Skip to content

MAINT Fixes test_fit_csr_matrix on py35 conda#15158

Merged
rth merged 2 commits intoscikit-learn:masterfrom
thomasjpfan:test_csr_matrix_fix
Oct 20, 2019
Merged

MAINT Fixes test_fit_csr_matrix on py35 conda#15158
rth merged 2 commits intoscikit-learn:masterfrom
thomasjpfan:test_csr_matrix_fix

Conversation

@thomasjpfan
Copy link
Copy Markdown
Member

Reference Issues/PRs

Fixes #14168

What does this implement/fix? Explain your changes.

Makes everything deterministic in test_fit_csr_matrix.

@thomasjpfan
Copy link
Copy Markdown
Member Author

@NicolasHug Most likely an easy review?

Copy link
Copy Markdown
Member

@rth rth left a comment

Choose a reason for hiding this comment

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

This certainly won't hurt, but that still means that the conditions we test for is too strict and won't pass for any random state. Though given that it repeatably fails only on py35 job (and passes elsewhere), I wonder if it's linked to numpy versions.

Copy link
Copy Markdown
Member

@NicolasHug NicolasHug left a comment

Choose a reason for hiding this comment

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

LGTM, but I agree with @rth's concern.

How stable is the test w.r.t the random seed?

@rth
Copy link
Copy Markdown
Member

rth commented Oct 20, 2019

I this there is no risk with merging it, as it's an improvement in any case. As to whether it will fix the corresponding issues, we'll see. If CI fails again we can always reopen (but I hope we will also drop py36 soon).

@rth rth merged commit 5a0043a into scikit-learn:master Oct 20, 2019
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

test_fit_csr_matrix failing on Linux py35_conda_openblas

3 participants