I applied a patch (orig.mbox), resolved conflicts, then amended it. I
got a weird patch (bad.mbox).  I then unrecorded and recorded, and got
a correct patch (good.mbox).  Before unrecording, changes -v reported
the same bad hunk as is in bad.mbox.

If this bug is considered interesting, I can provide more detail.

$ darcs --version
1.0.8 (release)
$ darcs --exact-version
darcs compiled on Jul  4 2006, at 20:08:54
# configured Fri Jun 16 14:55:21 EDT 2006
./configure --no-create --no-recursion

Context:

[TAG 1.0.8
Tommy Pettersson <ptp@lysator.liu.se>**20060616160213]
-- 
Trent Buck

Attachments

• bad.mbox
• good.mbox
• orig.mbox

Please provide more detail, perhaps also a sequence of commands.  I'm having
some trouble understanding what's going on.

On Wed, Jul 18, 2007 at 06:59:07AM -0000, Eric Kow wrote:
> Please provide more detail, perhaps also a sequence of commands.
> I'm having some trouble understanding what's going on.

I thought this was clear:

>> I applied a patch (orig.mbox), resolved conflicts, then amended
>> it. I got a weird patch (bad.mbox).  I then unrecorded and
>> recorded, and got a correct patch (good.mbox).  Before unrecording,
>> changes -v reported the same bad hunk as is in bad.mbox.

Allow me to paraphrase.  Each attached .mbox file is a mbox-formatted
mail archive with a single mail message, created by "darcs send".
What follows are not the verbatim commands I used (I can't remember
them), but an approximation.

    darcs apply ../orig.darcs-patch
    editor GNUMakefile
    echo yd | darcs amend -a
    darcs send -a -o ../bad.darcs-patch
    echo yd | darcs unrecord
    darcs record -am 'Localize for office.cyber.com.au.'
    darcs send -a -o ../good.darcs-patch

If you care to compare bad and good, which should have identical
hunks, you'll see that a bunch of extra deletions and re-insertions.
-- 
Trent W. Buck

Was the patch bad / weird as in broken so that darcs complained
about it, or just odd looking?

The amend code is different from the unrecord / record code.
Amend clashes old and new changes together, and then performs
some simple optimization tricks on the result, which can
sometimes result in not completely optimized patches.

But if the patch is broken then this is truly a bug.

I honestly don't remember.  Feel free to close this bug; if it happens
again I'll keep better records.
-- 
Trent W. Buck

Ok.  Please let us know if this comes up again.