I am one of the author of the R package xgboost. When I tried to pack it with R CMD build xgboost, there are 5 new compiled .o files inside src/.
I double checked that these files are not in the directory before. And strangely, this problem disappears when I run R CMD build --no-build-vignettes xgboost. I suspect there are something related to the process of compiling the vignettes. However we don't want to sacrifice the vignettes for it.
How can this be fixed? Thanks!
NOTE: To reproduce the problem, please run the following command in your console:
git clone https://github.com/tqchen/xgboost && cd xgboost && make Rpack
where make Rpack is a combination of file operations that move files to the correct places and R CMD operations. Edit the Rpack part in Makefile to control this flow.
OK after some attempts the following solution works for me:
Create a file named .Rbuildignore
Fill it with the file names (patterns) that you don't need.
An example is this file: https://github.com/dmlc/xgboost/blob/master/R-package/.Rbuildignore
Related
My .Rbuildignore specifies these files:
^inst/extdata$
This is a directory that holds some .csv files I use.
When I build my package with the "Install and Restart" button, these folders are included in the built package as demonstrated by:
> list.files(system.file(package = 'myPackage'), recursive = T, full.names = T)
...
[6] "C:/Users/JohnDoe/R/win-library/4.0/myPackage/extdata/df.csv"
...
However, if I build my package with the command devtools::install(), the inst/extdata folder is correctly ignored in the build process. This leads me to believe the discrepancy lies in the "Install and Restart" button. Is there any way to map this button to the devtools::install() command?
.Rbuildignore is used by the R CMD build command. If you install directly from a directory without building a tarball, it's not used. This is true for base R CMD INSTALL as well as the RStudio emulations of it.
So what you should do is to put the files somewhere else that isn't handled by R CMD INSTALL, for example in directory ignoreMe instead of a directory whose name means "install me".
If you don't want to do that, then it is possible to attach arbitrary R commands to keys in RStudio, but that's a lot of work. See http://rstudio.github.io/rstudioaddins/ if you want the details.
I'm exploring package building on Windows 7 but have been running into trouble after reaching the Command Prompt stage. The following thread appears to be on the same topic that I'm having trouble with, but the question remains unresolved:
Cannot install R package in Windows 7
The problem: whenever I try an R CMD command in Command Prompt (e.g. I've tried R CMD install pkgName, R cmd install pkgName, Rcmd install pkgName, where pkgName is a place-holder for an actual package that has been R CMD checked on OS X and also posted on CRAN, where I have the unzipped source in the current working directory), I get the following error message:
'R' is not recognized as an internal or external command,
operable program or batch file.
(If I try the Rcmd install pkgName command, then the error starts out as 'Rcmd' is not ....) Note that even if I type in just R in Command Prompt and hit enter, I get the same error, so I'm not even able to bring up R here, though it opens perfectly fine in its stand-alone interface.
I have examined the following guides without success in identifying the trouble:
Karl W Broman's website
Rob J Hyndman's website
Steven Mosher's blog
The most immediate thing that came to mind was that the path variables in Windows were not set right, but this also has not (yet) led to the answer. For example, I have triple-checked the paths to ensure they are correct, including for the R version number, that x64 in the path name is accurate, etc.
I realize this is a complex issue to remotely diagnose, but any advice and suggestions on resolving this trouble is appreciated. For reference, below are the steps I have taken already to try to resolve this issue:
Uninstall R and Rtools
Install the latest version of R (2.15.0)
Install Rtools215 to match R version 2.15.0
Originally I let Rtools specify the path variables (turning that option on during installation). No success after this method.
I updated the path variables on my computer, where the below paths were used:
C:\Rtools\bin; C:\Program Files\R\R-2.15.0\bin\x64;
I've also tried the R CMD (and Rcmd) commands with the following paths included (note that these paths don't lead to anything on my computer, so I tried doing R CMD / Rcmd with and without them specified):
C:\Rtools\perl\bin; C:\Rtools\MinGW\bin;
Tried adding/removing the following path, which goes to the only other folder visible within the Rtools folder:
C:\Rtools\gcc-4.6.3;
Tried multiple variations of the R CMD commands, as noted earlier.
I tried this all a few months ago when I had R version 2.14.1 and Rtools214 installed, and I had the exact same trouble.
Note: I already had Cygwin installed prior to R and Rtools, and I have not installed MikTeX (yet).
Have you closed the command prompt and opened a new one after adjusting the PATH?
