[Rd] R package Matrix failing to build with R-devel on Ubuntu AMD64

Martin Maechler maechler at stat.math.ethz.ch
Thu Jan 31 11:54:49 CET 2008


>>>>> "MWK" == Mark W Kimpel <mwkimpel at gmail.com>
>>>>>     on Tue, 29 Jan 2008 23:14:30 -0500 writes:

    MWK> Switched back to R-devel tonight after using R-2.6.1 for awhile and when 
    MWK> I updated my devel packages Matrix installation failed. Not critical for 
    MWK> me but thought I should report it. I am using Ubuntu Gusty AMD64 OS. 
    MWK> Tail of install output and sessionInfo follows. Mark

    MWK> Loading required package: methods
    MWK> Loading required package: lattice

    MWK> Attaching package: 'Matrix'


    MWK> The following object(s) are masked from package:stats :

    MWK> xtabs

    MWK> Error in new("dgCMatrix") :
    MWK> trying to generate an object from a virtual class ("dgCMatrix")
                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

    MWK> Calls: <Anonymous> ... eval.parent -> eval -> eval -> eval -> eval -> 
    MWK> new -> .Call
    MWK> Execution halted
    MWK> ERROR: installing package indices failed
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
    MWK> ** Removing '/home/mkimpel/R_HOME/site-library/Matrix'
    MWK> ** Restoring previous '/home/mkimpel/R_HOME/site-library/Matrix'

    MWK> The downloaded packages are in
    MWK> /tmp/RtmpJviXfF/downloaded_packages
    MWK> Warning message:
    MWK> In install.packages("Matrix") :
    MWK> installation of package 'Matrix' had non-zero exit status
    >> sessionInfo()
    MWK> R version 2.7.0 Under development (unstable) (2008-01-29 r44238)
    MWK> x86_64-unknown-linux-gnu

Thank you, Mark, for the report.

Indeed --- to my own very big astonishment --- 
I *can* reproduce the problem.

In the above messages, it's pretty funny that
"dgCMatrix" is said to be a virtual class (which it is clearly
not from the Matrix source).

The "daily" CRAN package checks don't show the problem yet,
but the checks' R version is

>>  R 2.7.0 Under development (unstable) (2008-01-27 r44199)

which is a bit older, than the one you used 
(and even older than the one I used ..... r44258)

I'm pretty sure I know around where the problem appeared..
More on this later.

Martin Maechler, ETH Zurich



More information about the R-devel mailing list