Re: [Rd] (PR#14181) Error description not prompted but R crashes
2010/1/12 Simon Urbanek > Even worse, this is a known bug long fixed in R 2.10.1 and hence a doubly > pointless report ... > Cheers, > Simon > > > You're right... have not seen that my daughter has a more recent version. Download 2.10.1 from KUL-Leuven mirror on my G5 and all is fine: you can close the ticket Thanks for this follow-up Phil [[alternative HTML version deleted]] __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] seq.int broken (seq as well) (PR#14169)
2010/1/11 Jens Oehlschlägel : > Petr, > This can have severe consequences like accessing beyond the limits of an > array. If C-code is involved, this can crash R. In the worst case algorithms > can silently do wrong. Being an admirer of R since its early days, I was > shocked to see this, and as a consequence, I suggest we do our homework and > suspend -- for a year or two -- any claims that R can be used productive such > as SAS. Invalid Argument Error What does SAS do in these cases? How do you know? How do you know if your SAS code isn't silently doing wrong? How do you know what caused your SAS session to crash? How do you know what algorithms SAS uses deep down inside? R gives you the source code, you can find these things out. The great irony here is that SAS have "THE POWER TO KNOW" as a registered trademark. Perhaps The R Foundation should trademark "THE POWER TO LEARN"? Barry __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Lightweight 'package' idea.
Going back a few months
I also thought "it would be nice if R had built into it some way of running
code in source packages possibly with degraded functionality to ease
development"
so building on Barry Rowlingsons start I came up with this:
---
loadDir <- function(.Root = choose.dir(getwd())){
require(tools, quietly=TRUE) # for Rd2HTML
.Package = basename(.Root) # package name defined by directory name
while(sum(search()==.Package)>0) detach(pos=which(search()==.Package)[1])
# if already attached, detach
attach(NULL, name=.Package) # attach empty environment
assign(".Root", .Root, pos=.Package) # insert .Root into .Package
# create reloadDir function and add into .Package
reloadDir <- function(.Package) { # .Package must be character
for (f in list.files(path=file.path(get(".Root", pos=.Package) , "R"),
# path is .Root/R/
pattern=".R$",full.names=TRUE,recursive=TRUE,ignore.case=TRUE))
# file type = .R
sys.source(f, envir=as.environment(.Package)) # source all such
files into .Package
invisible(.Package)} # invisibly return package name
assign(x = "reloadDir", value = reloadDir, pos=.Package)
# create help.'.Package' function and add into package
# (because function is created here, current value of .Root is in its
environment)
help.Package <- function(subj=""){# default subject is blank
if (substitute(subj)=="") subj = "*" else subj <- paste("*",
substitute(subj), "*", sep="") # get *subj*' as character
hfile <- list.files(path=file.path(.Root, "man"), # path is .Root/man/
pattern=paste(subj,"Rd$", sep="."), # file = subj.Rd
full.names=TRUE,recursive=TRUE,ignore.case=TRUE) # list of matching
files
if (length(hfile) != 1) # if not exactly one file, choose one
hfile <- choose.files(file.path(.Root, "man", paste(subj,"Rd",
sep=".")), multi=FALSE)
if(hfile != "") { # if exactly one file, open it
if(length(grep(".Rd$", hfile, ignore.case=TRUE)) ==1) { # if it's an
Rd file, create/update html
outfile <- sub("Rd$", "html", hfile, ignore.case=TRUE) # name of
corresponding html
Rd2HTML(hfile, sub("Rd$", "html", hfile, ignore.case=TRUE),
.Package) # convert Rd to html
hfile <- sub("Rd$", "html", hfile, ignore.case=TRUE)} # point to
html
shell.exec(shQuote(hfile))} # use operating system to open file of
any type
invisible(hfile)}# invisibly return help file name
assign(x = paste("help", .Package, sep="."), value=help.Package,
pos=.Package)
# use the reloadDir function to populate the environment
reloadDir(.Package)
invisible(.Package) # invisibly return .Package name
}
---
and a corresponding .Rd file
\name{loadDir}
\alias{reloadDir}
\alias{help.Package}
\title{Load an unbuilt package}
\description{
Loads code from a packages \file{\\R} subdirectory and gives access to help
files in the packages
\file{\\man} subdirectory (translating \file{.Rd} files to \file{.html}).
