devtools::check() is a wonderful tool, but it can take a lot of time on big packages.
If I have an error in one step, I often would like to check this step only, without having to compute all the previous ones.
For instance, I have this annoying no visible binding for global variable 'xxx' note when running the checking R code for possible problems step. I'm not sure which call of xxx is causing the note, so I would love to run only this step.
On this particular matter, I tried codetools::checkUsagePackage("mypackage") but there are far too many false positives so that is not CRAN-like.
Is there any way to run a single step of devtools::check() at a time?
Only a partial answer. You can set some of the arguments of devtools::check to FALSE to skip some specific checks/parts and you can also pass arguments to the underlying R CMD check (via args) to skip even more.
Have a look at the R CMD check utility help for the parameters you can pass to it:
$ R CMD check --help
Usage: R CMD check [options] pkgs
Check R packages from package sources, which can be directories or
package 'tar' archives with extension '.tar.gz', '.tar.bz2',
'.tar.xz' or '.tgz'.
A variety of diagnostic checks on directory structure, index and
control files are performed. The package is installed into the log
directory and production of the package PDF manual is tested.
All examples and tests provided by the package are tested to see if
they run successfully. By default code in the vignettes is tested,
as is re-building the vignette PDFs.
Options:
-h, --help print short help message and exit
-v, --version print version info and exit
-l, --library=LIB library directory used for test installation
of packages (default is outdir)
-o, --output=DIR directory for output, default is current directory.
Logfiles, R output, etc. will be placed in 'pkg.Rcheck'
in this directory, where 'pkg' is the name of the
checked package
--no-clean do not clean 'outdir' before using it
--no-codoc do not check for code/documentation mismatches
--no-examples do not run the examples in the Rd files
--no-install skip installation and associated tests
--no-tests do not run code in 'tests' subdirectory
--no-manual do not produce the PDF manual
--no-vignettes do not run R code in vignettes nor build outputs
--no-build-vignettes do not build vignette outputs
--ignore-vignettes skip all tests on vignettes
--run-dontrun do run \dontrun sections in the Rd files
--run-donttest do run \donttest sections in the Rd files
--use-gct use 'gctorture(TRUE)' when running examples/tests
--use-valgrind use 'valgrind' when running examples/tests/vignettes
--timings record timings for examples
--install-args= command-line args to be passed to INSTALL
--test-dir= look in this subdirectory for test scripts (default tests)
--no-stop-on-test-error do not stop running tests after first error
--check-subdirs=default|yes|no
run checks on the package subdirectories
(default is yes for a tarball, no otherwise)
--as-cran select customizations similar to those used
for CRAN incoming checking
The following options apply where sub-architectures are in use:
--extra-arch do only runtime tests needed for an additional
sub-architecture.
--multiarch do runtime tests on all installed sub-archs
--no-multiarch do runtime tests only on the main sub-architecture
--force-multiarch run tests on all sub-archs even for packages
with no compiled code
By default, all test sections are turned on.
Report bugs at <https://bugs.R-project.org>.
Related
I want to run valgrind on the tests, examples and vignettes of my package. Various sources insinuate that the way to do this should be:
R CMD build my-pkg
R CMD check --use-valgrind my-pkg_0.0.tar.gz
R CMD check seems to run fine, but shows no evidence of valgrind output, even after setting the environment variable VALGRIND_OPTS: --memcheck:leak-check=full. I've found sources that hint that R needs to run interactively for valgrind to show output, but R -d CMD check (or R -d "CMD check") seems to be the wrong format.
R -d "valgrind --tool=memcheck --leak-check=full" --vanilla < my-pkg.Rcheck/my-pkg-Ex.R does work, but only on the example files; I can't see a simple way to run this against my vignettes and testthat tests.
What is the best way to run all relevant scripts through valgrind? For what it's worth, the goal is to integrate this in a GitHub actions script.
Edit Mar 2022: The R CMD check case is actually simpler, running R CMD check --use-valgrind [other options you may want] will run the tests and examples under valgrind and then append the standard valgrind summary at the end of the examples output (i.e., pkg.Rcheck/pkg-Ex.Rout) and test output (i.e., pkg.Rcheck/tinytest.Rout as I use tinytest)_. What puzzles me now is that an error detected by valgrind does not seem to fail the test.
Original answer below the separator.
There is a bit more to this: you helps to ensure that the R build is instrumented for it. See Section 4.3.2 of Writing R Extensions:
On platforms where valgrind is installed you can build a version of R with extra instrumentation to help valgrind detect errors in the use of memory allocated from the R heap. The configure option is --with-valgrind-instrumentation=level, where level is 0, 1 or 2. Level 0 is the default and does not add anything. Level 1 will detect some uses117 of uninitialised memory and has little impact on speed (compared to level 0). Level 2 will detect many other memory-use bugs118 but make R much slower when running under valgrind. Using this in conjunction with gctorture can be even more effective (and even slower).
