[Rd] barplot can put legend in wrong place, request option to override that (PR#13265)

2008-11-07 Thread timhesterberg
This in an enhancement requestion, to add the capability to pass arguments through barplot() to legend(). I've created a version that does this. # Run this code; notice that the legend # is superposed over one bar: barplot(height = cbind(x = c(465,91) / 465 * 100, y = c(840,

[Rd] format.data.frame handles names in nested data frames badly (PR#12624)

2008-08-27 Thread timhesterberg
> d <- data.frame(x=1:3, y=2:4) > d[["r"]] <- d[1]/d[2] > d x y x 1 1 2 0.500 2 2 3 0.667 3 3 4 0.750 It appears that the name "x" was given instead of "r". That is not correct: > names(d) [1] "x" "y" "r" > showStructure(d) # from package splus2R list[3,3] S3 class: data.f

[Rd] save() should not overwrite a file if an error occurs (PR#12583)

2008-08-22 Thread timhesterberg
If save() fails because an object is not found, it should not overwrite an existing file. > a <- 1:9 > save(a, file = "a.rda") > rm(a) > load("a.rda") > a [1] 1 2 3 4 5 6 7 8 9 > rm(a) > save(a, file = "a.rda") Error in save(a, file = "a.rda") : object 'a' not found > load("a.rda") Error

[Rd] When a tests/*.R file is run, the package should automatically be loaded (PR#11950)

2008-08-07 Thread timhesterberg
Suppose that a package "foo" contains a tests directory with a number of *.R files. Currently, each of the .R files must begin with library("foo") or the library is not loaded. I suggest that the package be loaded automatically (at the beginning of each file, or before any of the files are ru

[Rd] browser() exits when not desired, in three different scenarios (PR#11623)

2008-06-10 Thread timhesterberg
Here are three different scenarios in which the browser() exits when I don't want it to. These are all related to the browser accepting either c or (return) to exit. I would like an option to turn off the (return) behavior in order to avoid these problems. # Case 1 - when calling help # Note,

[Rd] enhancement request for browser (PR#8706)

2006-03-24 Thread TimHesterberg
I would like a way to modify the behavior of browser(), so that entering a blank line does nothing rather than being equivalent to "c". I'm running R using emacs ESS. I often debug functions by inserting a browser, and stepping through one line at a time, sending a line at a time from a buffer wit