Re: othersrc conversion: call for RCS cleanup volunteers

[David Holland <dholland-tech%netbsd.org@localhost>]

On Mon, Jul 03, 2023 at 02:33:26PM +0200, Thomas Klausner wrote:
 > There are the following groups of problems:
 > 
 > Files with default branch, but no vendor branch
 > dist/cdk/BUGS
 >   ...
 > 
 > Disconnected revisions
 > dist/cdk/BUGS 1.1.1.1
 >   ...
 > usr.sbin/sushi/run.c 1.5.2.1
 > usr.sbin/sushi/scanform.c 1.21.2.1
 > usr.sbin/sushi/scanform.c 1.21.2.2
 > usr.sbin/sushi/scanform.c 1.21.2.3
 > usr.sbin/sushi/scanform.c 1.21.2.4
 > usr.sbin/sushi/sushi.c 1.9.2.1

The problem here seems to be that the ,v files were repo-copied from
basesrc to othersrc, and doing this seems to have caused all the tag
information to disappear. Why it would disappear, IDK. Maybe someone
deleted it on purpose.

For future reference the commits involved seem to be the following:

(1) import of cdk:
   Thu Jan  4 19:59:54 UTC 2001 by garbled with these tags:
      Vendor Tag:     CDK
      Release Tags:   CDK_4_9_9_20000407_myc3

(2) pullup of sushi/sushi.c into netbsd-1-6
   Mon Mar 29 05:35:41 UTC 2004 by jmc

(3) pullup of sushi/scanform.c into netbsd-1-6
   Mon Jul 29 14:37:56 UTC 2002 by lukem

(4) pullup of sushi/run.c into netbsd-1-6
   Mon Jul 29 14:38:48 UTC 2002 by lukem

(5) pullup of sushi/scanform.c into netbsd-1-6
   Mon Jul 29 14:39:55 UTC 2002 by lukem

(6) pullup of sushi/scanform.c into netbsd-1-6
   Mon Jul 29 14:52:08 UTC 2002

(7) pullup of sushi/scanform.c into netbsd-1-6
   Mon Jul 29 14:53:28 UTC 2002

(8) pullup of cdk/swindow.c into netbsd-3
   Mon Jun 13 22:02:36 UTC 2005 by tron

I'm sending you a diff with
   - massedit of the CDK import tags
   - addition of netbsd-1-6 and netbsd-1-6-base tags in the sushi files
   - addition of netbsd-3 and netbsd-3-base tags in cdk/swindow.c

There are other tags on these trees that could be pasted in from base,
but it's probably better not to. Although if we e.g. add netbsd-1-6
tags to sushi without later release tags and later release tags are
present elsewhere in the tree, that could cause complications because
it makes the global version graph inconsistent.

Maybe the right answer is to blow away all the pullup versions in the
othersrc copies of these trees; that history exists in base and
there's no particular reason to have it in othersrc, and trying to
keep it may create problems. (In fact, I might retcon the whole thing
by keeping just the import commit and a commit representing the
version removed from base; but that's a cleanup that would be much
easier to do after conversion.)

-- 
David A. Holland
dholland%netbsd.org@localhost