Skip to content

Test on prereleases instead of nightly - #3525

Merged
bkamins merged 3 commits into
mainfrom
andreasnoack-patch-1
Nov 22, 2026
Merged

bkamins merged 3 commits into
mainfrom
andreasnoack-patch-1

Conversation

@andreasnoack

Copy link
Copy Markdown
Member

Testing on nightly is, in my opinion, mostly a nuisance to package developers. Currently main is failing on nightly.

Testing on nightly is, in my opinion, mostly a nuisance to package developers.
@andreasnoack
andreasnoack requested a review from bkamins February 21, 2026 14:59
@bkamins

bkamins commented Feb 21, 2026

Copy link
Copy Markdown
Member

The tests on nightly fail due to changes in random number generation changes in Julia. In general this option to test on nightly was intentional, but we can discuss changing it to pre. @nalimilan - what do you think?

@bkamins bkamins added this to the patch milestone Feb 21, 2026
@bkamins bkamins added CI ecosystem Issues in DataFrames.jl ecosystem labels Feb 21, 2026
@nalimilan

Copy link
Copy Markdown
Member

I don't really have an opinion. We have moved some other packages to pre as indeed errors on nightly are often annoying, but they can also be interesting to detect in advance...

@anurag-mds

Copy link
Copy Markdown
Contributor

I'm also facing the nightly CI failure in my PR #3526
due to the same issue. Happy to see this being addressed!

@andreasnoack

Copy link
Copy Markdown
Member Author

Bumping this one. A CI log like this is a bad visual and is not informative about the state of this package

Screenshot 2026-11-13 at 08 54 25

@bkamins

bkamins commented Nov 13, 2026

Copy link
Copy Markdown
Member

The part of the code that fails both on nightly and on pre is:

    df = DataFrame(a=[1, 2], b=[missing, 3])
    dfr = [df[1, :], df[2, :], df[1, 1:1], df[2, 1:1]]
    nt = NamedTuple.(dfr)
    gk = [keys(groupby(df, [:a, :b])); keys(groupby(df, :a))]

However, I cannot reproduce this error on nightly locally.

The error produced is:

MethodError: no method matching issparse(::GroupedDataFrame{DataFrame})
    The function `issparse` exists, but no method is defined for this combination of argument types.

but I do not get it locally. What could be the reason?

I get:

julia> @which vcat(keys(groupby(df, [:a, :b])), keys(groupby(df, :a)))
vcat(V::AbstractVector{T}...) where T
     @ Base abstractarray.jl:1747

while in the tests it seems that the invoked funciton is:

vcat(X1::DataFrames.GroupKeys{GroupedDataFrame{DataFrame}}, X::DataFrames.GroupKeys{GroupedDataFrame{DataFrame}})
        @ SparseArrays /opt/hostedtoolcache/julia/1.14.0-beta1/x64/share/julia/stdlib/v1.14/SparseArrays/src/concatenation.jl:596

@andreasnoack

Copy link
Copy Markdown
Member Author

The failure on pre (MethodError: no method matching issparse(::GroupedDataFrame{DataFrame}) in test/dataframerow.jl) is caused by a change in SparseArrays, see JuliaSparse/SparseArrays.jl#1191. A fix is proposed in JuliaSparse/SparseArrays.jl#1192.

This comment was written by generative AI (Claude).

The ubuntu-latest runner image is now Ubuntu 26.04, which no longer
ships with Julia preinstalled.
@bkamins
bkamins merged commit f995e95 into main Nov 22, 2026
7 of 8 checks passed
@bkamins
bkamins deleted the andreasnoack-patch-1 branch November 22, 2026 19:11
@bkamins

bkamins commented Nov 22, 2026

Copy link
Copy Markdown
Member

Thank you!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CI ecosystem Issues in DataFrames.jl ecosystem

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants