Georgi Boshnakov
2014-Nov-25 14:42 UTC
[Rd] problem with setGroupGeneric from package methods
Hi,
There seems to be a problem with setGroupGeneric() from package
'methods'. The symptoms are somewhat erratic, in the sense that small
changes may lead to any of the following behaviours.
1.Package works without problems on Windows and installs on Linux but gives
error when loaded with library().
2.Package installs on both systems and gives error when loaded on any of them.
3. Package builds but fails to install on both systems.
4.Package fails to build on both systems.
Other combinations, depending also on the R version occur, as well.
The package in question is a pure R package with no OS specific code.
The error message in all cases is:
Error in .setupMethodsTables(generic) :
trying to get slot "group" from an object of a basic class
("NULL") with no slots
Error: package or namespace load failed for 'pctsData'
Similar problem has been reported before at
http://stackoverflow.com/questions/12368439/defining-group-generic-functions-in-an-r-package
A solution given by Romain Francois was to enclose the offending code in
evalqOnLoad() but this doesn't really solve the problem.
The code in that question can serve as a minimal example and I have packed it in
a package at
http://www.maths.manchester.ac.uk/~gb/Rpackages/grgen_1.0.tar.gz
but that package gives the error at installation time and I do not know how to
debug with the command line tools.
The example below is with my local package which installs and gives the same
error at attach time (I will make it available, together with its dependencies,
if needed).
> sessionInfo()
R version 3.1.2 (2014-10-31)
Platform: x86_64-unknown-linux-gnu (64-bit) (actually, Fedora)
locale:
[1] LC_CTYPE=en_GB.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_GB.UTF-8 LC_COLLATE=en_GB.UTF-8
[5] LC_MONETARY=en_GB.UTF-8 LC_MESSAGES=en_GB.UTF-8
[7] LC_PAPER=en_GB.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
> library(pctsData)
In one of the cases when the package installs the problem seems to appear after
cacheMetaData -> .updateMethodsInTable -> .updateMethodsInTable ->
.setupMethodsTables
in one of the recursive calls to updateMethodsInTable (the place is marked with
!!! towards the end of the code below).
At the time of the error in .setupMethodsTables(generic) 'generic' is
NULL and generic at group fails. The calling stack shown by the recover
facility is this:
==========================================Enter a frame number, or 0 to exit
1: library(pctsData)
2: try({
ns <- loadNamespace(package, c(which.lib.loc, lib.loc))
env <-
3: tryCatch(expr, error = function(e) {
call <- conditionCall(e)
if (!is
4: tryCatchList(expr, classes, parentenv, handlers)
5: tryCatchOne(expr, names, parentenv, handlers[[1]])
6: doTryCatch(return(expr), name, parentenv, handler)
7: loadNamespace(package, c(which.lib.loc, lib.loc))
8: methods:::cacheMetaData(ns, TRUE, ns)
9: .updateMethodsInTable(fdef, where, attach)
10: .updateMethodsInTable(getGeneric(g), where, attach)
11: .setupMethodsTables(generic)
=============================================
On a different Linux machine and R version (the standard faculty wide
installation), library(pctsData) completes fine but package "grgen"
mentioned above still gives the error at installation time.
The session info for this case:
R version 3.0.2 (2013-09-25)
Platform: x86_64-redhat-linux-gnu (64-bit)
locale:
[1] LC_CTYPE=en_GB.UTF-8 LC_NUMERIC=C
[3] LC_TIME=en_GB.UTF-8 LC_COLLATE=en_GB.UTF-8
[5] LC_MONETARY=en_GB.UTF-8 LC_MESSAGES=en_GB.UTF-8
[7] LC_PAPER=en_GB.UTF-8 LC_NAME=C
[9] LC_ADDRESS=C LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_GB.UTF-8 LC_IDENTIFICATION=C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] pctsData_0.0-3
loaded via a namespace
===================================================
Further details are after the signature below.
Georgi
--
Dr Georgi Boshnakov tel: (+44) (0)161 306 3684
School of Mathematics fax: (+44) (0)161 306 3669
Alan Turing Building 1.125
The University of Manchester email: Georgi.Boshnakov at manchester.ac.uk
Oxford Road
Manchester M13 9PL
UK
===================================================
Here is the relevant code from my package code.
