this is with two branches of a darcs1 repo that had many unresolved 
conflicts before it has been converted to darcs2.

i've made sure that i pull all the patches (that i could) into a merged 
darcs1 repo, then used darcs 2.3.1 to convert this merge repo to darcs2, 
and then i reconstructed the two darcs2 branches from the darcs2 version 
of the merge repo using unpull and applying+recording patches.

darcs convert in 2.4 became very slow, that's why i've used 2.3.1 for the 
conversion.

unfortunately the repo is not public. it contains unicode filenames and 
file contents.

i've tried with 2.3.1, but it behaved the same.


alendvai@ed101:/tmp/nafi$ darcs --version
2.4 (release)
alendvai@ed101:/tmp$ darcs get ~/workspace/nafi/
Copying patches, to get lazy repository hit ctrl-C...
Finished getting.                                             
alendvai@ed101:/tmp$ darcs get /opt/ebr42-nafi/workspace/ebr42/
Copying patches, to get lazy repository hit ctrl-C...
Finished getting.                                             
alendvai@ed101:/tmp$ cd nafi/                                 
alendvai@ed101:/tmp/nafi$ darcs push ../ebr42/
The remote repository has 3 patches to pull.
Fri Feb 12 15:13:35 CET 2010  tomi.borbely@gmail.com
  * new slots in támogatási-kifizetés class
Shall I push this patch? (1/120)  [ynWsfvplxdaqjk], or ? for help: q
Push cancelled.
alendvai@ed101:/tmp/nafi$ darcs pull ../ebr42/
Pulling from "/tmp/ebr42"...
Wed Mar 10 17:32:14 CET 2010  tomi.borbely@gmail.com
  * fixek a 2. negyedévi támogatások számításához
Shall I pull this patch? (1/3)  [ynWvplxdaqjk], or ? for help: n
Wed Mar 10 19:35:41 CET 2010  tomi.borbely@gmail.com
  * notes
Shall I pull this patch? (2/3)  [ynWsfvplxdaqjk], or ? for help: n
Thu Mar 11 14:06:04 CET 2010  attila.lendvai@gmail.com
  * felmeres persistent singletons are exported
Shall I pull this patch? (3/3)  [ynWsfvplxdaqjk], or ? for help: y
Backing up ./source/model/közoktatási-egyedi-megállapodás.lisp(-darcs-
backup0)
We have conflicts in the following files:
./source/model/közoktatási-egyedi-megállapodás.lisp
Finished pulling and applying.
alendvai@ed101:/tmp/nafi$ darcs pull ../ebr42/
Pulling from "/tmp/ebr42"...
darcs: Inconsistent patch:
conflictor [
hunk ./source/model/közoktatási-egyedi-megállapodás.lisp 13
-                ,@(iter (for entry :in entries)
-                        (collect (bind (((name-part &optional (variable-
name-part name-part)) (ensure-list entry))
-                                        (name (format nil "Adatgyűjtés a 
közoktatási egyedi megállapodásokhoz (~A)" name-part))
-                                        (constant-name (symbolicate 
'#:+közoktatási-egyedi-megállapodások-felmérése/ (string-upcase variable-
name-part) "+"))
-                                        (variable-name (symbolicate 
'#:*közoktatási-egyedi-megállapodások-felmérése/ (string-upcase variable-
name-part) "*")))
-                                   `(progn
-                                      (def (constant e :test #'equal)
-                                          ,constant-name ,name)
-                                      (def singleton-persistent-instance 
,variable-name
-                                          (aprog1
-                                              (select-felmérés :név 
,name)
-                                            (assert it))))))))))
-  (x "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december"
