Might be a duplicate... nevertheless, I ran into this today running buildbot-try in the 
adventure branch.

Yours,
   Petr.

darcs: mergeConflictingNons failed in geteff with ix
{{
|hunk ./darcs.cabal 684
|-                   hashed-storage == 0.5.2,
|+                   hashed-storage >= 0.5.2 && < 0.6,
|:
hunk ./darcs.cabal 675
-    build-depends: base >= 3,
-                   bytestring >= 0.9.0 && < 0.10,
-                   haskeline    >= 0.6.2.2 && < 0.7,
-                   text >= 0.3,
-                   old-time   == 1.0.*,
-                   directory  == 1.0.*,
-                   process    == 1.0.*,
-                   containers >= 0.1 && < 0.4,
-                   array      >= 0.1 && < 0.4,
-                   hashed-storage >= 0.5.2 && < 0.6,
-                   random     == 1.0.*
+  build-depends: base >= 3,
+                 bytestring >= 0.9.0 && < 0.10,
+                 haskeline    >= 0.6.2.2 && < 0.7,
+                 text >= 0.3,
+                 old-time   == 1.0.*,
+                 directory  == 1.0.*,
+                 process    == 1.0.*,
+                 containers >= 0.1 && < 0.4,
+                 array      >= 0.1 && < 0.4,
+                 hashed-storage >= 0.5.2 && < 0.6,
+                 random     == 1.0.*
|:
hunk ./darcs.cabal 675
-    build-depends: base >= 3,
+    build-depends: base >= 4 && <5,
}}
xx
hunk ./darcs.cabal 675
-    build-depends: base >= 3,
+v v v v v v v
+    build-depends: base >= 4 && <5,
hunk ./darcs.cabal 687
+*************
+  build-depends: base >= 3,
+                 bytestring >= 0.9.0 && < 0.10,
+                 haskeline    >= 0.6.2.2 && < 0.7,
+                 text >= 0.3,
+                 old-time   == 1.0.*,
+                 directory  == 1.0.*,
+                 process    == 1.0.*,
+                 containers >= 0.1 && < 0.4,
+                 array      >= 0.1 && < 0.4,
+                 hashed-storage >= 0.5.2 && < 0.6,
+                 random     == 1.0.*
+^ ^ ^ ^ ^ ^ ^

Nice.  Ganesh, is this one for your shrinker?

I'll have a go when I have time. Petr, I'm not sure there's enough info 
here to reproduce easily - what state were HEAD and adventure in at the 
time?

Ganesh Sittampalam <bugs@darcs.net> writes:
> I'll have a go when I have time. Petr, I'm not sure there's enough
> info here to reproduce easily - what state were HEAD and adventure in
> at the time?
Still happens. I think getting a copy of adventure (the last bundle,
presumably), applying it and running buildbot-try (you don't need actual
buildbot. First,

darcs init
darcs pull -a --intersection ../adventure http://darcs.net

produces a conflict, even while neither of these repos is
conflicted. Apparently, the same conflict is resolved by different
patches in the two branches. When on top of this (marked, unresolved)
conflict, I pull the resolution from adventure I get the bug.

I don't get the bug if I revert the markers first.

Yours,
   Petr.

The inscrutable code for darcs-2 conflictors makes it very unlikely 
that we ever fix this.