Enter the command PATH on the command line and ensure that you are properly updating your path
Try navigating to the directory where R.exe is saved and running it locally -- cd C:\Program...\bin\x64, R.exe.
Ensure that you're running on Windows' Command Prompt and not Cygwin's terminal
If you actually have a space after the semi-colon in your PATH, you'll need to remove that: /bin;C:/Program.... Try combining that with Joshua's solution.
I attach my whole solution here:
Install Rtools which is compatible with my R version
Add PATH: C:\Program Files\R\R-2.15.1\bin\x64;C:\Rtools\bin;C:\Rtools\gcc-4.6.3\bin
In cmd, enter: R CMD INSTALL packagename
Click "enter", done.
I had a lot of trouble getting this to work, and finally found the golden nugget on r-project (R doesn't like spaces in paths) so here's how I finally did all this:
I installed R to "c:\programs" instead of "c:\program files...". My final path to RSCRIPT.exe is "C:\Programs\R\R-2.15.2\bin\Rscript.exe"
I added that directory to my PATH, but that doesn't seem to help
RStudio is also installed under c:\Programs (e.g. c:\Programs\RStudio), but RTools is installed under the "c:\" root directory (e.g. c:\RTools).
I downloaded all the packages to "c:\downloads\r"
I opened a command prompt in "c:\downloads\r" and ran the R CMD install from this downloads directory, specifying the path to R
e.g. "C:\Programs\R\R-2.15.2\bin\x64\R CMD INSTALL colorspace_1.2-0.zip"
It appears that the CMD processor is Case Sensitive. You have to use INSTALL vs. install
Simple solution, use quotes:
c:> "C:\Program Files\WinZip\wzunzip.exe"
if you run this, you don't need to worry about the space in Program Files.
Hope this help.
I encountered a similar problem while trying to use SparkR from Windows Power Shell. Each time I would do ".\sparkR.cmd" (I was inside the folder containing this file), it would throw an error message like,
cmd is not recognized as an internal and external
command...blah...blah...blah.
I solved the problem by adding "C:\Windows\System32" to my PATH.
I restarted my system and then repeated the process and this time, I saw a new error,
R is not recognized as an internal and external
command...blah...blah...blah.
I tried all sorts of approaches, including the ones mentioned above. NOTHING WORKED, I was probably wasn't efficient enough to make it work.
Then I came across this place at code.google. I simply copied the R.bat onto a notepad, named it as R.bat. Kept it inside the folder where sparkR.cmd was located. For me it was,
C:\Apache\spark-1.5.1-bin-hadoop2.6\bin\
The I placed it on my PATH as :
C:\Apache\spark-1.5.1-bin-hadoop2.6\bin\R.bat
Restarted my laptop, opened my Power Shell as an administrator and typed :
.\R.bat help
It worked fine. Then I typed,
.\sparkR.cmd
and there it was.
It worked.
I hope it helps to people new to R. Thank you G. Grothendieck.
In the .bat file add a line of code specifying the path of the folder where R is installed and then give the path of the script as follows -
#ECHO OFF
PATH C:\Program Files\R\R-3.5.1\bin
R CMD BATCH D:\project_abc\helloworld.R
Following this, run the .bat file and you will see the code in the script will run smoothly.
In my case the issue was solved by using Edit instead of New while selecting PATH in the window below:
enter image description here
After selecting Edit I added the directory below:
C:\Program Files\R\R-4.0.2\bin\x64
This solved the issue in my case.
Suppose I have a project for which I have developed an R package. The hierarchy might look something like this.
/project
---Makefile
---workflow.R
---test.R
---/mypackage
---DESCRIPTION
---NAMESPACE
---/R
---func1.R
---func2.R
workflow.R depends on the latest version of mypackage being installed. However, I only want to re-build the package if any file inside of it has been modified.
Currently, in my Makefile, I have:
PACKAGE=$(wildcard mypackage/**/*)
all: install test workflow
install: $(PACKAGE)
R CMD INSTALL mypackage
workflow: install
Rscript workflow.R
test: install
Rscript test.R
However, this will re-install the package every time I run make test, even if nothing inside the package has changed. Is there a clean way to avoid this?
The install rule does not create a file named install in the current directory, so make tries to remake it each time. This looks like it should be a .PHONY target, but that itself won't fix the issue as it will still execute the recipes.
One solution is to have another rule that creates a stub file:
.PHONY: all install test workflow
all: install test workflow
install: install.done
install.done: $(PACKAGE)
R CMD INSTALL mypackage
touch $#
Or you could just make install the stub file itself and make it a non-.PHONY rule.
