Skip to content

building the development version

11 messages · Deborah Swayne, Brian Ripley, A.J. Rossini +2 more

#
I'm trying to build the current development version of R on an SGI
running IRIX6.5.  Everything proceeds merrily until I reach the eda
package.  This is the error message I get if I cd to the appropriate
directory and type 'make' or 'make all.'  Does this make sense to
anyone?  -- Thanks, Debby

gmake[3]: Entering directory `/l/fsc/dfs/src/R/src/library/eda'
building package 'eda'
../../../library/eda/R/eda is unchanged
../../../library/eda/man/eda.Rd is unchanged
/home/dfs/src/R/src/library/eda/src
gmake[4]: Entering directory `/l/fsc/dfs/src/R/src/library/eda/src'
config.status: error: invalid argument: src/library/eda/src/Makefile
#
Are you rebuilding?  

    ./config.status 

might need to be run to remake the Makefile, or:

      ./config.status --recheck ; ./config.status

(from top level directory in the R tree).  I've had the same problems
before. 

Deborah Swayne <dfs at research.att.com> writes:

  
    
#
On Wed, 18 Feb 2004, Deborah Swayne wrote:

            
There should be no such directory.  Did you do this in a clean build 
directory, as this is what would happen if you try to rebuild over 
sources from a couple of months ago?  Please try again starting in an 
empty build directory.

  
    
#
Prof Brian Ripley <ripley at stats.ox.ac.uk> writes:
This directory exists on a clean anonCVS checkout, cvs co, followed
with a cvs update -Pd.  Done AFTER Debbie's post, on a new machine.

I verified Debbie's results, and the solution I suggested (2
incantations, with a weird incantation and the removal of the eda
subdir) got me most of the way there.  The second incantation was
needed to cope with a problem in the ctest directory.

best,
-tony
#
On Wed, 18 Feb 2004, A.J. Rossini wrote:

            
Looks like there is a problem with anonCVS.  That directory is empty
in the master CVS tree and hence does not appear in the daily snapshots.
I've just checked: it is not there in R-devel_2004-02-18.tar.bz2.

What is in it under anonCVS?
Looks like we need to take some preventative measures in any case ....
#
rossini at blindglobe.net (A.J. Rossini) writes:
There should actually be an eda directory. On my system, with a fresh
checkout from the developer CVS, it ends up building a package
containing only

.noGenerics <- TRUE

.First.lib <- function(lib, pkg)
{
    have.stats <- "package:stats" %in% search()
    if(!have.stats) require("stats")
    warning("package ", sQuote("eda"), " has been merged into ",
            sQuote("stats"), call. = FALSE)
}

I can't reproduce Debbie's problem though...
#
On 18 Feb 2004, Peter Dalgaard wrote:

            
Yes, but no eda/src nor eda/man.

  
    
#
Prof Brian Ripley <ripley at stats.ox.ac.uk> writes:
OK, that I can confirm. They do turn up on a checkout without the -P
option, but disappear after "cvs up -Pd". Tony, Debbie, did you
perhaps configure before attempting to prune the empty dirs? Could you
try 

cvs -d ...blabla... co -P R

        -p

BTW: Make dist appears to be broken?
#
Peter Dalgaard writes:
 > 
 > OK, that I can confirm. They do turn up on a checkout without the -P
 > option, but disappear after "cvs up -Pd". Tony, Debbie, did you
 > perhaps configure before attempting to prune the empty dirs? Could you
 > try 
 > 
 > cvs -d ...blabla... co -P R

I've use rsync; I don't know if that makes any difference:
   rsync -rC rsync.r-project.org::r-devel R

In any case, I deleted all the old source code and started over,
and the development version emerged without any difficulty.

Thanks for helping me think about this.

Debby
#
Deborah Swayne <dfs at research.att.com> writes:
Hmm. It might mean that the rsync tree is built without the -P. If so,
we should probably fix it.
Sounds like this was after Brian fixed the Makefiles.
#
Peter Dalgaard <p.dalgaard at biostat.ku.dk> writes:
I think it is rather that Deborah's rsync call did not include a
--delete option.  The crontab entries that update the rsync trees all
use -Pd (which I manage to remember because of your login).

11  *  * * *    (cd /spare1/R-src/r-patched; cvs -Q up -Pd -r R-1-8-patches; cd ../r-devel; cvs -Q up -Pd)

22 02 * * *     (cd /spare1/ggobi-src/RSggobi; cvs -Q up -Pd; cd ../ggobi; cvs -Q up -Pd)

22 05 * * *     (cd /spare1/R-project-web; cvs -Q up -Pd)

19 01 * * *     (cd /spare1/Bioconductor-devel/madman; cvs -Q up -Pd)