R help - Aug 2001 - Warning when building packages and update suggestion

I want to share this knowledge to save you the hours it took me to
troubleshoot the following problem. Maybe you already know about this, but I
certainly didn't. In the end I also suggest an update fix.

PROBLEM:
I have been building packages for a couple of months now without any
problems. However, a couple of hours ago I ran into a strange problem, which
I first couldn't understand or fix:

  >Rcmd build --binary R.base
  ...
  make[1]: *** [/tmp/Rbuild.468075/R.base/zzzz] Error 1
  make[1]: Leaving directory `/users/braju.com.R/public_html/R.base/R.base'
  make: *** [pkg-R.base] Error 2
  make: Leaving directory `/progs/R/rw1030/src/gnuwin32'
  *** Installation of R.base failed ***
  ...

REASON:
As usual you think that *you* didn't do anything different than usual.
However, yesterday I added a library(R.base) in my ~/R.profile file, i.e.
every time [R] starts the library R.base is automatically loaded. After
imitating the same make calls as 'Rcmd build --binary R.base' does and
looking into on of the temporary files created, I saw that during the build
process [R] was started. I turned out that [R] failed because it tried to
load the library I was currently building and from the previous build I did
an error was introduced. So, the installed R.base didn't work, which made
any try rebuild it to fail.

SOLUTION:
My solution was to remove ~/R.profile and everything worked as usual.


UPDATE SUGGESTION:
I don't know much about the package building process, but couldn't [R]
*always* be started with the options --no-site-file (Suppress reading of the
site-wide startup profile) and --no-init-file (Suppress reading of the
directory or user's .Rprofile file) when Rcmd build is used? I was trying to
look into the make files where this should be done, but I couldn't find it.


Henrik Bengtsson
Lund University/UC Berkeley

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

First, this is a comment on *one* way of building packages under *Windows*.
So it is hardly universal.  And you did not even mention the OS.

Second, is not R.base a confusing name?  There is already an R base
package, and the solution is already used for that!

Third, the sources are all there for you to investigate, and the
information is in the documentation.  R is run to create the time stamp.
Make your package by

make STAMP=NO pkg-R.base

to avoid this, or alter the line

        @echo  "cat('Built: R ', R.version[['major']],
'.',
R.version[['minor']], ';  Win32;  ', format(Sys.time()),
sep='')" |
$(RHOME)/bin/Rterm --slave >> ${DPKG}/DESCRIPTION

in src/gnuwin32/MakePkg as you wish.  I have added --vanilla.

BDR

On Fri, 3 Aug 2001, Henrik Bengtsson wrote:
> I want to share this knowledge to save you the hours it took me to
> troubleshoot the following problem. Maybe you already know about this, but
I
> certainly didn't. In the end I also suggest an update fix.
>
> PROBLEM:
> I have been building packages for a couple of months now without any
> problems. However, a couple of hours ago I ran into a strange problem,
which
> I first couldn't understand or fix:
>
>   >Rcmd build --binary R.base
>   ...
>   make[1]: *** [/tmp/Rbuild.468075/R.base/zzzz] Error 1
>   make[1]: Leaving directory
`/users/braju.com.R/public_html/R.base/R.base'
>   make: *** [pkg-R.base] Error 2
>   make: Leaving directory `/progs/R/rw1030/src/gnuwin32'
>   *** Installation of R.base failed ***
>   ...
>
> REASON:
> As usual you think that *you* didn't do anything different than usual.
> However, yesterday I added a library(R.base) in my ~/R.profile file, i.e.
> every time [R] starts the library R.base is automatically loaded. After
> imitating the same make calls as 'Rcmd build --binary R.base' does
and
> looking into on of the temporary files created, I saw that during the build
> process [R] was started. I turned out that [R] failed because it tried to
> load the library I was currently building and from the previous build I did
> an error was introduced. So, the installed R.base didn't work, which
made
> any try rebuild it to fail.
>
> SOLUTION:
> My solution was to remove ~/R.profile and everything worked as usual.
>
>
> UPDATE SUGGESTION:
> I don't know much about the package building process, but couldn't
[R]
> *always* be started with the options --no-site-file (Suppress reading of
the
> site-wide startup profile) and --no-init-file (Suppress reading of the
> directory or user's .Rprofile file) when Rcmd build is used? I was
trying to
> look into the make files where this should be done, but I couldn't find
it.
>
>
> Henrik Bengtsson
> Lund University/UC Berkeley
>
>
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
> r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
>
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
>
-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._