-     ("I. negyedév"   1.negyedév)
-     ("II. negyedév"  2.negyedév)
-     ("III. negyedév" 3.negyedév)
-     ("IV. negyedév"  4.negyedév)
-     "beszámoló"))
+                ,@(iter (for name :in names)
+                        (collect `(def (constant e :test #'equal)
+                                      ,(format-symbol *package* 
"+KÖZOKTATÁSI-EGYEDI-MEGÁLLAPODÁSOK-FELMÉRÉSE/~A+" (string-upcase name))
+                                      ,(format nil "Adatgyűjtés a 
közoktatási egyedi megállapodásokhoz (~A)" (string-downcase name))))))))
+  (x "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december" 
"beszámoló"))
hunk ./source/model/közoktatási-egyedi-megállapodás.lisp 13
-                ,@(iter (for name :in names)
-                        (collect `(def (constant e :test #'equal)
-                                      ,(format-symbol *package* 
"+KÖZOKTATÁSI-EGYEDI-MEGÁLLAPODÁSOK-FELMÉRÉSE/~A+" (string-upcase name))
-                                      ,(format nil "Adatgyűjtés a 
közoktatási egyedi megállapodásokhoz (~A)" (string-downcase name))))))))
-  (x "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december" 
"beszámoló"))
+                ,@(iter (for entry :in entries)
+                        (collect (bind (((name-part &optional (variable-
name-part name-part)) (ensure-list entry))
+                                        (name (format nil "Adatgyűjtés a 
közoktatási egyedi megállapodásokhoz (~A)" name-part))
+                                        (constant-name (symbolicate 
'#:+közoktatási-egyedi-megállapodások-felmérése/ (string-upcase variable-
name-part) "+"))
+                                        (variable-name (symbolicate 
'#:*közoktatási-egyedi-megállapodások-felmérése/ (string-upcase variable-
name-part) "*")))
+                                   `(progn
+                                      (def (constant e :test #'equal)
+                                          ,constant-name ,name)
+                                      (def singleton-persistent-instance 
,variable-name
+                                          (aprog1
+                                              (select-felmérés :név 
,name)
+                                            (assert it))))))))))
+  (frob "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december"
+        ("I. negyedév"   1.negyedév)
+        ("II. negyedév"  2.negyedév)
+        ("III. negyedév" 3.negyedév)
+        ("IV. negyedév"  4.negyedév)
+        "beszámoló"))
]
|:
hunk ./source/model/k�zoktat�si-egyedi-meg�llapod�s.lisp 21
-                                      (def singleton-persistent-instance 
,variable-name
+                                      (def (singleton-persistent-instance 
e) ,variable-name
m doesn't conflict with mm in is_consistent
alendvai@ed101:/tmp/nafi$ darcs check
The repository is consistent!
alendvai@ed101:/tmp/nafi$ darcs check
The repository is consistent!                                                   
alendvai@ed101:/tmp/nafi$ darcs show repo
          Type: darcs
        Format: hashed, darcs-2
          Root: /tmp/nafi
      Pristine: HashedPristine
         Cache: thisrepo:/tmp/nafi, cache:/home/alendvai/.darcs/cache, 
cache:/home/ati/.darcs/cache, repo:/home/alendvai/workspace/nafi, 
repo:/home/alendvai/Desktop/conversion/merge.without-deadly-
conflicts.darcs2
boringfile Pref: .boring
Default Remote: /tmp/ebr42
   Num Patches: 3004
alendvai@ed101:/tmp/nafi$

