I'd like to fix a package, but can't figure out how to edit it. I can download the source from R-Forge as a .tar.gz and unzip it. There is an "R" directory with the source and also a "tests" directory.
How do I include the sources in my own project to test my edits?
How do I run the tests? The tests each start with "library(blotter)". How do I make that load the library from the sources I've downloaded.
What I have done is the following (IIRC) remove the old package remove.packages. Re zip your new package to a tar.gz file again (with your changed sources). Install the new package with install.packages with pkg = path to your zipped library, and repos=NULL.
The recommended process is described in some detail in the manual 'Writing R Extensions' that came with your R installation.
There are also numerous tutorials all over the web.
I found that I could load all the sources to save me from the ridiculous repackaging by doing:
for (file in dir("../R", pattern="*.R", full.names=TRUE)) {
source(file)
}
No need to re-zip the source. Just load the package again from the source on your drive:
install.packages(/path/to/package, repos = NULL, type="source")
your method of doing the for loop to loop across R files will work in some situations but in others it may not. For example, if there is compiled non R code then looping across the *.R files may not work properly. So it's generally better to just install.packages() again.
BTW, if you clone a version control repo to your hard drive you generally don't get the zip file. So the same method is used to install.
When I'm doing testing on a package, I use a little bash script that looks like this:
#!/bin/bash
#build the package from source
R CMD build ../pkgdirectory/
#remove the old version and install the new one
R CMD REMOVE pkgname
R CMD INSTALL pkgname_0.7.tar.gz
Replace package name with the appropriate names. Save it as "make" and then run it whenever you finish a major edit. Then your testing scripts can use library(pkgname) just like normal.