[Rd] Warning: strange behaviour in RStudio in R 4.5.x .

[Kevin Ushey]

... and of course, immediately after clicking send, I realize you're
using macOS, not Windows. I don't think we need this sub-directory
approach on non-Windows operating systems, so if that's affecting the
behaviour here, that's surely an RStudio bug.

On Mon, Jun 30, 2025 at 10:20 AM Kevin Ushey <kevinushey using gmail.com> wrote:
>
> Hi Duncan,
>
> We haven't investigated yet, but I strongly suspect this is an RStudio issue.
>
> For extra context, RStudio is doing something similar to R's own
> staged installation here, where the package is initially installed
> into a _build sub-directory, and then moved to its final intended
> installation location following a successful installation. This
> approach can be useful on Windows because, at least in some contexts,
> it's possible to move a directory from location A to location B, even
> if files in location B are currently open / in-use in some
> application. For RStudio, this can help avoid an installation failure
> when installing a package which is already loaded in the current
> RStudio R session.
>
> In any event, would you be willing to file an issue at
> https://github.com/rstudio/rstudio/issues so we can properly track and
> investigate this? It will also be useful to know what version of
> Windows you're using, just in case the behavior described above
> depends on the Windows version being used.
>
> Thanks,
> Kevin
>
> On Sat, Jun 28, 2025 at 12:22 PM Duncan Murdoch
> <murdoch.duncan using gmail.com> wrote:
> >
> > I'm running RStudio on MacOS Sonoma 14.7.1 on an M4 Macbook.  I haven't
> > updated either RStudio or the OS for several months.
> >
> > However, I updated R to version 4.5.0 recently, and to 4.5.1 today.  In
> > both of those versions I'm seeing some strange behaviour when I ask
> > RStudio to install a package that I'm developing.  I haven't seen this
> > before, and I haven't noticed it in other packages.
> >
> > The behaviour I see is that the package is installed
> > and RStudio restarts the session and runs `library(pkg)`, but it loads
> > the old version of the package.
> >
> > I don't use the devtools option, so RStudio says it is running
> >
> >    R CMD INSTALL --preclean --no-multiarch --with-keep.source pkg
> >
> > If I manually restart the R session and run `library(pkg)` I get the new
> > one, so it appears that RStudio is loading the package before it is
> > fully installed.
> >
> > Has anyone else seen this?  I'd especially like to know if people have
> > seen it outside of RStudio.  I suspect it's a bug there, but as I said,
> > I haven't updated it recently.
> >
> > BTW, I think I know how to stop this in RStudio:  In the "Global Options
> > | Packages | Development" dialog, there's an option "Use alternate
> > library path when building package".
> >
> > If checked, this builds to
> >
> >
> > /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/library/_build
> >
> > instead of to
> >
> >    /Library/Frameworks/R.framework/Versions/4.5-arm64/Resources/library
> >
> > then moves the package into place afterwards.  I've only seen the
> > strange behaviour when I have that option checked, which is what makes
> > me think this is an RStudio bug --- but somehow triggered by something
> > new in R 4.5.x.
> >
> > Duncan Murdoch
> >
> > ______________________________________________
> > R-devel using r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-devel