytest (check)
- python-scipy (python-scipy-mkl-binAUR, python-scipy-mklAUR, python-scipy-gitAUR) (check)
Required by (3)
- python-control (check)
- python-control (optional)
- python-pymor (optional)
Latest Comments
1 2 Next › Last »
Jona commented on 2023-09-16 11:39 (UTC)
I can confirm that the changes done in PKGBUILD are working ok on aarch64 (RaspberryPi 400, but for any other board the result should be the same), using makepkg.
Jona commented on 2023-09-14 19:12 (UTC) (edited on 2023-09-14 19:37 (UTC) by Jona)
For those struggling with this package not building with current numpy, if the maintainer has some spare time, he can update the PKGBUILD to get this working:
bnavigator commented on 2022-07-09 16:22 (UTC)
scikit-build does not work with numpy 1.23.0.
You have to have numpy 1.22 installed during the build of python-slycot. Upgrading to numpy 1.23.0 is possible after installation of python-slycot.
carlosal1015 commented on 2021-10-15 19:35 (UTC)
Thanks @bnavigator. I post the issue here.
bnavigator commented on 2021-10-15 18:32 (UTC)
@carlosal1015
Looks like a new or different version of BLAS/LAPACK returns a differently scaled value for a eigensolution. Not necessarily wrong. Just not expected by the Slycot tests. Could you please open an issue upstream at https://github.com/python-control/Slycot/issues so that the developers (which I am a member of) can look at it?
carlosal1015 commented on 2021-10-15 17:51 (UTC)
Hi, I have the following message error from the last pkgrel changed.
bnavigator commented on 2020-10-02 14:16 (UTC) (edited on 2020-10-02 15:03 (UTC) by bnavigator)
I am not saying we do not support MKL upstream. In fact the conda packages used to build against MKL. I do not know the current status, it looks like they reverted that. Upstream still has the recipe in the repository with
python setup.py build_ext -lmkl -DBLA_VENDOR=Intel10_64lp
Removing the check is not an option, it makes sure that the built package actually works. Did you try to use the installed slycot yet? Chances are, that the link to your installed MKL lib is broken there too.
To disable the check use
makepkg --nocheck
(oryay --mflags --nocheck
). To force a specific LAPACK implementation, set BLA_VENDOR.acxz commented on 2020-10-02 13:14 (UTC) (edited on 2020-10-02 13:22 (UTC) by acxz)
According to upstream: https://github.com/python-control/Slycot#dependencies last bullet
They mention that MKL is a valid BLAS/LAPACK package to use.
Although installing MKL and rebuilding
python-slycot
finishes successfully. Thx for you help in debugging this. I post an issue upstream. (https://github.com/python-control/Slycot/issues/138)In the meanwhile, would it be possible to comment out the
check
function to prevent this issue from appearing for other users. Right now when installing/updating python-slycot I will need to uninstall intel-mkl and reinstall it after python-slycot has been installed.bnavigator commented on 2020-10-02 13:01 (UTC) (edited on 2020-10-02 13:05 (UTC) by bnavigator)
Yes those versions are the same (ignore the plugins, the are not used here). More interesting would be the numpy and blas/lapack versions.
Edit: Yep you are trying to use a custom lapack/blas implementation: Intel MKL. Archlinux cannot help here. Maybe I can help you upstream when you open an issue on GitHub and provide as much details about your installation as possible, but be aware that it might be just your custom setup which is not supported.
acxz commented on 2020-10-02 12:45 (UTC) (edited on 2020-10-02 12:56 (UTC) by acxz)
Hello @bnavigator, thx for maintaining this package! I am having some trouble when installing this package, in particular the
check()
is failing for me. Commenting out thecheck()
functions allows me to complete the successful install of the package.Here is the error with the check function:
Honestly doesn't seem very informative. Maybe I should also post upstream? Can you confirm that the versions under
test session starts
are the same for you? ThanksEdit: Looking a bit deeper (i.e. running the test manually by adding a main and calling the test function directly in the file) I see the follow error:
1 2 Next › Last »