Skip to content

Conversation

ricardoV94
Copy link
Member

@ricardoV94 ricardoV94 commented Sep 23, 2025

Some major import changes in the latest release


📚 Documentation preview 📚: https://pymc--7910.org.readthedocs.build/en/7910/

@ricardoV94 ricardoV94 added release pytensor major Include in major changes release notes section labels Sep 23, 2025
@ricardoV94 ricardoV94 changed the title Update pytensor dependency Update PyTensor dependency Oct 9, 2025
@ricardoV94 ricardoV94 marked this pull request as ready for review October 9, 2025 09:59
@ricardoV94
Copy link
Member Author

ugh loads of broken stuff

@ricardoV94
Copy link
Member Author

Failures had to do with functions that guessed the inputs for the user. The order changed since the last release and now gradients / forward were not aligned.

This was not a great idea to begin with, so I now raise anytime such method is called in a way that the input order is ambiguous.

@ricardoV94 ricardoV94 force-pushed the bump_pytensor_dep branch 3 times, most recently from 4caaaca to 3ddcecc Compare October 13, 2025 18:16
@ricardoV94
Copy link
Member Author

CI broken by pymc-devs/pytensor#1660 which means we'll need another release (which is also a major release as it drops old numpy / python) :)

@ricardoV94
Copy link
Member Author

getting close. @maresb @lucianopaz any luck with the windows blas issue on conda?

@ricardoV94 ricardoV94 force-pushed the bump_pytensor_dep branch 3 times, most recently from 04d3c35 to 6f626fa Compare October 14, 2025 12:55
@ricardoV94 ricardoV94 changed the title Update PyTensor dependency Update PyTensor dependency (any functions that inferred inputs of the graph fail if there is more than one input, including Model.compile_fn) Oct 15, 2025
@ricardoV94
Copy link
Member Author

The verbose PR title is just to remind to add a note in the release notes

@ricardoV94 ricardoV94 changed the title Update PyTensor dependency (any functions that inferred inputs of the graph fail if there is more than one input, including Model.compile_fn) Update PyTensor dependency (drop numpy <2.0, and Python 3.10. any functions that inferred inputs of the graph fail if there is more than one input, including Model.compile_fn) Oct 15, 2025
@ricardoV94 ricardoV94 force-pushed the bump_pytensor_dep branch 2 times, most recently from 45b4034 to 787179b Compare October 15, 2025 09:00
@ricardoV94
Copy link
Member Author

I'm not testing on Windows because the release is held at conda-forge. I'll open an issue to revert the relevant commit once conda-forge/pytensor-suite-feedstock#177 is fixed

ricardoV94 and others added 2 commits October 15, 2025 11:14
Drops support for Python 3.10 and Numpy<2.0

Co-authored-by: Jesse Grabowski <[email protected]>
@ricardoV94 ricardoV94 force-pushed the bump_pytensor_dep branch 2 times, most recently from be479e8 to 9d1f69f Compare October 15, 2025 09:17
@maresb
Copy link
Contributor

maresb commented Oct 15, 2025

Ya, sorry, I'm working on it, but not sure yet what the problem is. Will try to follow Luciano's advice

Copy link

codecov bot commented Oct 15, 2025

Codecov Report

❌ Patch coverage is 88.67925% with 6 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.47%. Comparing base (6993a2f) to head (70a0df8).

Files with missing lines Patch % Lines
pymc/pytensorf.py 54.54% 5 Missing ⚠️
pymc/model/core.py 91.66% 1 Missing ⚠️
Additional details and impacted files

Impacted file tree graph

@@            Coverage Diff             @@
##             main    #7910      +/-   ##
==========================================
- Coverage   91.63%   91.47%   -0.16%     
==========================================
  Files         116      116              
  Lines       18882    18904      +22     
==========================================
- Hits        17302    17292      -10     
- Misses       1580     1612      +32     
Files with missing lines Coverage Δ
pymc/distributions/custom.py 95.45% <100.00%> (ø)
pymc/distributions/shape_utils.py 91.87% <100.00%> (ø)
pymc/distributions/timeseries.py 94.62% <100.00%> (+0.01%) ⬆️
pymc/distributions/transforms.py 100.00% <100.00%> (ø)
pymc/initial_point.py 99.20% <100.00%> (+<0.01%) ⬆️
pymc/logprob/basic.py 100.00% <100.00%> (ø)
pymc/logprob/mixture.py 95.72% <100.00%> (+0.01%) ⬆️
pymc/logprob/rewriting.py 100.00% <100.00%> (ø)
pymc/logprob/tensor.py 94.26% <100.00%> (ø)
pymc/logprob/utils.py 90.07% <100.00%> (+0.07%) ⬆️
... and 8 more

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@maresb
Copy link
Contributor

maresb commented Oct 15, 2025

Ya, sorry, I'm working on it, but not sure yet what the problem is. Will try to follow Luciano's advice

Solved! 🎉

@ricardoV94
Copy link
Member Author

reverted the commit so we test on windows as usual

@ricardoV94
Copy link
Member Author

Uhuh tests are passing

Copy link
Member

@lucianopaz lucianopaz left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think the changes look OK, but maybe someone else should have a look too.

@ricardoV94 ricardoV94 merged commit f80e896 into pymc-devs:main Oct 16, 2025
26 checks passed
@ricardoV94 ricardoV94 deleted the bump_pytensor_dep branch October 16, 2025 19:47
@ricardoV94 ricardoV94 changed the title Update PyTensor dependency (drop numpy <2.0, and Python 3.10. any functions that inferred inputs of the graph fail if there is more than one input, including Model.compile_fn) Update PyTensor dependency (drop numpy <2.0, and Python 3.10.) Oct 16, 2025
@maresb
Copy link
Contributor

maresb commented Oct 17, 2025

Oops! We forgot to update python_requires:

pymc/setup.py

Line 73 in f80e896

python_requires=">=3.10",

@maresb maresb mentioned this pull request Oct 17, 2025
11 tasks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

major Include in major changes release notes section pytensor release

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants