In the survival package the survConcordance function was replaced by concordance a while ago, the latter works for any ordered response (continuous, binary, survival, ...). I deprecated the old one a couple of years ago. How long should I have this persist before converting it to Defunct and removing the old function? Terry -- Terry M Therneau, PhD Department of Quantitative Health Sciences Mayo Clinic therneau at mayo.edu "TERR-ree THUR-noh" [[alternative HTML version deleted]]
>>>>> Therneau, Terry M , Ph D via R-devel >>>>> on Wed, 30 Jul 2025 09:04:39 -0500 writes:> In the survival package the survConcordance function was replaced by concordance a while > ago, the latter works for any ordered response (continuous, binary, survival, ...). > I deprecated the old one a couple of years ago. > How long should I have this persist before converting it to Defunct and removing the old > function? > Terry > -- > Terry M Therneau, PhD > Department of Quantitative Health Sciences > Mayo Clinic > therneau at mayo.edu > "TERR-ree THUR-noh" Making it defunct is different than removing it. You'd replace the old function by a much shorter version survConcordance <- function(formula, data, weights, subset, na.action) .Defunct("concordance") and also update help files, and possibly examples / demos / tests. For such an important and long-history package as `survival` I'd then keep this little bit of code and docs "forever". If in 20 years, somebody tries to run R code from a script or a book that is 30 years old, they would still be happy to be shown why their code calling survConcordance() does not run, and how they can amend it .. Martin
On Wed, 30 Jul 2025 09:04:39 -0500 "Therneau, Terry M., Ph.D. via R-devel" <r-devel at r-project.org> wrote:> In the survival package the survConcordance function was replaced by > concordance a while ago, the latter works for any ordered response > (continuous, binary, survival, ...). I deprecated the old one a > couple of years ago.For a less painful switch from Deprecated to Defunct, it should help to minimise its use in existing CRAN and Bioconductor packages. On CRAN, it's still used by the following packages: - 'arsenal', one vignette, CRAN version from 2021, last change on GitHub in 2024 - 'blockForest', R code, CRAN version from 2023 - 'distcomp', R code, CRAN version from 2022 - 'grpreg', \link to "survival-deprecated" in documentation - 'MTLR', R code, CRAN version from 2019, last change on GitHub in 2020 (A few remaining search results consist of commented out code and news entries about having avoided the deprecation warning.) On Bioconductor, it's two more packages, 'iNETgrate' and 'messina' (both of which seem to be stale imports, not actual uses of the function): https://code.bioconductor.org/search/search?q=survConcordance The maintainers may need some help removing these last few uses of the function. -- Best regards, Ivan