[Rd] .onLoad failing because could not find function "loadMethod"

[Martin Morgan]

----- Original Message -----
> The thing that happens during the check is that the methods package is
> not among the automatically loaded and attached packages.
> 
> When the pkgB packages is attached, the Namespaces of pkgA and methods
> are loaded but not attached.
> 
> Note that one of the checks tells you:
> 
> "A namespace must be able to be loaded with just the base namespace
> loaded: otherwise if the namespace gets loaded by a saved object, the
> session will be unable to start."
> 
> 
> If you like to see into the details more easily, just start R with
> 
> R R_DEFAULT_PACKAGES=NULL
> and then try to
> library("pkgB")
> and you are able to reproduce the check messages.

yep, but .onLoad code can be written as

  a = new("A")
  length(a)

the error is on length(a) -- after methods::new has been found. Also, inserting a browser() call in .onLoad and then 

Browse[1]> names(getNamespaceInfo(parent.env(environment()), "imports"))
[1] "base"    "methods" "pkgA"    "pkgA"   

or 

Browse[1]> findFunction("loadMethod")
[[1]]
<environment: 0x752d90>
attr(,"name")
[1] "imports:pkgB"

shows that the methods name space is where it should be. It is the dispatch on the promoted length (in pkgA) that is awry.

Martin

> 
> You can avoid n-1 Warnings/Errors in the checks by adding
> library("methods")
> in your .onLoad function, but of course you cannot get rid of the
> check
> that checks that the package can be loaded with the base namespace
> only.
> 
> Best,
> Uwe Ligges
> 
> 
> 
> 
> 
> 
> 
> 
> On 15.10.2011 02:46, Hervé Pagès wrote:
> > Hi,
> >
> > This strange warning happen sometimes when running 'R CMD check':
> >
> > * checking whether the package can be loaded with stated
> > dependencies
> > ... WARNING
> >
> > Error : .onLoad failed in loadNamespace() for ‘pkgB’, details:
> > call: length(new("A"))
> > error: could not find function "loadMethod"
> > Error: package/namespace load failed for ‘pkgB’
> > Execution halted
> >
> > It looks like this package (or one of its dependent packages) has an
> > unstated dependence on a standard package. All dependencies must be
> > declared in DESCRIPTION.
> > See the information on DESCRIPTION files in the chapter ‘Creating R
> > packages’ of the ‘Writing R Extensions’ manual.
> >
> > The same warning is reported during the 2 next checks:
> >
> > * checking whether the package can be unloaded cleanly ... WARNING
> > <same error here>
> > * checking whether the namespace can be loaded with stated
> > dependencies
> > ... WARNING
> > <same error here>
> >
> > It seems to happen when pkgB has code in its .onLoad hook that
> > tries to call a method defined in pkgA (pkgB imports pkgA) on an
> > object
> > 'x' that is an instance of a class defined in pkgA. For some obscure
> > reason it seems that, at the time the .onLoad hook is executed, the
> > methods package (which is imported in both packages) is not working
> > properly.
> >
> > I've put 2 toy packages here that illustrate the problem:
> >
> > https://rapidshare.com/files/4043144391/pkgA_1.0.tar.gz
> >
> > https://rapidshare.com/files/1959508331/pkgB_1.0.tar.gz
> >
> > I'm using R version 2.14.0 alpha (2011-10-10 r57211).
> >
> > The .onLoad hook in pkgB is:
> >
> > .onLoad <- function(libname, pkgname)
> > {
> > showMethods("length")
> > length(new("A"))
> > }
> >
> > Class A and the length method for this class is defined in pkgA.
> >
> > Because of the showMethods("length") statement, you should see this
> > when running R CMD check pkgB_1.0.tar.gz:
> >
> > * checking whether the package can be loaded with stated
> > dependencies
> > ... WARNING
> > Function: length (package base)
> > x="A"
> >
> > Error : .onLoad failed in loadNamespace() for ‘pkgB’, details:
> > call: length(new("A"))
> > error: could not find function "loadMethod"
> > Error: package/namespace load failed for ‘pkgB’
> > Execution halted
> >
> > Any insight on this would be appreciated.
> >
> > Thanks,
> > H.
> >
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel