[Bioc-devel] return value of BiocManager::install()

[Martin Morgan]

require is expensive and detach() fraught with subtle difficulties. Using withCallingHandlers is more elegant than capture.output(). See and direct further commentary to the issue.

Martin

On 1/22/19, 4:41 AM, "Bioc-devel on behalf of Levi Waldron" <bioc-devel-bounces using r-project.org on behalf of lwaldron.research using gmail.com> wrote:

    BTW I just noticed the alternative approach of searching through the
    warnings, used at
    https://github.com/Bioconductor/bioc_docker/blob/5be6d37ded4f97995640cbb3a7d201165234b1e4/src/core/install.R.in#L42
    :
    
    if (!is.null(warnings()))
    {
        w <- capture.output(warnings())
        if (length(grep("is not available|had non-zero exit status", w)))
            quit("no", 1L)
    }
    
    
    On Tue, Jan 22, 2019 at 10:19 AM Levi Waldron <lwaldron.research using gmail.com>
    wrote:
    
    > I opened an issue on BiocManager (
    > https://github.com/Bioconductor/BiocManager/issues/39) somewhat related
    > to Sean's recent post, about what BiocManager::install() should return, and
    > it occurs to me that I should mention it here too for feedback:
    >
    > > Having `BiocManager::install()` return the "pkgs" argument is not
    > especially useful. Could it instead return success/error codes? Otherwise,
    > how can you tell in a script whether the requested packages were installed
    > successfully or not?
    >
    > The most basic return value I can think of would be like `require()`, TRUE
    > if successful and FALSE if unsuccessful. More sophisticated behavior could
    > give different error codes for different failures, although I don't know
    > how difficult that would be to implement. This came up from wanting to test
    > a Docker container for ability to install many different packages within
    > it, and doing this by making a (hacked and specific to my purpose of
    > installing/testing one package at a time) wrapper around
    > `BiocManager::install()` that uses require() to test whether the
    > installation was successful. That said, the workaround was not that
    > difficult and it's probably not a priority for too many other users if
    > you're hesitant to change the return value of BiocManager::install().
    >
    > installcheck <- function(x){
    >   if (x %in% installed.packages())
    >     return(TRUE)
    >   BiocManager::install(x, ask=FALSE, update=FALSE)
    >   res <- require(x, character.only = TRUE)
    >   detach(paste0("package:", x), unload=TRUE, character.only = TRUE)
    >   return(res)
    > }
    >
    >
    >
    
    -- 
    Levi Waldron
    http://www.waldronlab.io
    Associate Professor of Biostatistics    CUNY School of Public Health
    US: +1 646-364-9616                                           Skype:
    levi.waldron
    
    	[[alternative HTML version deleted]]
    
    _______________________________________________
    Bioc-devel using r-project.org mailing list
    https://stat.ethz.ch/mailman/listinfo/bioc-devel