[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Conflict between R 4.5 and some (mostly r-bioc 3.20) R packages



r-bioc-pwalign / r-bioc-shortread: this appears to be https://github.com/Bioconductor/pwalign/issues/11 ; I have pushed the upstream fix.

r-cran-rlang: I have pushed a version with an r-base-core (>= 4.5.0~) Depends.

r-bioc-bioccheck: the explicit version check is actually in r-cran-biocmanager (at https://sources.debian.org/src/r-cran-biocmanager/1.30.25+dfsg-1/R/version.R/?hl=329#L329 ), which is used by BiocCheck:::checkSingleColon and BiocCheck:::checkIsPackageNameAlreadyInUse. I have pushed a version removing the tests of those functions. (I also tried the override mechanism described at https://cran.r-project.org/web/packages/BiocManager/vignettes/BiocManager.html#unsupported-r-bioconductor-versions + https://sources.debian.org/src/dh-r/20241218/scripts/pkg-r-autopkgtest/?hl=32#L32 , but it didn't work.)

This version check will also fail BiocManager::install(), BiocManager::available(), BiocManager::repositories(), BiocCheck:::checkSingleColon, BiocCheck:::checkIsPackageNameAlreadyInUse if actual users try to use them, which raises the question of whether we should change the check message to say that this is expected in Debian, and that users can uninstall r-bioc-* and run BiocManager::install(version = "3.21") if they want an upstream-supported version.

(I *don't* think we should disable the check entirely: we *are* doing something upstream don't recommend, and the fact that I've done so much of the work is *not* intended to be an opinion on whether it was ever a good idea.)


Reply to: