This is the output I got:

 merge (twfp): merge commute: [Failed]
*** Failed! (after 68 tests): 
mergeCommute 2 failed
x
duplicate
rotcilfnoc [
move ./Z.txt ./Us/zUY.txt
]
:
rmdir ./Us
move ./Us/zUY.txt ./Z.txt
:
rmdir ./Us
y
conflictor {{
:
move ./Z.txt ./Us/zUY.txt
:
rmdir ./Us
}} []
:
move ./Z.txt ./IX.txt
x'
duplicate
rotcilfnoc {{
:
move ./Z.txt ./IX.txt
:
move ./Z.txt ./Us/zUY.txt
}} []
:
rmdir ./Us
rotcilfnoc [
move ./Z.txt ./IX.txt
]
:
move ./Z.txt ./Us/zUY.txt
:
rmdir ./Us
y'
conflictor {{
:
move ./Z.txt ./Us/zUY.txt
:
rmdir ./Us
}} []
:
move ./Z.txt ./IX.txt
Sealed (WithStartState V1Model{ "Us" "Z.txt"["r f" "Q R" "u e" "p" "o K" 
"O b" "E P" "a z" "C" "d" "d l" "O" "w f" "q" "w P" "s" "L j" "M" "S D" 
"o" "M X" "H" "J f" "Q s"] } (TWFP 4 (ParTree (ParTree (SeqTree (DP 
(fp2fn "./Us") RmDir) NilTree) (SeqTree (DP (fp2fn "./Us") RmDir) 
NilTree)) (ParTree (SeqTree (Move (fp2fn "./Z.txt") (fp2fn 
"./Us/zUY.txt")) (SeqTree (FP (fp2fn "./Us/zUY.txt") (TokReplace "A-Za-
z_0-9" "O" "I")) NilTree)) (SeqTree (DP (fp2fn "./Us") RmDir) (SeqTree 
(Move (fp2fn "./Z.txt") (fp2fn "./IX.txt")) (SeqTree (FP (fp2fn 
"./IX.txt") (Hunk 1 ["r f","Q R","u e","p","o K","O b","E P","a 
z","C","d","d l","O","w f","q","w P","s"] ["b q"])) NilTree)))))))
(used seed 1036523552)

My operating system is Ubuntu:
3.11.0-18-generic #32-Ubuntu SMP i686 GNU/Linux.

Attached is the exact version of darcs 2.9.8.

Attachments

• exact_version.txt

I think this test may occasionally fail depending on the random cases 
generated. I'm adding Owen in case he remembers anything as I think he 
was looking at it some time ago.

In any case, we should probably just disable it to avoid confusion for 
now. I doubt anything broke it recently.

Yes, conflictors and duplicates don't play well together. I don't
remember the details well (and annoyingly can't find the bug report),
but I figured out what the issue was at the April sprint in Southampton
a few years ago.

I agree it should be disabled for now.

Ah, here it is: http://bugs.darcs.net/issue2047

I'm not sure it's exactly the same issue, but it's at least very related.

Typical RepoPatchV2 bug, unfixable.