Re: Stable CVS Version 1.11.13 Released!

[Mark D. Baushke]

Hi Paul,

> Looks like the mailing list to newsgroup is working
> again.  I wonder if it is still bidirectional?

I have no idea. I suppose it will depend on where the
newsgroup gets its feed.

> I submitted this fix before, but I don't think it ever made
> it to your end.  I'm resubmitting against current version...

The patch you submitted had whitespace problems. When I
manually apply it to the TRUNK sources I see that it almost
immediately (sanity.sh test 95) triggers an assertion
failure in the RCS_checkout() function on line 3972:

    assert (rev == NULL || isdigit ((unsigned char) *rev));

with rev being garbage rather than NULL. This may be directly
traced to the fact that join_rev2 is NULL, but rev2 is not NULL
and contains garbage.

(This is on a Redhat 7.3 GNU/Linux system.)

The patch, in its present state is therefore not yet ready to
be accepted into the TRUNK sources.

If/when you have another version of the patch, feel free to
let us know. It would also be highly desirable for you to
include a suggested ChangeLog entry for the change and (if
possible) either a patch to sanity.sh to test the new
functionality, or a step-by-step description of how to test
it by hand that one of us could form into a new sanity.sh
test.

        Thanks,
        -- Mark

Here is the patch I attempted to use and which I am rejecting.
(It is in unified rather than context diff format.)

Index: src/update.c
===================================================================
RCS file: /cvs/ccvs/src/update.c,v
retrieving revision 1.226
diff -u -p -r1.226 update.c
--- update.c    2 Feb 2004 21:10:28 -0000       1.226
+++ update.c    15 Feb 2004 10:33:37 -0000
@@ -2210,12 +2210,17 @@ join_file (struct file_info *finfo, Vers
            return;
        }
 
-       /* If only one join tag was specified, and the user file has
-           been changed since the greatest common ancestor (rev1),
+       /* If the file has been changed, genuinely changed,
+           in comparison to the greatest common ancestor (rev1),
            then there is a conflict we can not resolve.  See above for
            the rationale.  */
-       if (join_rev2 == NULL
-           && strcmp (rev1, vers->vn_user) != 0)
+       if ((strcmp (rev1, vers->vn_user) != 0)
+           && RCS_cmp_file(vers->srcfile,
+                           rev1,
+                           (char **)NULL,
+                           rev2,
+                           vers->options,
+                           finfo->file))
        {
            if (jdate2 != NULL)
                error (0, 0,