output of pulling with --debug-verbose

alendvai@ed101:/tmp/nafi$ darcs pull --debug-verbose 
Beginning identifying repository .
Done identifying repository .
Identified darcs-2 repo: /tmp/nafi
Beginning identifying repository /tmp/ebr42
Done identifying repository /tmp/ebr42
Pulling from "/tmp/ebr42"...
Reading patch file: Wed Mar  3 12:28:05 CET 2010  
attila.lendvai@gmail.com
  tagged ebr42-final
I'm doing copyFileUsingCache on patches/0000128939-
c834503164efed88da7a3507bce46eb82553056cf0db062e10a2c31853a3af9c
I'm doing copyFileUsingCache on inventories/0000217653-
e256a27c4997adc59802f25a69b61e057f31d9de91aab560aa08aa1003998629
Reading patch file: Wed Mar 10 18:25:31 CET 2010  
attila.lendvai@gmail.com
  * don't use *nafi...*
I'm doing copyFileUsingCache on patches/0000000432-
1711f59cae461ef7deefe46974a42f67d93d6cd68e0f472fcaf39c3b20621579
Reading patch file: Mon Mar  8 10:54:32 CET 2010  
attila.lendvai@gmail.com
  * update .boring, new darcs follows symlinks
I'm doing copyFileUsingCache on patches/0000000195-
85491ca53b37f8a00206e874062b3f9d4517cbc36a4665a10fba168d159ef930
Reading patch file: Mon Mar  8 15:23:39 CET 2010  tomi.borbely@gmail.com
  * kifizetés-szétválasztás: special case for pászok
I'm doing copyFileUsingCache on patches/0000002122-
68ab9c12880e220b0dea7dc01ec8cd41a790e6a23572206cba8e931981fee43e
Reading patch file: Wed Mar 10 10:08:58 CET 2010  tomi.borbely@gmail.com
  * márciusi kifizetések csv
I'm doing copyFileUsingCache on patches/0000010744-
4b20412909f48ded89a323aea25c0337b84a3e2575e820bcb20266b0b8d3e4f9
Reading patch file: Wed Mar 10 11:31:47 CET 2010  tomi.borbely@gmail.com
  * integrált oktatásért alapítvány plusz kifizetése
I'm doing copyFileUsingCache on patches/0000000304-
b14879895f4c1cb1093e64539a5bfb05ab1284675e20d577bb13c150fecf701b
Reading patch file: Thu Mar 11 14:06:04 CET 2010  
attila.lendvai@gmail.com
  * felmeres persistent singletons are exported
I'm doing copyFileUsingCache on patches/0000004501-
db71db76fede6b96a62d343967e4a150d37dbdabe441a2719a9899a0133242fa
darcs: Inconsistent patch:
conflictor [
hunk ./source/model/közoktatási-egyedi-megállapodás.lisp 13
-                ,@(iter (for entry :in entries)
-                        (collect (bind (((name-part &optional 
(variable-name-part name-part)) (ensure-list entry))
-                                        (name (format nil "Adatgyűjtés 
a közoktatási egyedi megállapodásokhoz (~A)" name-part))
-                                        (constant-name (symbolicate 
'#:+közoktatási-egyedi-megállapodások-felmérése/ (string-upcase 
variable-name-part) "+"))
-                                        (variable-name (symbolicate 
'#:*közoktatási-egyedi-megállapodások-felmérése/ (string-upcase 
variable-name-part) "*")))
-                                   `(progn
-                                      (def (constant e :test #'equal)
-                                          ,constant-name ,name)
-                                      (def singleton-persistent-
instance ,variable-name
-                                          (aprog1
-                                              (select-felmérés :név 
,name)
-                                            (assert it))))))))))
-  (x "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december"
-     ("I. negyedév"   1.negyedév)
-     ("II. negyedév"  2.negyedév)
-     ("III. negyedév" 3.negyedév)
-     ("IV. negyedév"  4.negyedév)
-     "beszámoló"))
+                ,@(iter (for name :in names)
+                        (collect `(def (constant e :test #'equal)
+                                      ,(format-symbol *package* 
"+KÖZOKTATÁSI-EGYEDI-MEGÁLLAPODÁSOK-FELMÉRÉSE/~A+" (string-upcase name))
+                                      ,(format nil "Adatgyűjtés a 
közoktatási egyedi megállapodásokhoz (~A)" (string-downcase name))))))))
+  (x "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december" 
"beszámoló"))
hunk ./source/model/közoktatási-egyedi-megállapodás.lisp 13
-                ,@(iter (for name :in names)
-                        (collect `(def (constant e :test #'equal)
-                                      ,(format-symbol *package* 
"+KÖZOKTATÁSI-EGYEDI-MEGÁLLAPODÁSOK-FELMÉRÉSE/~A+" (string-upcase name))
-                                      ,(format nil "Adatgyűjtés a 
közoktatási egyedi megállapodásokhoz (~A)" (string-downcase name))))))))
-  (x "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december" 
"beszámoló"))
+                ,@(iter (for entry :in entries)
+                        (collect (bind (((name-part &optional 
(variable-name-part name-part)) (ensure-list entry))
+                                        (name (format nil "Adatgyűjtés 
a közoktatási egyedi megállapodásokhoz (~A)" name-part))
+                                        (constant-name (symbolicate 
'#:+közoktatási-egyedi-megállapodások-felmérése/ (string-upcase 
variable-name-part) "+"))
+                                        (variable-name (symbolicate 
'#:*közoktatási-egyedi-megállapodások-felmérése/ (string-upcase 
variable-name-part) "*")))
+                                   `(progn
+                                      (def (constant e :test #'equal)
+                                          ,constant-name ,name)
+                                      (def singleton-persistent-
instance ,variable-name
+                                          (aprog1
+                                              (select-felmérés :név 
,name)
+                                            (assert it))))))))))
+  (frob "terv" "január" "február" "március" "április" "május" "június" 
"július" "augusztus" "szeptember" "október" "november" "december"
+        ("I. negyedév"   1.negyedév)
+        ("II. negyedév"  2.negyedév)
+        ("III. negyedév" 3.negyedév)
+        ("IV. negyedév"  4.negyedév)
+        "beszámoló"))
]
|:
hunk ./source/model/k�zoktat�si-egyedi-meg�llapod�s.lisp 21
-                                      (def singleton-persistent-
instance ,variable-name
+                                      (def (singleton-persistent-
instance e) ,variable-name
m doesn't conflict with mm in is_consistent

