HenrikBengtsson/matrixStats

Packages that break with new useNames=TRUE

HenrikBengtsson opened this issue · 9 comments

With the new useNames = TRUE as the default, there is 1 broken Bioconductor release package and 2 broken Bioconductor devel packages (2023-07-31). All CRAN packages are ok.

CRAN

Last updated: 2023-06-01

Bioconductor

Last updated: 2023-07-31

Regardless what the new default will be, these should be updated to explicitly set useNames = FALSE in the locations where useNames = TRUE break.

For at least some of the packages, the regression is related to the fact that the test code observes that one vector is named, whereas the other is not and thus reports a failure.

Can someone help reaching out to the above 6 Bioconductor packages?

They need to update the parts of the code/tests that break if useNames = TRUE becomes the new default. Note that the default might become FALSE, so they need to be updates to handle either case. The easiest is for them to specify useNames = FALSE where the code/tests break, because that seems to be what they expect the case to be right now.

Hi Henrik,

I have been following this issue, I just have been very busy the past two weeks and an ok holidays until the end of next week. But I plan to update sparseMatrixStats when I am back :)

Several Bioconductor packages on the Bioc devel branch are still broken. But since it's been 1.5 months since matrixStats 1.0.0 was released, I'll stop monitoring these broken packages.

We've got a patches for MatrixGenerics and DelayedMatrixStats (and I think sparseMatrixStats) ready for BioC-devel.
I'm waiting to hear back from Herve (who I think is on leave) and then we'll incorporate.

That's good news. Thanks for the update.

I've just pushed the patched versions of MatrixGenerics and DelayedMatrixStats to Bioconductor's git server.
I think @const-ae has already pushed his updates to sparseMatrixStats - is that right @const-ae ?

I'll keep an eye on the next build report but these build & check successfully locally for me.

I think @const-ae has already pushed his updates to sparseMatrixStats - is that right @const-ae ?

Yes :)

I think @const-ae has already pushed his updates to sparseMatrixStats - is that right @const-ae ?

Yes :)

@const-ae , it looks like the Bioc devel version still needs to be fixed.