On VCS-independent ways of identifying commits, and why they are desirable.
authorEric S. Raymond <esr@thyrsus.com>
Thu, 16 Jan 2014 05:50:26 +0000 (00:50 -0500)
committerEric S. Raymond <esr@thyrsus.com>
Thu, 16 Jan 2014 05:50:26 +0000 (00:50 -0500)
No code changes.

Makefile.in
admin/ChangeLog
admin/notes/commits

index e69998e..852a134 100644 (file)
@@ -385,6 +385,7 @@ src: Makefile FRC
        cd $@ || exit;                                                  \
        boot=bootstrap-emacs$(EXEEXT);                                  \
        [ ! -x "$$boot" ] || boot='';                                   \
+       echo "VCSWITNESS: $$vcswitness";\
        $(MAKE) all $(MFLAGS)                                           \
          CC='${CC}' CFLAGS='${CFLAGS}' CPPFLAGS='${CPPFLAGS}'          \
          LDFLAGS='${LDFLAGS}' MAKE='${MAKE}' BOOTSTRAPEMACS="$$boot"   \
index 36a10fa..111d644 100644 (file)
@@ -1,3 +1,8 @@
+2014-01-16  Eric S. Raymond  <esr@thyrsus.com>
+
+       * notes/commits: Add a 'graph on VCS-independent ways of
+       identifying commits and the desirability thereof.
+
 2014-01-15  Paul Eggert  <eggert@cs.ucla.edu>
 
        Fix copyright license notices for Adobe Unicode mapping files.
index 2c6f80c..6e2aabe 100644 (file)
@@ -45,6 +45,15 @@ Date:         Tue, 31 Mar 2009 12:21:20 +0900
     for modern source-control systems with a global log, it's better to
     have something like "Regenerate configure".
 
+(4) (Added in 2014) In commit comments, and ChangLog files, it is best
+    to use ways of identifying revisions that are not dependent on a
+    particular version control system.  (At time of writing Emacs is
+    about to move to its third VCS and another move in the future is
+    not impossible.)  An excellent way to identify commits is by
+    quoting their summary line.  Another is with an action stamp - an
+    RFC3339 date followed by ! followed by the committer's email - for
+    example, "2014-01-16T05:43:35Z!esr@thyrsus.com". Often, "my
+    previous commit" will suffice.
 
 Followup discussion:
 http://lists.gnu.org/archive/html/emacs-devel/2010-01/msg00897.html