
1. A common case is that a patch has been merged in Linus' tree.
   'pushpatch' will then fail, but 'pushpatch -m' invokes wiggle, which
   realizes that nothing needs to be applied. When you then run refpatch,
   you end up with an empty patch, and you would like to do poppatch and
   delete the patch from 'series', but poppatch thinks something is wrong
   and refuses to pop the empty patch.

2. Wiggle is quite optimistic when applying patches. Most of the time, it
   gets things right or creates conflict markers, but occasionally it does
   the wrong thing without any warning messages being generated. One
   example is the ppc64-sections patch, which adds the new file
   include/asm-ppc64/sections.h. The same file has also been added in
   -bk6, but is slightly different. In this case wiggle decided to
   concatenate the two versions and also included one line from the diff
   header in the ouput. No warning was generated by wiggle.

   The same thing happened with the do_div-comment patch. Linus' comment
   and your comment were both added to the file without any warning from
   wiggle. I guess the moral is that you can't trust wiggle to always to
   the right thing, so you have to manually check the files it
   touches. pushpatch -m now reports which files are processed by wiggle.

3. import_patch has problems with patch-2.6.0-test2-bk6.bz2, because the
   patch deletes a line in sk98lin.txt that begins with
   '-- CheckLinkState', so the patch looks like '--- CheckLinkState' which
   confuses touched-by-patch and makes the .pc file wrong.

