2016-08-02 18:25 AEST by Arthur Barrett - If a client wants to 'get' a revision, it would be helpful for the server to provide some unique id that 
could be used.  e.g.: you 'cvs log' a file 'batman.docx'

From the log you could have an id that identifies revision 2.

Using that you are then sent brucewayne.doc

ie: you wouldn't need to know the branch, tag, path, filename, or anything - just the 'id'.

I suggest the ID would be the full RCS path and revision number, encrypted with blowfish.

The server would then:
- decrepit the id
- check it is valid (in the repo tree)
- check it exists 
- go ahead


This overcomes complexities with rename when a client wants to get an earlier revision of the same 
file but has no idea of where in the repository it was or it's old name.

This idea came from bug 7050