reinstalling function

Reinstalling binary dependencies

The glmmTMB package depends on several upstream packages, which it uses in a way that depends heavily on their internal (binary) structure. Sometimes, therefore, installing an update to one of these packages will require that you re-install a binary-compatible version of glmmTMB, i.e. a version that has been compiled with the updated version of the upstream package.

• If you have development tools (compilers etc.) installed, you should be able to re-install a binary-compatible version of the package by running install.packages("glmmTMB", type="source"). If you want to install the development version of glmmTMB instead, you can use remotes::install_github("glmmTMB/glmmTMB/glmmTMB"). (On Windows, you can install development tools following the instructions at https://cran.r-project.org/bin/windows/Rtools/; on MacOS, see https://mac.r-project.org/tools/.)

• If you do not have development tools and can't/don't want to install them (and so can't install packages with compiled code from source), you have two choices:

  • revert the upstream package(s) to their previous binary version. For example, using the checkpoint package:

    ## load (installing if necessary) the checkpoint package
    while (!require("checkpoint")) install.packages("checkpoint")
    ## retrieve build date of installed version of glmmTMB
    bd \\<- as.character(asDateBuilt(
     packageDescription("glmmTMB",fields="Built")))
    oldrepo \\<- getOption("repos")
    use_mran_snapshot(bd) ## was setSnapshot() pre-checkpoint v1.0.0
    install.packages("TMB")
    options(repos=oldrepo) ## restore original repo
    

    A similar recipe (substituting Matrix for TMB and TMB for glmmTMB) can be used if you get warnings about an incompatibility between TMB and Matrix.

  • hope that the glmmTMB maintainers have posted a binary version of the package that works with your system; try installing it via install.packages("glmmTMB",repos="https://glmmTMB.github.io/glmmTMB/repos",type="binary")

    If this doesn't work, please file an issue (with full details about your operating system and R version) asking the maintainers to build and post an appropriate binary version of the package.