this might be relevant: after the darcs2 conversion i haven't cleaned my 
~/.darcs/cache but i put the darcs2 repos to the exact same location 
where the darcs1 repos were before.

err, it has nothing to do with darcs convert. i suspect it's much more due to 
something like having non-ascii in file names that are being removed and later 
conflicted.

the following happens with two branches of a freshly initialized darcs2 repo that 
i've created after i gave up on darcs convert-ing (see earlier messages in this 
bug report):

what happens below: there are two branches: old and new. i try to pull over a 
patch from old into new. this patch changes a file in old that has been deleted in 
new. pulling marks the conflict. trying to pull again from the old repo ends up in 
the assert both with 2.3.1 and 2.4.





alendvai@ed101:/tmp/ebr42.new-branch$ darcs push ../ebr42.old-branch/
Pushing to "/tmp/ebr42.old-branch"...
Fri Mar 12 15:47:16 CET 2010  attila.lendvai@gmail.com
  * rename src to source
Shall I push this patch? (1/38)  [ynWvplxdaqjk], or ? for help: q
Push cancelled.
alendvai@ed101:/tmp/ebr42.new-branch$ darcs pull ../ebr42.old-branch/
Pulling from "/tmp/ebr42.old-branch"...
Tue Mar 16 12:01:35 CET 2010  tomi.borbely@gmail.com
  * fix beszámoló pdf
Shall I pull this patch? (1/5)  [ynWsfvplxdaqjk], or ? for help: v
[fix beszámoló pdf
tomi.borbely@gmail.com**20100316110135
 Ignore-this: b136ac646a51ed2a84f2ff01e187fc79
] hunk ./src/presentation/kitöltés.lisp 963
-                                                                             
(list (list fenntartó) validity current-time)))))
+                                                                             
(list (list fenntartó) (list *gyökér-támogatási-jogcím/egyedi-megállapodások*) 
validity current-time)))))
Shall I pull this patch? (1/5)  [ynWsfvplxdaqjk], or ? for help: y
Fri Mar 19 09:59:34 CET 2010  attila.lendvai@gmail.com
  * renames around támogatási-kifizetés/átutalt-támogatási-összeg
Shall I pull this patch? (2/5)  [ynWvplxdaqjk], or ? for help: d
We have conflicts in the following files:
./source/presentation/kitöltés.lisp
Finished pulling and applying.
alendvai@ed101:/tmp/ebr42.new-branch$ darcs w -l
M ./source/presentation/kitöltés.lisp -1 +1
alendvai@ed101:/tmp/ebr42.new-branch$ darcs pull ../ebr42.old-branch/
Pulling from "/tmp/ebr42.old-branch"...
darcs: Inconsistent patch:
conflictor [
hunk ./source/presentation/kitöltés.lisp 1
-(in-package :ebr42-presentation)
-
-;;;;;;
-;;; naturális mutató kitöltés title component
-

