PyFMI 2.13.0 - problem with conda-forge installation #238
Comments
|
@jschueller Do you have an idea what could be the issue here? From my understanding of the error message (from similar reported issue with numpy), this is likely due to a mismatch of the (Assimulo?) version build used to build PyFMI, vs the one actually installed. |
|
pyfmi just needed a rebuild, should be ok now |
|
Good! I can confirm that a few test cases I run on Google Colab
Works just fine using PyFMI 2.13-0 with Assimulo 3.5.1.
Jan Peter
PS Test cases you have here https://github.com/janpeter19
… On 23 May 2024, at 12:49, Julien Schueller ***@***.***> wrote:
pyfmi just needed a rebuild, should be ok now
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35L2G3YQGGBIF3X2JMODZDXCSXAVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMRWHAYDGOJYGQ>.
You are receiving this because you authored the thread.
|
|
By the way, is Python 3.10.12 a good environment for PyFMI 2.13-0
or time to step up to Python 3.11 or even 3.12?
FMPy 0.3.20 I run in Python 3.12.3 (at least in Windows).
… On 23 May 2024, at 12:49, Julien Schueller ***@***.***> wrote:
pyfmi just needed a rebuild, should be ok now
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35L2G3YQGGBIF3X2JMODZDXCSXAVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMRWHAYDGOJYGQ>.
You are receiving this because you authored the thread.
|
|
[3.8-3.11] is ok but 3.12 is not supported yet because of assimulo fortran codes |
|
Thanks. / Jan Peter
… On 23 May 2024, at 14:00, Julien Schueller ***@***.***> wrote:
[3.8-3.11] is ok but 3.12 is not supported yet because of assimulo fortran codes
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35LZGEHRZ72Q6IEGOXVTZDXK4XAVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMRWHEZTGMBYG4>.
You are receiving this because you authored the thread.
|
|
@jschueller Thanks for fixing this. Am I missing something with the Assimulo fortran codes? The codes should be fine to compile when using std:legacy, or is there some other issues you came across while working on replacing setuptools? /Peter |
|
same issue: the fortran codes prevent us from using setuptools and force us to find a replacement for distutils |
|
We do intend to keep the Fortran solvers, can you elaborate a bit on the issues with Fortran? (Maybe more suitable in the Assimulo repo in some issue or MR) |
|
see modelon-community/Assimulo#98 |
|
Good to hear. The earlier fix helped well with Python 3.10.x. Do you suggest to try installation with cones-forge and Python 3.12 and do I need to bother about handling disabled assimulo Fortran solvers? Jan PeterSkickat från min iPhone8 juni 2024 kl. 13:54 skrev Julien Schueller ***@***.***>:
see modelon-community/Assimulo#98
we have pyfmi with python 3.12 now (with assimulo fortran solvers disabled though)
I believe this can be closed now
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>
|
|
I think you can pick any version if you dont need the solvers explicitly |
|
I use CVODE by default. Will it work?Linux FMU ME?Skickat från min iPhone8 juni 2024 kl. 14:18 skrev Julien Schueller ***@***.***>:
I think you can pick any version if you dont need the solvers explicitly
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>
|
|
yes, this is fine |
|
Hi,
I made a test using Google Colab which is my usual “demo infrastructure” and upgrading to Python 3.12 did not work at all.
If you like you can try it out yourself.
https://github.com/janpeter19/BPL_TEST2_Perfusion/blob/main/xBPL_TEST2_Perfusion_colab.ipynb <https://github.com/janpeter19/BPL_TEST2_Perfusion/blob/main/xBPL_TEST2_Perfusion_colab.ipynb>
Note, that the second simulation in this notebook gives not the right results after 30 hours, but
This is due to the FMU ME compilation from OpenModelica and the key people there are informed.
For the time being I stay with Python 3.10 and let your competitor FMPy work with Python 3.12.
Would be nice to have them both on Python 3.12, but not a big deal.
I have at least one example that PyFMI handles better than FMPy.
