If you have a repo with thousands of files and only plan to change 5 or 6, branching the whole repository of files would mean having to re-integrate a large number of files other than those you changed.
I would prefer to create a branch, and then selectively update my 5 or 6 files to that branch and then merge those 5 or 6 files back, greatly reducing my merge headache.
So, should I simply rtag the repo (which is done without reference to a working copy and without making anything immediately sticky) and then cvs update -r $rtag_name on a file-by-file basis?
And then when I'm ready to checkin. I can move to trunk via cvs update -A and then merge from there via cvs update -j $rtag_name ?