}
\usage{
loadDir(.Root = choose.dir(getwd()))
reloadDir(.Package)
help.'.Package'(subj="")
}
\arguments{
\item{.Root}{character, scalar. The package directory (containing
subdirectories \file{\\R} and \file{\\man}). \cr
\code{basename(.Root)} is taken as the package name
(\code{.Package}).}
\item{.Package}{character, scalar. Package name to be reloaded}
\item{subj}{character, scalar. File name to be searched for in
\file{\\man} subdirectory}
}
\details{ \describe{
\item{\code{loadDir}}{ attaches an environment at the second position on the
search list with name \code{basename(.Root)} (after detaching any existing
entries with that name). Into that environment it sources all \file{.R}
files in the \file{\\R} subdirectory, searching recursively. \cr
In that environment it also places \code{.Root} and \code{.Package
<-basename(.Root)} so that \code{get(".Root", pos=.Package)} can be used to
retrieve the original file path. \cr
In that environment it also places function \code{reloadDir} (q.v.) \cr
In that environment it also places a function named '\code{help.}' followed
by the name of the package (see \code{help.'.Package'})}
\item{\code{reloadDir}}{ re-sources all \file{.R} files in the packages
\file{\\R} subdirectory, searching recursively.}
\item{\code{help.'.Package'}}{ recursively searches the packages
\file{\\man} subdirectory for files named \code{subj}, initially searching
for files of type \file{.Rd}. \cr
If there is not exactly one such file it opens a \code{file.choose} dialog
to choose a single file of any type. \cr
If the single file is of type \file{.Rd} it is translated to a
correspondingly named \file{.html} file in the same folder, which is opened
by the operating system's file associations. \cr
If the file chosen is of any other type it is opened by the operating
system's file associations.} }}
\value{ \d
[Rd] bug in callNextMethod (PR#14185)
Hi,
there seems to be a possible bug in callNextMethod in conjunction with
the [-operator.
Relevant info, minimal example and sessionInfo follow below:
###
setClass("foo", representation = representation(a = "numeric"))
setClass("bar", contains = "foo")
setMethod(
f = "[",
signature = signature("foo"),
def = function(x,i,j,...,drop=TRUE) {
cat("drop in foo-[ :", drop, "\n")
return(1)
}
)
setMethod(
f = "[",
signature = signature("bar"),
def = function(x,i,j,...,drop=TRUE) {
cat("drop in bar-[ :", drop, "\n")
# we lose the value of drop here, when the call gets
dispatched to the super method by callNextMethod
callNextMethod()
}
)
x <- new("foo")
x[1, drop=FALSE]
# drop in foo-[ : FALSE
y <- new("bar")
y[1, drop=FALSE]
# drop in bar-[ : FALSE
# drop in foo-[ : TRUE
###
#?callNextMethod
#The statement that the method is called with the current arguments is
more precisely as follows. Arguments that were missing in the current
call are still missing (remember that "missing" is a valid class in a
method signature). For a formal argument, say x, that appears in the
original call, there is a corresponding argument in the next method call
equivalent to x = x. In effect, this means that the next method sees the
same actual arguments, but arguments are evaluated only once.
###
#S3 gets this right:
'[.foo' <- function(x, i, j, ..., drop=TRUE) {
cat("drop in foo-[ :", drop, "\n")
}
'[.bar' <- function(x, i, j, ..., drop=TRUE) {
cat("drop in bar-[ :", drop, "\n")
NextMethod()
}
x <- 1
class(x) <- c("bar", "foo")
x[1, drop=FALSE]
# drop in bar-[ : FALSE
# drop in foo-[ : FALSE
###
> sessionInfo()
R version 2.11.0 Under development (unstable) (2010-01-12 r50970)
i386-pc-mingw32
locale:
[1] LC_COLLATE=German_Germany.1252 LC_CTYPE=German_Germany.1252
[3] LC_MONETARY=German_Germany.1252 LC_NUMERIC=C
[5] LC_TIME=German_Germany.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
loaded via a namespace (and not attached):
[1] tools_2.11.0
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Is this "strftime" behavior (on WinXP) what's expected?
