Created attachment 1747 [details] R CMD check toto_1.0.tar.gz reproduces the issue [Sorry for reporting this bug again but my original bug report (https://bugs.r-project.org/bugzilla/show_bug.cgi?id=16153) was closed (NOT REPRODUCIBLE) before I was given a chance to provide a reproducible example (I did provide a reproducible example after the bug was already closed but the system would only let me change the status to RESOLVED or VERIFIED so I didn't change it). 8 days have passed now, and it looks unlikely that someone will re-open the issue. So here it is again.] In the context of refactoring a function, some arguments might be added, dropped, and/or renamed. 'R CMD check' has always been a great tool to help keep the documentation in sync with the code after this type of change. I just ran into the following glitch though: documented arguments that are not in \usage are usually detected but it seems that ... is not. Running 'R CMD check toto_1.0.tar.gz' (package attached) reproduces the issue. The toto() function (defined in the package) has only 1 arg: > args(toto) function (x) NULL In the man page for the toto() function, I've added entries for non-existing args 'y' and '...': \arguments{ \item{x}{} \item{y}{} \item{...}{} } but 'R CMD check toto_1.0.tar.gz' only complains about 'y' not being in \usage: * checking Rd \usage sections ... WARNING Documented arguments not in \usage in documentation object 'toto': ‘y’ It says nothing about '...' not being in \usage. Thanks, H.
Yes I can confirm that I tried this and there was no complaint for `...` but there was for another parameter that was not in the signature.
Should be fixed in the trunk with c77755.