[**** loads of line deletion skipped here ****]

-                                                                                  
(név-of it))
-                                                                        :titulus 
(vezetői-titulus *felmérési-alany*)
-                                                                        
:intézmény (név-of *felmérési-alany*)))))))))
rmfile ./source/presentation/kitöltés.lisp
]
|:
hunk ./source/presentation/kit�lt�s.lisp 963
-                                                                             
(list (list fenntartó) validity current-time)))))
+                                                                             
(list (list fenntartó) (list *gyökér-támogatási-jogcím/egyedi-megállapodások*) 
validity current-time)))))
m doesn't conflict with mm in is_consistent








after reverting and unpulling the patch:

alendvai@ed101:/tmp/ebr42.new-branch$ ~/bin/darcs-2.4 check
The repository is consistent!
Hash mismatch(es)!                           
source/presentation
    index: 8f54cfa7b1f8e51b1289394451e27edf26c505c67d5774ed6dbd2353f970852d
  working: 618508e7a997b332df6194b6a1b9be388f247f843cfea8fb78e5d888b8cbf736
source
    index: a910343fe1fafaa78e968cad0a631f220b7ee9eff7a7d84040ec9d99ba1e452d
  working: 452e74e53d3b1675cc83357be71212fb7942a77b179b205e1c14fb991a119863

Bad index.




which is another bug i've reported and might be related: 
http://bugs.darcs.net/issue1756

as far as i can see 2.3.1 ad 2.4 behaves exactly the same, including the bad index 
(except that only 2.4 detects the bad index).

hope this helps!

Adding Petr because of the index mismatch

This seems to have to do with the fn2ps and ps2fn stuff. Darcs encodes
filenames in UTF-8 when it serializes them to the patch file, and it
seems to do too much or too little of that at some point.

This seems to have to do with the fn2ps and ps2fn stuff. Darcs encodes
filenames in UTF-8 when it serializes them to the patch file, and it
seems to do too much or too little of that at some point.

this is causing us regular headaches... :/

any hope for a fix?

I did a little bit of darcs changes research during the sprint.  In case
it helps, here are the results:

Check out also the issue706 patch below.

 darcs changes src/Darcs/Patch/Prim.lhs --match 'hunk NewFormat' --reverse

No context in
  http://lists.osuosl.org/pipermail/darcs-devel/2008-January/

maybe the month before

Changes to src/Darcs/Patch/Prim.lhs:

Thu Jan 10 23:04:13 CET 2008  David Roundy <droundy@darcs.net>
  * make darcs-2 repositories store patches in more-nicely-formated format

Tue Feb 12 20:44:56 CET 2008  David Roundy <droundy@darcs.net>
  * add new "chunk" patch format (currently only used in darcs-2
repositories).

Thu Feb 14 18:15:58 CET 2008  David Roundy <droundy@darcs.net>
  * roll back changes in hunk handling.
  These changes introduced a bug in the reading of patch bundles, and I
don't
  have time to track that bug down, so I'm rolling them back wholesale.
  Right now, we just don't have time for new regressions.

  rolling back:

  Sun Feb 10 17:30:44 EST 2008  David Roundy <droundy@darcs.net>
    * rewrite hunks to store solid chunks rather than lists of lines.

      M ./src/Darcs/Patch/Apply.lhs -51 +48
      M ./src/Darcs/Patch/Commute.lhs -5 +5
      M ./src/Darcs/Patch/Prim.lhs -54 +50
      M ./src/Darcs/Patch/Viewing.lhs -13 +13
      M ./src/FastPackedString.hs -24 +3
      M ./src/OldFastPackedString.hs -23 +55
  Sun Feb 10 18:15:46 EST 2008  David Roundy <droundy@darcs.net>
    * fix bug in new code that broke darcs check on darcs' own repo.

