[R] What is the development cycle where there are code in tests/ for package development?

Peng Yu pengyu.ut at gmail.com
Wed Dec 9 23:34:30 CET 2009


I see 'library(stats)' at the beginning of
R-2.10.0/src/library/stats/tests/nls.R.

I'm wondering if I am developing my own package 'mypackage' whether I
should put 'library(mypackage)' in a .R file in mypackage/tests/? If I
do, then it seems awkward to me, because to use 'library(mypackage)',
I have to first get 'mypackage' installed.

So the development cycle is: try test cases in tests-> see bugs in
'mypackage' -> modify the code in 'mypackage' -> install
'mypackage'->try test cases in tests again....

But I think it would faster if the step of installing the package is
avoid. So instead of using 'library(mypackage)', I'd think to use
'source(some_file_in_mypackage.R)' in any file in tests/. Could
somebody let me know what is the current standard way of developing
package. Why 'library(mypackage)' rather than
'source(some_file_in_mypackage.R)' is used?




More information about the R-help mailing list