Jan Peter
… On 8 Jun 2024, at 16:01, Julien Schueller ***@***.***> wrote:
yes, this is fine
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35L55NRMBJHIBFS3HQNTZGMFB7AVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNJWGA2DONRYHA>.
You are receiving this because you authored the thread.
|
|
Hi @janpeter19 has this issue been resolved? |
|
Hi
No the issue is not resolved.
However, the link to the notebook was a bit confusing, since when your started it,
a slightly older version with Python 3.10 was loaded and works.
Now the example in the link is fixed so your really get Python 3.12 and then
you see that pyfmi that is loaded has severe problems.
Hope you can get it all working with Python 3.12 since “nice” and same as FMPy standard,
but not at all critical for me, and my users.
Jan Peter
… On 12 Jun 2024, at 14:02, Robin Andersson ***@***.***> wrote:
Hi @janpeter19 <https://github.com/janpeter19> has this issue been resolved?
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35L7YCAWYYVJ66S2EZULZHA2HHAVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRSHAZTKNBYGQ>.
You are receiving this because you were mentioned.
|
|
Hi @janpeter19, now I understand, thanks for the clarification! |
Closed
|
You script is incorrect: your are mixing packages from defaults channel, I suggest to use miniforge instead of miniconda |
|
A couple of questions
1) Is here not a way to use Miniconda so that “mixing packages from default channel” is avoided?
(After all Miniconda is well spread and accepted and if PyFMI gets dependent of the (smaller?) alternative miniforge
that may question the use of PyFMI in favour of FMPy…)
2) I looked at Github Miniforge and want to at least try it:
a) Installers is with Python 3.10 - does not that prevent usage of Python 3.12 that was the goal?
b) How to bring in Miniforge in an automated way to Colab notebook? My current commands are:
!wget https://repo.anaconda.com/miniconda/Miniconda3-py310_23.1.0-1-Linux-x86_64.sh <https://repo.anaconda.com/miniconda/Miniconda3-py310_23.1.0-1-Linux-x86_64.sh>
!chmod +x Miniconda3-py310_23.1.0-1-Linux-x86_64.sh
!bash ./Miniconda3-py310_23.1.0-1-Linux-x86_64.sh -b -f -p /usr/local
import sys
sys.path.append('/usr/local/lib/python3.10/site-packages/')
The wget line I can update to use of Miniforge, but after that hard to understand for me how to do.
Jan Peter
… On 12 Jun 2024, at 17:40, Julien Schueller ***@***.***> wrote:
You script is incorrect: your are mixing packages from defaults channel, I suggest to use miniforge instead of miniconda
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35L2WNIY4NSVTUOQTJNLZHBTW7AVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRTGM2TGOBRGA>.
You are receiving this because you were mentioned.
|
|
Good. A brief look show a certain focus on MacOS. I guess it can directly be applied to Ubuntu 22.04 right?
(Well, I am a Mac user but run Modelica stuff on virtual machines for Windows or Linux and my Github demos
use Google Colab which currently have Ubuntu 22.04) / Jan Peter
… On 13 Jun 2024, at 11:01, Julien Schueller ***@***.***> wrote:
see https://conda-forge.org/docs/user/tipsandtricks/ <https://conda-forge.org/docs/user/tipsandtricks/>
—
Reply to this email directly, view it on GitHub <#238 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIP35L2O6QYLY5ZIZZPE34TZHFNVXAVCNFSM6AAAAABIEH3AD6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRVGA2TMNBRGY>.
You are receiving this because you were mentioned.
|
Closed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi,
I install PyFMI 2.13.0 using conda from conda-forge into Ubuntu 22.04 LTS and it does not work.
(I use Python 3.10.12 and Numpy 1.26.4 and Assimulo 3.5.1).
The problem is that when the FMU is loaded I get an error message about Assimulo as follows:
ValueError: assimulo.problem.cProblem size changed, may indicate binary incompatibility. Expected 32 from C header, got 24 from PyObject
It seems also not possible to now use the earlier PyFMI version 2.12.0.
The text was updated successfully, but these errors were encountered: