R CMD check dependency simplification

Friedrich.Leisch@ci.tuwien.ac.at Friedrich.Leisch@ci.tuwien.ac.at
Sun, 6 Oct 2002 09:47:47 +0200

>>>>> On Tue, 01 Oct 2002 12:50:28 -0400,
>>>>> Paul Gilbert (PG) wrote:

  > I am trying to re-organizing my Makefile for testing my R packages and I
  > am having trouble finding a structure I like. The problem seems to be
  > that "R CMD check", which does wonderful things, does not allow
  > dependency information to flow through very gracefully. That is, since
  > everything is copied into the pkg.Rcheck directory, it is difficult to
  > make any changes to source files without having to completely re-do "R
  > CMD check pkg". 

  > For example, I would like to be able to make changes to
  > src/pkg1/man/foo.Rd and have dependencies that indicate only codoc and
  > examples need to be re-run, not tests and vignettes. (R CMD check can to
  > do this, the problem is moving the appropriate new pieces, in the
  > appropriate way, into the pkg1.Rcheck directory.)

I absolutely agree that this would be nice to have, but we don't have
any tools for that yet (and I don't think it is all that simple to get
it right).

If there is a clear dependency structure like you describe, i.e., pkg3
depends on pkg2 which depends on pkg1 one here is what I currently do:

check pkg1, when it passes INSTALL it into a development library and
add that to your R_LIBS variable. Then check pkg2 and do the
same. Then check pkg3.

  > The difficulty is exacerbated for me because I have pkg3 depends on pkg2
  > depends on pkg1. If I make a change to src/pkg1/inst/doc/guide.tex then
  > I really should only need to re-do the vignettes for pkg1, but instead I
  > have to re-do the whole "R CMD check" for pkg1, pkg2, and pkg3.

Why would you run R CMD check on pkg2 if you only change the vignette
of pkg1?


r-devel 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-devel-request@stat.math.ethz.ch