[Rd] Suggestion: Add DESCRIPTION 'Date' to R CMD check log header

Martin Maechler maechler at stat.math.ethz.ch
Wed Sep 15 09:49:13 CEST 2010


Hi Henrik

>>>>> "HB" == Henrik Bengtsson <hb at stat.berkeley.edu>
>>>>>     on Tue, 14 Sep 2010 18:26:31 -0700 writes:

    HB> Hi,
    HB> in R CMD check, the version of the package being checked is reported, e.g.

    HB> Thu Sep  9 05:02:30 2010: Checking package R.utils (SVN revision 399) ...
    HB> * using log directory ‘/srv/R/R.check/R-devel/PKGS/R.utils.Rcheck’
    HB> * using R version 2.12.0 Under development (unstable) (2010-09-07 r52876)
    HB> * using platform: x86_64-unknown-linux-gnu (64-bit)
    HB> * using session charset: UTF-8
    HB> * checking for file ‘R.utils/DESCRIPTION’ ... OK
    HB> * this is package ‘R.utils’ version ‘1.5.2’
    HB> ...

    HB> I'd like to request/suggest that the 'Date' in the DESCRIPTION file is
    HB> also added, e.g.

    HB> * this is package ‘R.utils’ version ‘1.5.2’ ('2010-09-14')


    HB> WHY?
    HB> This would be particular useful when you work toward sites like
    HB> R-forge and Bioconductor when you may commit your day's work on
    HB> package when you update the 'Date' but you do not really want to
    HB> update the 'Version' because you're going to put in more work
    HB> tomorrow.  With the 'Date' information you'll be able to see what
    HB> "version" of your updates have been checked by the servers.  I
    HB> understand that this may be an odd process to follow even for devel
    HB> branches and you may argue that you should always bump the version
    HB> number whenever you do an SVN commit (e.g. '1.5.2.1' for temporary
    HB> commits).  Either way, I find it useful to see the date as well.

I agree that this is useful.
Of course, for all those cases, where there's no  "Date:",
nothing (i.e. no "()") should be written.

If you (or someone) provide patches against  R-devel
( https://svn.r-project.org/R/trunk/ )
and they pass 'make check-all', I'd add this feature.

Martin



More information about the R-devel mailing list