WinXP, R version 2.10.1:
> s1<- strftime("2009/12/15 11:55:00 PM", format="%Y/%m/%d %I:%M:%S %p",
> tz="GMT", usetz=FALSE)
> print(s1)
[1] "2009/12/15 11:55:00 AM"
> s2<- strftime("2009/12/16 12:00:00 AM", format="%Y/%m/%d %I:%M:%S %p",
> tz="GMT", usetz=FALSE)
> print(s2)
[1] "2009/12/16 12:00:00 PM"
> s3<- strftime("2009/12/16 00:00:00 AM", format="%Y/%m/%d %I:%M:%S %p",
> tz="GMT", usetz=FALSE)
> print(s3)
[1] "2009/12/16 12:00:00 AM"
>
--
Jan Galkowski, 617.444.4995
""Eppur si muove." --Galilei
[[alternative HTML version deleted]]
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Rd output garbled in some circumstances
I'm having trouble getting correct help output in some circumstances for
a package I've created. Though this is not an issue with the current R,
I would like my package to work with previous ones as well.
I'm looking for suggestions about how I could rework my .Rd file so that
it will work with prior R's. In particular, R 2.7 is in the latest
stable release of Debian, so I'd like to solve the problem for 2.7.
The .Rd file is for a function and has an arguments section like this
\arguments{
\item{formula}{ A formula giving the vectors containing
## skipped
covariates. }
## skipped
\item{stepdenominator}{See \code{stepnumerator} just above.}
\item{do.what}{\describe{
\item{1}{By default, calculates a maximimum likelihood. To evaluate
a single likelihood, set all parameters to fixed. }
\item{0}{Count number of paths and related statistics without
evaluating the likelihood.}
\item{-1}{Get detailed counts (but no likelihoods) associated with
each case. The return value is a matrix.}
\item{10}{Use the model to generate a random path for each
case. returning a \code{data.frame} with simulated observed states
and times and all other data as observed.}
}}
\item{testing}{This argument is only for use by developers. Set it
## etc
This comes out fine in a pdf, but ?mspath (the function) produces, in
part,
stepdenominator: See 'stepnumerator' just above.
1 By default, calculates a maximimum likelihood. To evaluate a
single likelihood, set all parameters to fixed.
0 Count number of paths and related statistics without evaluating
the likelihood.
in R 2.7. The "do.what" header has vanished. In R 2.10 it's fine.
Is there an error in my documentation format?
Even if not, is there some change I could make that would get R 2.7 to
work better?
The R change log doesn't show anything obviously related to this, though
it has several references to unspecified fixes to the documentation
system. I also tried looking at the bug tracker, but couldn't find
anything--in fact I had trouble identifying bugs in the documentation
system as opposed to bugs in the documentation.
Thanks.
Ross Boylan
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] bug in callNextMethod (PR#14185)
Yes, a bug prevented callNextMethod() from detecting the special case of
the `[` call, for which it did have code.
Should be fixed in the current 2.11.0 (r 50976).
Thanks for the helpful bug report.
John Chambers
bernd_bis...@gmx.net wrote:
Hi,
there seems to be a possible bug in callNextMethod in conjunction with
the [-operator.
Relevant info, minimal example and sessionInfo follow below:
###
setClass("foo", representation = representation(a = "numeric"))
setClass("bar", contains = "foo")
setMethod(
f = "[",
signature = signature("foo"),
def = function(x,i,j,...,drop=TRUE) {
cat("drop in foo-[ :", drop, "\n")
return(1)
}
)
setMethod(
f = "[",
signature = signature("bar"),
def = function(x,i,j,...,drop=TRUE) {
cat("drop in bar-[ :", drop, "\n")
# we lose the value of drop here, when the call gets
dispatched to the super method by callNextMethod
callNextMethod()
}
)
x <- new("foo")
x[1, drop=FALSE]
# drop in foo-[ : FALSE
y <- new("bar")
y[1, drop=FALSE]
# drop in bar-[ : FALSE
# drop in foo-[ : TRUE
###
#?callNextMethod
#The statement that the method is called with the current arguments is
more precisely as follows. Arguments that were missing in the current
call are still missing (remember that "missing" is a valid class in a
method signature). For a formal argument, say x, that appears in the
original call, there is a corresponding argument in the next method call
equivalent to x = x. In effect, this means that the next method sees the
same actual arguments, but arguments are evaluated only once.
