On Mon, Mar 28, 2011 at 09:44:34AM +0200, Maarten Billemont wrote: > I've created a series of patches for my local commits using: > > git format-patch -M -C ... > > The resulting patch files contain hunks like: > > diff --git a/file1 b/file2 > similarity index 93% > rename from file1 > rename to file2 > index 7cf3a09..88c9c13 100644 > --- a/file1 > +++ b/file2 > [...] > > git-am fails at applying these hunks, failing with: > > Applying: [commit description] > fatal: git apply: bad git-diff - inconsistent old filename on line 109 > Patch failed at [HEAD commit] I can't replicate your problem using this script: git init repo && cd repo perl -le 'print for(1 .. 100)' >file1 git add file1 git commit -m 'add file1' sed 50d file1 >file2 rm file1 git add -A git commit -m 'move file1 to file2, with tweak' git format-patch -M --stdout -1 >patch git checkout -b other HEAD^ git am patch which generates a patch like the one you mention above. Does that script work for you? If not, what git version are you using? If it does work, then perhaps there is something more complex going on in your real repo. Can you show us a more complete example, with the real filenames? > Additionally, I'd like to report that the line number reported by > git-am in this error message does not reflect the actual line number > of the problem in the patch file. The reported line number is offset > by -[amount of lines before the [---] line in the patch file], which > makes understanding the error message hard, confusing and potentially > misleading. Yes. "git am" splits the email into pieces and just feeds the patch portion to "git apply". The lines it is referencing are in .git/rebase-apply/patch. It would perhaps be useful if "git apply" gave the filename in addition to the line number (alternatively, "git am" could mention the filename after a failed apply). -Peff -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html