Re: bzr repository ready?

[Karl Fogel]

Jason Earl <address@hidden> writes:
> Thanks for looking over the conversion.  That's a big deal.

So, I've found a possible problem in the conversion.  I'd like others'
opinions, though; maybe there's some explanation that has to do with
exactly how the CVS->git->bzr conversion process works.

In the CVS tree, we have this change to lisp/cus-start.el:

   ----------------------------
   revision 1.50.2.39
   date: 2008-06-12 01:34:43 -0400;  author: miles;  state: Exp; \
   lines: +0 -45;  commitid: AfSvpHSiE9VuuC6t;
   Merge from emacs--devo--0
   
   Revision: address@hidden/emacs--lexbind--0--patch-88
   ----------------------------

That's on the 'lexbind' branch.  That change was part of a commit to
many files.  I'm using a verbose cvs2cl.pl-generated ChangeLog (see
http://www.red-bean.com/kfogel/emacs-bzr-switchover/cvs2cl-ChangeLog) to
pick changes to spot-check, and it shows how the above is part of a
larger change (note we're in UTC this time):

   2008-06-12 05:34  miles

        * lisp/cus-start.el, lisp/finder.el, lisp/window.el,
        lisp/ChangeLog, lisp/bindings.el, lisp/ffap.el, lisp/menu-bar.el,
        lisp/vc-cvs.el, lisp/longlines.el, lisp/vc.el, lisp/Makefile.in,
        lisp/help-fns.el, lisp/help.el, lisp/replace.el,
        doc/misc/gnus.texi, lisp/mail/sendmail.el, lisp/nxml/nxml-mode.el,
        lisp/nxml/nxml-rap.el, lisp/nxml/nxml-util.el, doc/misc/ChangeLog,
        src/data.c, src/window.c, src/buffer.c, src/window.h, src/coding.c,
        src/ChangeLog, src/dispnew.c, src/xdisp.c, src/lisp.h,
        lisp/gnus/ChangeLog, lisp/gnus/gnus-art.el, lisp/gnus/gnus-util.el,
        lisp/gnus/nnir.el, lisp/gnus/mm-decode.el,
        lisp/language/hanja-util.el, leim/quail/hangul.el, etc/NEWS
        (lexbind.[39,19,36,224,57,29,50,33,32,68,48,57,54,65,13,32,8,5,5,\
                  20,47,92,73,21,59,203,55,140,77,138,76,37,2,39,3,12,176]):

        Merge from emacs--devo--0

        Revision: address@hidden/emacs--lexbind--0--patch-88

For those new to cvs2cl.pl: the "(lexbind.[NUMBER,NUMBER,...])" at the
end shows the branch name and then, for each listed file in order, the
revision number of file's corresponding change on that branch.  You can
see that the "39" matches the "1.50.2.39" in the 'cvs log' output, and
'cvs log' confirms that "1.50.0.2" is the branch number for "lexbind"
(ignore the intermediate ".0", it's just a CVS idiosyncracy).  You can
use 'cvs2cl.pl -b -r -t -S --utc' to generate this data yourself.

So: given that this change exists as a discrete commit in CVS, I'd
expect it to show up somewhere in the 'bzr log --long' output (see
http://www.red-bean.com/kfogel/emacs-bzr-switchover/bzr-log.out).

But it doesn't.  At least, I can't find it.  Searching for the string
"address@hidden/emacs--lexbind--0--patch-88" turns up nothing, even
though I can find similar strings from other, similar commits, such as
this one:

    ------------------------------------------------------------
    revno: 52501.1.1886
    committer: Miles Bader <address@hidden>
    timestamp: Wed 2008-01-30 07:57:28 +0000
    message:
      Merge from emacs--devo--0
      
      Revision: address@hidden/emacs--unicode--0--patch-324
    ------------------------------------------------------------

Similarly, searching for a change by Miles Bader on 2008-06-12 05:34
(UTC) turns up nothing.

Should we worry?  Is there some innocent explanation for this?

-Karl