R install.package() bug

[Markus Hönicka]

> Stephen P Carrier via Cygwin <cygwin@cygwin.com> hat am 2023-11-28 06:01 CET geschrieben:
> 
>  
> I would like to report an issue with install.packages() experienced today
> using the current R package.
> 
> I installed R 4.3.0 on system:
> 
> # uname -rs
> CYGWIN_NT-10.0-20348 3.4.9-1.x86_64
> 
> I tried to install the DBI package, DBI_1.1.3.tar.gz, downloaded from a
> mirror.
> 
> I used
> 
> install.packages("/path/to/DBI_1.1.13.tar.gz", repos = NULL);
> 
> What happens is:
> 
> -----
> Installing package into ‘/usr/lib/R/site-library’
> (as ‘lib’ is unspecified)
> During startup - Warning message:
> package ‘stats’ in options("defaultPackages") was not found
> * installing *source* package ‘DBI’ ...
> ** package ‘DBI’ successfully unpacked and MD5 sums checked
> ** using staged installation
> Error : Invalid DESCRIPTION file
> 
> Malformed maintainer field.
> 
> See section 'The DESCRIPTION file' in the 'Writing R Extensions'
> manual.
> ERROR: installing package DESCRIPTION failed for package ‘DBI’
> -----
> 
> I looked in the manual to see what DESCRIPTION files should look like.  The
> encoding is specified as UTF-8 in the file so the umlauts sprinkled
> throughout should be fine.  The 'Maintainer:' line contains an umlaut, I
> removed it, after which the package installed without error.  Umlauts
> elsewhere in the DESCRIPTION file are not a problem.
> 
> Seems like a bug.  I don't have R 4.3.0 installed on a Linux system to test
> if the bug is upstream from Cygwin, and my immediate problem is solved with
> a simple workaround.
> 
> Thanks to all for maintaining this fine product (Cygwin kernel and all the
> packages).
> 
> --Stephen Carrier
> 

Hi Stephen,
I came across this type of error with R on Cygwin, but not on FreeBSD or Linux. I'm still running R version 3.6.3 on Cygwin (I don't have root permissions on this box, so I'm forced to upgrade very infrequently) and I have positively seen this error when I set up my current installation. I use the same R packages on my Unix boxes at home and have never seen such an error. I don't think this is an upstream error. Unfortunately this is above my head to debug.
regards,
Markus