Updating Packages Within a Packrat Project

185 views
Skip to first unread message

Jeff Bruce

unread,
Aug 28, 2015, 11:42:51 AM8/28/15
to packrat-discuss
Hi all,

I'm a new packrat user.  I have a project directory that I'm working from and successfully called packrat::init() within this project directory.  I notice, however, that it installs an outdated version of the shiny package (v. 0.11.1), compared to the current version which is 0.12.2.  I need the newer version of shiny for my project because I plan on using the interactive table widget from the htmlwidgets package, which R claims that I need the newer version of shiny for the table to render properly.  So, I call update.packages(), and for each package I am asked whether I want to update it.  I tell R that I only want to update shiny, so it goes off and downloads/installs shiny and its associated dependencies.  The update fails with the following error messages:

Error in library(RMINC) : there is no package called ‘RMINC’
Calls: .First -> library
Execution halted
Error in library(RMINC) : there is no package called ‘RMINC’
Calls: .First -> library
Execution halted
Error in library(RMINC) : there is no package called ‘RMINC’
Calls: .First -> library
Execution halted
Error in library(RMINC) : there is no package called ‘RMINC’
Calls: .First -> library
Execution halted

It attempts to install httpuv, mime, and jsonlite, since I believe they are dependencies for shiny.  RMINC is an R library which is used system-wide at my workplace.  I'm not sure why RMINC is even being referenced at all because RMINC cannot be found in any of the directories listed in my new .libPaths() (.libPaths() values change as a result of packrat::init()).  Any ideas why this might be happening or how I can get around this?

Thanks.

Kevin Ushey

unread,
Sep 10, 2015, 5:48:52 PM9/10/15
to packrat-discuss
Hi Jeff,

Is it possible that RMINC is being loaded by a global .Rprofile, e.g. at ~/.Rprofile? Or perhaps it's made available by some other mechanism? The fact that the failure occurs in a call to .First seems to imply it's coming from a .Rprofile somewhere.

Thanks,
Kevin
Reply all
Reply to author
Forward
0 new messages