      M ./src/Darcs/Patch/Apply.lhs -1 +3
  Tue Feb 12 14:44:56 EST 2008  David Roundy <droundy@darcs.net>
    * add new "chunk" patch format (currently only used in darcs-2
repositories).

      M ./src/Darcs/Patch/Prim.lhs -2 +32
      M ./src/Darcs/Patch/Read.lhs -1 +25

Tue Mar  4 17:16:15 CET 2008  David Roundy <droundy@darcs.net>
  * resolved issue706: encode spaces in filenames for darcs-2 format.

The bug is assigned to me, but I'm not really available ATM. If anyone
would like to take over, we'd appreciate it!

fyi, i've managed to compile a "2.1.0 (+ 70 patches)" which fails the 
failing-issue1763-pull-fails-on-non-ascii-filenames.sh the same way.

compiling earlier versions is a headache due to too fresh ghc on my 
ubuntu.

The following patch makes the test pass. It may work as a stopgap 
measure, it's however unlikely to be free of side effects... Please do 
not apply this yet without further research.

diff -rN -u -p old-mainline/src/Darcs/Patch/Prim.lhs new-
mainline/src/Darcs/Patch/Prim.lhs
--- old-mainline/src/Darcs/Patch/Prim.lhs	2010-05-27 
13:08:38.000000000 +0200
+++ new-mainline/src/Darcs/Patch/Prim.lhs	2010-05-27 
13:08:45.000000000 +0200
@@ -242,8 +242,8 @@ instance Show (Prim C(x y))  where
 
 data FileNameFormat = OldFormat | NewFormat
 formatFileName :: FileNameFormat -> FileName -> Doc
-formatFileName OldFormat = packedString . fn2ps
-formatFileName NewFormat = text . encodeWhite . fn2fp
+formatFileName _ = packedString . fn2ps
+-- formatFileName NewFormat = text . encodeWhite . fn2fp
 
 showPrim :: FileNameFormat -> Prim C(a b) -> Doc
 showPrim x (FP f AddFile) = showAddFile x f
diff -rN -u -p old-mainline/src/Darcs/Patch/Read.hs new-
mainline/src/Darcs/Patch/Read.hs
--- old-mainline/src/Darcs/Patch/Read.hs	2010-05-27 
13:08:38.000000000 +0200
+++ new-mainline/src/Darcs/Patch/Read.hs	2010-05-27 
13:08:45.000000000 +0200
@@ -113,8 +113,8 @@ readSplit x = do
   return $ Split `mapSeal` ps
 
 readFileName :: FileNameFormat -> B.ByteString -> FileName
-readFileName OldFormat = ps2fn
-readFileName NewFormat = fp2fn . decodeWhite . BC.unpack
+readFileName _ = ps2fn
+-- readFileName OldFormat = fp2fn . decodeWhite . BC.unpack
 
 readHunk :: ParserM m => FileNameFormat -> m (Prim C(x y))
 readHunk x = do

Forget what I just said. I have tracked down the bug and there's a 
(hopefully; needs review) proper fix now, see patch252. Due to simplicity 
of the patch, I nominate this for a 2.4 point-release.

The following patch updated the status of issue1763 to be resolved:

* Resolve issue1763: use correct filename encoding in conflictors. 
Ignore-this: 62d98022e62edf926d601ba3074eaf0b

The following patch updated issue issue1763 with status=resolved;resolvedin=2.5.0 (current)

* Resolve issue1763: use correct filename encoding in conflictors. 
Ignore-this: 62d98022e62edf926d601ba3074eaf0b

Note that the test currently fails on MacOS X, but this appears to be
due to issue1880.

Further notes: issue1880 also fails on Linux, so I've not got the right 
diagnosis exactly.

We need to run the test on MacOS X by hand, study the output and 
determine the first spot where the test is doing something unexpected.  
The buildbot output is making me wonder if it's just something stupid 
like assumptions about the encoding of the shell script itself.

I don't think this should be open? (I don't see any failures in tests 
related to this anywhere.) Eric if you disagree please reopen.

Huh! the test passes on the Mac buildslave.  I'm a bit confused by that, 
but I guess I'll have to open a new ticket if further investigation 
reveals something.