When I narrowed down the cause of the problem to setGroupGeneric, I found the
following commented out excerpt, probably the error was the same. A change in
the syntax back then removed the problem.
## 2011-07-07 I was not able to compile pcts properly since R moved from 2.7.x
to 2.8.x.
## Today I finally investigated and discovered that the reason is
something to do
## with the group generic. I looked in the sources of package
"methods" for the
## way group generics are defined and switched to using argument
"known members"
## (see below) instead of the following commands.
## setGroupGeneric("pcfData", function(x) NULL)
## setGeneric("pc.nseasons" , group="pcfData") # ima
setGeneric i pri pcSeason. Check!
## setGeneric("pc.nepochs" , group="pcfData")
## setGeneric("pc.nvariables" , group="pcfData")
## setGeneric("pc.namesofseasons" , group="pcfData")
## setGeneric("pc.namesofvariables", group="pcfData")
## setGeneric("pc.data.matrix" , group="pcfData")
## setGeneric("pc.data.vec" , group="pcfData")
## setGeneric("pc.data.Vec" , group="pcfData")
## setGeneric("pc.data.tsvec" , group="pcfData")
## setGeneric("pc.data.tsVec" , group="pcfData")
# setGeneric("",group="pcfData")
setGeneric("pc.nepochs" )
setGeneric("pc.nvariables" )
setGeneric("pc.namesofseasons" )
setGeneric("pc.namesofvariables")
setGeneric("pc.data.matrix" )
setGeneric("pc.data.vec" )
setGeneric("pc.data.Vec" )
setGeneric("pc.data.tsvec" )
setGeneric("pc.data.tsVec" )
## 2014-11-24 modifying since causes problems on Linux (cannot load the
package).
## no luck.
## setGroupGeneric("pcfData", function(x) NULL
## , knownMembers = c("pc.nepochs"
## , "pc.nvariables"
## , "pc.namesofseasons"
## , "pc.namesofvariables"
## , "pc.data.matrix"
## , "pc.data.vec"
## , "pc.data.Vec"
## , "pc.data.tsvec"
## , "pc.data.tsVec"
## ))
pcfData <- function(x) NULL
setGeneric("pcfData")
setGroupGeneric("pcfData", # def = function(x) NULL,
knownMembers = c("pc.nepochs"
, "pc.nvariables"
, "pc.namesofseasons"
, "pc.namesofvariables"
, "pc.data.matrix"
, "pc.data.vec"
, "pc.data.Vec"
, "pc.data.tsvec"
, "pc.data.tsVec"
),
where = topenv()
)
#########################################################
Related code from "methods" package:
cacheMetaData <-
function(where, attach = TRUE, searchWhere = as.environment(where),
doCheck = TRUE)
{
## a collection of actions performed on attach or detach
## to update class and method information.
pkg <- getPackageName(where)
classes <- getClasses(where)
for(cl in classes) {
cldef <- (if(attach) get(classMetaName(cl), where) # NOT getClassDef,
it will use cache
else getClassDef(cl, searchWhere))
if(is(cldef, "classRepresentation")) {
if(attach) {
.cacheClass(cl, cldef, is(cldef,
"ClassUnionRepresentation"), where)
}
else if(identical(cldef at package, pkg)) {
.uncacheClass(cl, cldef)
.removeSuperclassBackRefs(cl, cldef, searchWhere)
}
}
}
generics <- .getGenerics(where)
packages <- attr(generics, "package")
if(length(packages) < length(generics))
packages <- rep(packages, length.out = length(generics))
if(attach && exists(".requireCachedGenerics", where,
inherits = FALSE)) {
others <- get(".requireCachedGenerics", where)
generics <- c(generics, others)
packages <- c(packages, attr(others, "package"))
}
## check for duplicates
dups <- duplicated(generics) & duplicated(packages)
generics <- generics[!dups]
for(i in seq_along(generics)) {
f <- generics[[i]]
fpkg <- packages[[i]]
if(!identical(fpkg, pkg) && doCheck) {
if(attach) {
env <- as.environment(where)
## All instances of this generic in different attached packages
must
## agree with the cached version of the generic for consistent
## method selection.
if(exists(f, envir = env, inherits = FALSE)) {
def <- get(f, envir = env)
fdef <- .genericOrImplicit(f, fpkg, env)
if(is.function(def)) {
## exclude a non-function of the same name as a
primitive with methods (!)