###
#S3 gets this right:
'[.foo' <- function(x, i, j, ..., drop=TRUE) {
cat("drop in foo-[ :", drop, "\n")
}
'[.bar' <- function(x, i, j, ..., drop=TRUE) {
cat("drop in bar-[ :", drop, "\n")
NextMethod()
}
x <- 1
class(x) <- c("bar", "foo")
x[1, drop=FALSE]
# drop in bar-[ : FALSE
# drop in foo-[ : FALSE
###
> sessionInfo()
R version 2.11.0 Under development (unstable) (2010-01-12 r50970)
i386-pc-mingw32
locale:
[1] LC_COLLATE=German_Germany.1252 LC_CTYPE=German_Germany.1252
[3] LC_MONETARY=German_Germany.1252 LC_NUMERIC=C
[5] LC_TIME=German_Germany.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
loaded via a namespace (and not attached):
[1] tools_2.11.0
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Rd output garbled in some circumstances
On 13/01/2010 6:15 PM, Ross Boylan wrote:
I'm having trouble getting correct help output in some circumstances for
a package I've created. Though this is not an issue with the current R,
I would like my package to work with previous ones as well.
I'm looking for suggestions about how I could rework my .Rd file so that
it will work with prior R's. In particular, R 2.7 is in the latest
stable release of Debian, so I'd like to solve the problem for 2.7.
The .Rd file is for a function and has an arguments section like this
\arguments{
\item{formula}{ A formula giving the vectors containing
## skipped
covariates. }
## skipped
\item{stepdenominator}{See \code{stepnumerator} just above.}
\item{do.what}{\describe{
\item{1}{By default, calculates a maximimum likelihood. To evaluate
a single likelihood, set all parameters to fixed. }
\item{0}{Count number of paths and related statistics without
evaluating the likelihood.}
\item{-1}{Get detailed counts (but no likelihoods) associated with
each case. The return value is a matrix.}
\item{10}{Use the model to generate a random path for each
case. returning a \code{data.frame} with simulated observed states
and times and all other data as observed.}
}}
\item{testing}{This argument is only for use by developers. Set it
## etc
This comes out fine in a pdf, but ?mspath (the function) produces, in
part,
stepdenominator: See 'stepnumerator' just above.
1 By default, calculates a maximimum likelihood. To evaluate a
single likelihood, set all parameters to fixed.
0 Count number of paths and related statistics without evaluating
the likelihood.
in R 2.7. The "do.what" header has vanished. In R 2.10 it's fine.
Is there an error in my documentation format?
Even if not, is there some change I could make that would get R 2.7 to
work better?
I would avoid nesting the \describe within \arguments. Both basically
use the same formatting code, and 2.7 probably doesn't support nesting
properly. There was no real parser there, just a fallible pattern
matching approach.
A better solution is to say your package requires a recent version of R,
but maybe that's not feasible for you.
Duncan Murdoch
The R change log doesn't show anything obviously related to this, though
it has several references to unspecified fixes to the documentation
system. I also tried looking at the bug tracker, but couldn't find
anything--in fact I had trouble identifying bugs in the documentation
system as opposed to bugs in the documentation.
Thanks.
Ross Boylan
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Mksetup() limited to hashing with 32 bits
The MKsetup() in unique.c throws an error if the vector to be hashed is
longer than (2^32)/8:
if(n < 0 || n > 536870912) /* protect against overflow to -ve */
error(_("length %d is too large for hashing"), n);
I occasionally work with vectors longer than this on 64-bit builds. Would it be
too much to ask that R can take advantage of all 64 bits for hashing when
compiled as such?
Thanks
Ben
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