So you probably want to build yourself a Docker-ized version of R to call from your GitHub Action. I think the excellent 'sumo' container by Winston has a valgrind build as well so you could try that as well. It's huge at over 4gb:
edd#rob:~$ docker images| grep wch # some whitespace edit out
wch1/r-debug latest a88fabe8ec81 8 days ago 4.49GB
edd#rob:~$
And of course if you test dependent packages you have to get them into the valgrind session too...
Unfortunately, I found the learning curve involved in dockerizing R in GitHub actions, per #dirk-eddelbuettel's suggestion, too steep.
I came up with the hacky approach of adding a file memcheck.R to the root of the package with the contents
devtools::load_all()
devtools::run_examples()
devtools::build_vignettes()
devtools::test()
(remembering to add to .Rbuildignore).
Running R -d "valgrind --tool=memcheck --leak-check=full" --vanilla < memcheck.R then seems to work, albeit with the reporting of some issues that appear to be false positives, or at least issues that are not identified by CRAN's valgrind implementation.
Here's an example of this in action in a GitHub actions script.
(Readers who know what they are doing are invited to suggest shortcomings of this approach in the comments!)
I'm writing a package containing several functions to make running and evaluating models more streamlined.
I have a function that I'm going to make the first function within my package detailed with roxygen2 comments, which I can include into this write-up as an edit if necessary, but my issue is more with Package Creation.
I've created a separate .R file for the function and it lives within the R folder in within my package folder. I've run R CMD build pkgname and R CMD INSTALL pkgname successfully.
At the document() stage I run it (from console or whether in my terminal using R -e 'library(devtools);document()', deleting the existing NAMESPACE file first) and I get the following error: Try removing ‘/Library/Frameworks/R.framework/Versions/ 3.5/Resources/library/00LOCK-pkgname.
I've already seen the [issue posted here][1] and haven't had success after deleting the 00LOCK-pkgname folder, for two reasons: when I run document(), even when it throws the above error, it doesn't stop running, it just keeps looping (that happens whether I run this in R or use the Terminal). Additionally, no matter how many times I delete the folder, it keeps re-appearing even though I've stopped running the function.
Any insight into why that error is being thrown and the document() function continually runs in a loop?
Best answer I've found is in this blog post: Hilary Parker R-Package Blog Post
The steps I follow to document and install are as follows:
Within the project that contains my package, open a new R Script and run setwd('..')
Run devtools::document()
Run devtools::install()
This works for me when initially installing my package and also updating it.
I want to use testthat for integration tests in an R package being installed into a Jupyter notebook environment. Practically, that means I don't want the tests run when the package is installed, but rather manually later, when the system is running.
I think that means I shouldn't put tests into a tests/testthat directory, but rather the R directory.
So, I still want all the expect_ functions, but how do I run a test? I'm not sure if I can specify a file or directory, because it will be after the package is installed, and I don't know what the current working directory or install directory will be.
From the testthat package . . .
If you're using testthat in a package, you should put your tests in
tests/testthat. Each test file should start with test and end in .R or
.r. To ensure R CMD check runs your tests, place the following code in
tests/testthat.R:
library(testthat) ;
library(yourpackage)
test_check("yourpackage")
It sounds like you should follow all the setup but leave the testthat.R file out of the directory. You can run the test_check("yourpackage") from the console or a script.
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 have installed Rtools v.31 and MKTeX2.9 on my laptop before building my R package. After I ran R CMD check, a directory called mypackage.Rcheck and the source file mypackage_1.x.tar.gz appeared, and while RStudio build package function is checking everything, there is no warning or error message, but at the end when it said (as follows), the corresponding mypackage.Rcheck directory which is supposed to contain the PDF manual disappeared:
checking PDF version of manual ... OK
DONE
NOTE: There were 3 notes.
See
'C:/.../package/mypackage.Rcheck/00check.log'
for details.
R CMD check succeeded
I thought it might be a version conflict between Rtools and R because I'm using R13.1.2 and Rtools v.31, where on the http://cran.r-project.org/bin/windows/Rtools/ website, it said the image isn't frozen (which I don't understand what it means). But I tried all the suggested compatible version pairs between Rtools and R, and still couldn't get it working. So I think this disappearing of mypackage.RCheck is not due to version conflict.
I don't think it's MiKTeX problem because apparently pdf is generated, but was consequently removed. I also tried using --options in R CMD check, but there is no way to force generation of pdf manual, plus pdf was generated at one point. I also read one post on this website, suggesting the use of Rutils function Rd2pdf: "R CMD Rd2pdf mypackage", but I can't run in on windows, and Rutils package is not available for R13.1.2.
can someone suggest the source of the problem or if they have seen this problem before and was able to fix it?
In RStudio: Tools -> Global Options -> Packages. There is a setting "Cleanup output after successful R CMD Check". If this is checked it removes the package.Rcheck folder after a successful check. If you uncheck this setting, the package.Rcheck folder is not removed and you will find the reference manual pdf there.
A common problem for mysterious disappearances of files from R tarballs is to accidentally specify that they should be ignored in the .Rbuildignore file. The lines in that file are regular expressions, so leaving off the anchors ^ and $ can cause them to match more files than you intended. For example,
vig.e
matches vignette.pdf (because there are no anchors to the start and end of the name, and the
dot means "any character".