if(identical(environment(def), environment(fdef)))
next # the methods are identical
else if( is(fdef, "genericFunction")) {
.assignOverBinding(f, fdef, env, FALSE)
}
} # else, go ahead to update primitive methods
}
else # either imported generic or a primitive
fdef <- getGeneric(f, FALSE, searchWhere, fpkg)
}
else
fdef <- getGeneric(f, FALSE, searchWhere, fpkg)
}
else
fdef <- getGeneric(f, FALSE, searchWhere, fpkg)
if(!is(fdef, "genericFunction"))
next ## silently ignores all generics not visible from searchWhere
if(attach)
.cacheGeneric(f, fdef)
else
.uncacheGeneric(f, fdef)
methods <- .updateMethodsInTable(fdef, where, attach) ###
!!!
cacheGenericsMetaData(f, fdef, attach, where, fdef at package, methods)
}
.doLoadActions(where, attach)
invisible(NULL) ## as some people call this at the end of functions
}
###############
# add objects to the generic function's environment that allow
# table-based dispatch of methods
.setupMethodsTables <- function(generic,
initialize = !exists(".MTable", envir = env, inherits = FALSE))
{
env <- environment(generic)
if(initialize || !exists(".SigLength", envir = env, inherits =
FALSE)) {
nsig <- 1
## check that groups of generics agree on .SigLength; otherwise
## labels won't match
for(gp in generic at group) { ### !!!
gpDef <- getGeneric(gp)
if(is(gpDef, "genericFunction")) {
.getMethodsTable(gpDef) # force initialization
nsig <- max(nsig, get(".SigLength", envir =
environment(gpDef)))
}
}
assign(".SigLength", nsig, envir = env)
}
argSyms <- lapply(generic at signature, as.name)
assign(".SigArgs", argSyms, envir = env)
if(initialize) {
mlist <- generic at default # from 2.11.0: method, primitive or NULL,
not MethodsList
mtable <- .mlistAddToTable(generic, mlist) # by default, adds to an
empty table
assign(".MTable", mtable, envir = env)
}
else ## the current .MTable
mtable <- getMethodsForDispatch(generic)
.resetInheritedMethods(env, mtable)
if(is(generic, "groupGenericFunction")) {
for(gp in generic at groupMembers) {
gpDef <- getGeneric(gp)
if(is(gpDef, "genericFunction"))
.getMethodsTable(gpDef) # force initialization w. group methods
}
}
NULL
}
###############
.updateMethodsInTable <- function(generic, where, attach) {
fenv <- environment(generic)
reset <- identical(attach, "reset")
if(!exists(".MTable", envir = fenv, inherits = FALSE))
.setupMethodsTables(generic) ### !!!
mtable <- get(".MTable", envir = fenv)
if(!reset) {
env <- as.environment(where)
tname <- .TableMetaName(generic at generic, generic at package)
if(exists(tname, envir = env, inherits = FALSE)) {
.mergeMethodsTable(generic, mtable, get(tname, envir = env), attach)
}
## else used to warn, but the generic may be implicitly required
## by class inheritance, without any explicit methods in this package
}
if(length(generic at group)) {
groups <- as.list(generic at group)
generics <- vector("list", length(groups))
for(i in seq_along(groups))
generics[[i]] <- getGeneric(groups[[i]])
.checkGroupSigLength(groups, generics)
}
if(is(generic, "groupGenericFunction")) {
.checkGroupSigLength(list(generic at generic), list(generic))
for(g in getGroupMembers(generic))
.updateMethodsInTable(getGeneric(g), where, attach) ### !!!
}
.resetInheritedMethods(fenv, mtable)
mtable
}
Maybe Matching Threads
Wisdom of the Ancients