It sounds like you want to treat the installation as an intermediate step. You can do this by adding
.INTERMEDIATE: install
to your makefile.
The make manual explains (link):
If an ordinary file b does not exist, and make considers a target that depends on b, it invariably creates b and then updates the target from b. But if b is an intermediate file, then make can leave well enough alone. It won’t bother updating b, or the ultimate target, unless some prerequisite of b is newer than that target or there is some other reason to update that target.
I am interested in providing a command line interface to an R package called Slidify that I am authoring. It uses Rscript and I think that would make it cross-platform. The scripts are stored in the subdirectory inst/slidify. In order to use the script from any directory, I added its path to my .bash_profile as I am on a Mac.
My question is
How should I handle installation of the script in an automated cross-platform way?
How can I make sure that the file permissions are retained in this process?
What should the shebang line for the script be? I am currently using
#!/usr/bin/Rscript --vanilla --slave
I would appreciate pointers on how to handle this and any examples of R packages that already do it. Just to make sure, I am clear on how this would work, a user would be able to generate a slide deck from slides.Rmd by just running slidify generate slides.Rmd from the command line.
UPDATE:
Here is how I install it on a Mac from the command line. I use the excellent sub library by 37 signals to create the scripts.
echo "$(path/to/clidir/slidify init -)" >> ~/.bash_profile
exec bash
Two follow up questions
Can I package these commands into an R function install_slidify_cli?
How can I mirror these commands for Windows users?
Lovin' slidify so would be glad to help.
But in short, you can't.
R packages simply cannot install outside of $R_HOME or the chosen library folder. Ship the script in the package, and tell users to copy it. If there was a better way, out littler package with predecessor / alternative to Rscript would long have used it, and roxygen / roxygen2 would also have shipped something.
I am writing a package in R and I would appreciate some example bash code to process the Rd files to latex and then to pdf.
It is in the directory ~/mypkg/dev/. I have generated the file structure and Rd templates.
from ~/mypkg/dev/man, I have tried
R CMD Rdconv -o mypkg-package.tex --type=latex mypkg-package.Rd
mypkg-package.tex file is generated, but
pdflatex mypkg-package.tex
generates tex without any preamble.
I have read the documentation in "Writing R extensions" and "R CMD Rdconv --help" on this subject, but no examples are provided.
Thank you
There are two issues here:
First, the Rdconv command only transforms one Rd file at a time; your question suggests that you want the full manual.
Second, the Rd2dvi command is your friend. I just ran the following on a local package:
R CMD Rd2dvi --pdf --title='Test of foo' -o /tmp/foo.pdf man/*.Rd
That should be what you asked for.
Try this. It worked for me.
found from Making an R package PDF manual using devtools
pack <- "name_of_your_package"
path <- find.package(pack)
system(paste(shQuote(file.path(R.home("bin"), "R")),"CMD", "Rd2pdf", shQuote(path)))
0. From question, it is assumed that you have .Rd files.
You may have obtained these .Rd files via roxygen package or some other ways. You need .pdf of your package. One way to do this is from Windows's command line.
1. Open Windows Command Prompt (in Administrator mode, if possible).
Start - type "cmd" - (optional: right click the appearing icon - Run as administrator)
2. Pass to the R's current working directory (it can be found via "getwd()" in R's console) in the command propmt. R's current working directory contains the folder with your package source.
cd C:\Users\erdogan\Documents\Revolution
3. For the sake of argument, say the package folder was called "causfinder". Run the following command in Windows command prompt:
(be sure that you have not got any causfinder.pdf in R's working directory (You may have obtained some R-development-incompatible causfinder.pdf with some other ways outside R). If there exists, delete causfinder.pdf first.
Otherwise, you get this error: "file 'causfinder.pdf' exists; please remove it first")
R CMD Rd2pdf causfinder/
This command performs .Rd --> LateX --> .pdf process automatically. You obtain causfinder.pdf in R's working directory.
This is further described in the manual on Writing R Extensions under the section on "Processing documentation files"
I have created a basic bash function, so I can just run rdoc from my command line. It generates the Rd files, creates the pdf, and opens it. I only use it for testing, not for creating the final documentation. You can add it by adding the following function to your .bashrc (or whichever you use) file
rdoc() {
echo -e "devtools::document()" | R --no-save
rm /tmp/rdoc.pdf
R CMD Rd2pdf -o /tmp/rdoc.pdf man/*.Rd
open /tmp/rdoc.pdf
}