Rsamtools fails to install if a message is printed from .Rprofile. (I had print("JDEnvironment loaded") in my .Rprofile)

Bioconductor version 3.12 (BiocManager 1.30.12), R 4.0.3 (2020-10-10)
Installing package(s) 'Rsamtools'
trying URL 'https://bioconductor.org/packages/3.12/bioc/src/contrib/Rsamtools_2.6.0.tar.gz'
Content type 'application/x-gzip' length 2906572 bytes (2.8 MB)
==================================================
downloaded 2.8 MB
[1] "JDEnvironment loaded"
* installing *source* package ‘Rsamtools’ ...
** using staged installation
** libs
gcc -std=c99 -I"/usr/local/Cluster-Apps/R/R.4.0.3/include" -DNDEBUG [1] "JDEnvironment loaded" -D_FILE_OFFSET_BITS=64 -[...]
gcc: error: [1]: No such file or directory

Obviously, this message is passed on to gcc as debug message without correct quotation. Installation works fine after removing this message.

Perhaps it is bad practice to print from .Rprofile; however, I have yet not encountered another package having difficulties being installed with this.

A similar issue has been reported recently for rhdf5filters (https://github.com/grimbough/rhdf5filters/issues/11) and rhdf5 (https://github.com/grimbough/rhdf5/issues/81). They all use the same mechanism for extracting the flags used during compilation, and fall over when .Rprofile leads to some printing.

Out of interest, are the messages printed using cat() or message() (or something else)?

I wonder if something something has changed recently, as this code has been stable for a few years and I've never seen the issue reported before. Now there's been a flurry of reports across different packages and users in the space of a few weeks.

I'd suggest opening an issue on the Rsamtools GitHub page (https://github.com/Bioconductor/Rsamtools/issues/). I asked for some guidance on the developers mailing list here and came up with hopefully a workable solution, feel free to include that in the issue.