Is the rev field in the CVSROOT/history file for "R" records wrong?

[Richard Geiger]

Hi. I've been writing a tool that examines the history file, and I've
noticed notice that, for type "R" records (a file has been deleted),
the value in the revision field is not what I was expecting (and
wanting!): the number of the newly created "dead" revision.

Rather, it seems as if the revision number of the branch point is
recorded.

For example, if I have a file on a branch with two revisions:

  1.2.2.1
  1.2.2.2

and I delete the file (thus creating the "dead" revision 1.2.2.3), the
history entry gives "1.2" as the revision in the "R" record written to
the history file. My expectation was for "1.2.2.3".

Is this a bug? (If so, I already have a fix coded and tested :-).

I'm certainly HOPING that it's a bug, and that nothing anybody knows
of depends on the would-be buggy behavior...

Thanks,

  - Richard Geiger
    rmg@foxcove.com