Many people use Biostrings without wanting to be forced to effectively call e.g. library(S4Vectors). Various non-Biostrings functions being imported in a user's R session leads to confusing problems such as e.g. dplyr's rename() being superseded by plyr's rename() if and only if Biostrings is loaded after dplyr. Since functions such as plyr's rename do not seem required to use Biostrings, wouldn't it be better to put the offending packages as Imports? I think it is general best practice for R packages to use Imports over Depends for this reason. (I may very well be missing some obvious reasons for why these packages are in Depends and perhaps the Bioconductor team has considered all this and decided it is a necessary evil, but posting as I didn't see an issue on it yet).
Edit: closing this because the plyr::rename issue should probably be brought up at S4Vectors and I should do more digging to see if Biostrings is designed to only be used with the Depends packages.
Many people use Biostrings without wanting to be forced to effectively call e.g. library(S4Vectors). Various non-Biostrings functions being imported in a user's R session leads to confusing problems such as e.g. dplyr's rename() being superseded by plyr's rename() if and only if Biostrings is loaded after dplyr. Since functions such as plyr's rename do not seem required to use Biostrings, wouldn't it be better to put the offending packages as Imports? I think it is general best practice for R packages to use Imports over Depends for this reason. (I may very well be missing some obvious reasons for why these packages are in Depends and perhaps the Bioconductor team has considered all this and decided it is a necessary evil, but posting as I didn't see an issue on it yet).
Edit: closing this because the plyr::rename issue should probably be brought up at S4Vectors and I should do more digging to see if Biostrings is designed to only be used with the Depends packages.