Re: c99-to-c89 patch... oops?

[Jim Meyering]

Matthew Woehlke <address@hidden> wrote:
> ...is there a reason the leading directories are inconsistent, or is
> this an oversight? Most of the patch applied with -p2 except shred.c,
> which needed -p0. This is from Jim's 6.7+ snapshot from Jan-26.

Thanks for the testing and report.
I didn't notice because the automatic patch-check
uses patch with no -p option, and that works fine.

Here's what I've done to fix it:

        * src/c99-to-c89.diff: Make shred.c Index: and a/b prefixes
        consistent, so this can be applied with patch -p0.
        Reported by Matthew Woehlke.

        * Makefile.maint (patch-check): Use patch with its -p2 option,
        since that makes this check slightly more strict.

diff --git a/Makefile.maint b/Makefile.maint
index d526baa..53b61b6 100644
--- a/Makefile.maint
+++ b/Makefile.maint
@@ -329,7 +329,7 @@ sc_useless_cpp_parens:
 patch-check:
        rm -rf src-c89 address@hidden address@hidden
        cp -a src src-c89
-       (cd src-c89; patch -V never --fuzz=0) < src/c99-to-c89.diff \
+       (cd src-c89; patch -p2 -V never --fuzz=0) < src/c99-to-c89.diff \
          > address@hidden 2>&1
        if test "$${REGEN_PATCH+set}" = set; then \
          diff -upr src src-c89 > new-diff || : ; fi
diff --git a/src/c99-to-c89.diff b/src/c99-to-c89.diff
index 04db772..997b009 100644
--- a/src/c99-to-c89.diff
+++ b/src/c99-to-c89.diff
@@ -83,13 +83,13 @@ index 364a21c..7a24014 100644
 +  }
 +  }
  }
-Index: shred.c
+Index: src/shred.c
 ===================================================================
 RCS file: /fetish/cu/src/shred.c,v
 retrieving revision 1.130
 diff -u -p -r1.130 shred.c
---- shred.c     3 Sep 2006 02:53:16 -0000       1.130
-+++ shred.c     3 Oct 2006 13:48:24 -0000
+--- a/src/shred.c     3 Sep 2006 02:53:16 -0000       1.130
++++ b/src/shred.c     3 Oct 2006 13:48:24 -0000
 @@ -464,7 +464,7 @@ dopass (int fd, char const *qname, off_t
                     out.  Thus, it shouldn't give up on bad blocks.  This
                     code works because lim is always a multiple of