% Copyright (C) 2002-2003 David Roundy % % This program is free software; you can redistribute it and/or modify % it under the terms of the GNU General Public License as published by % the Free Software Foundation; either version 2, or (at your option) % any later version. % % This program is distributed in the hope that it will be useful, % but WITHOUT ANY WARRANTY; without even the implied warranty of % MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the % GNU General Public License for more details. % % You should have received a copy of the GNU General Public License % along with this program; see the file COPYING. If not, write to % the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, % Boston, MA 02110-1301, USA. \begin{code}
{-# OPTIONS_GHC -cpp -fglasgow-exts -fno-warn-orphans #-}
-- , TypeOperators #-}

#include "gadts.h"

module Darcs.Patch.Commute ( fromPrims,
                             merge, elegantMerge,
                             merger, unravel,
                             publicUnravel, mangleUnravelled,
                             CommuteFunction, Perhaps(..),
                             -- for other commutes:

import Control.Monad ( MonadPlus, mplus, msum, mzero, guard )

import Darcs.Patch.FileName ( FileName, fn2fp, fp2fn )
import Darcs.Patch.Info ( invertName, idpatchinfo )
import Darcs.Patch.Patchy ( Commute(..), Invert(..), toFwdCommute, toRevCommute )
import Darcs.Patch.Core ( Patch(..), Named(..),
                          joinPatchesFL )
import Darcs.Patch.Prim ( Prim(..), FromPrims(..),
                          Conflict(..), Effect(..),
                          is_filepatch, sortCoalesceFL,
                          FilePatchType(..), DirPatchType(..),
                          primIsHunk, modernizePrim )
import qualified Data.ByteString.Char8 as BC (pack, last)
import qualified Data.ByteString       as B (null, ByteString)
import Data.Maybe ( isJust )
import Data.List ( intersperse, sort )
import Darcs.Patch.Permutations ( headPermutationsRL, simpleHeadPermutationsFL )
import Printer ( text, vcat, ($$) )
import Darcs.Patch.Patchy ( invertRL )
import Darcs.Patch.Show ( showPatch_ )
import Data.List ( nub, nubBy )
import Darcs.Witnesses.Sealed ( unsafeUnseal, unsafeUnsealFlipped )
import Darcs.Utils ( nubsort )
#include "impossible.h"
import Darcs.Witnesses.Sealed ( Sealed(..), mapSeal, unseal, FlippedSeal(..), mapFlipped, unsealFlipped )
import Darcs.Witnesses.Ordered ( mapFL, mapFL_FL, unsafeCoerceP,
                             unsafeCoercePStart, unsafeCoercePEnd,
                             FL(..), RL(..),
                             (:/\:)(..), (:<)(..), (:\/:)(..), (:>)(..),
                             lengthFL, mapRL,
                             reverseFL, reverseRL, concatFL,
                             MyEq, unsafeCompare, EqCheck(..), (=\/=)

--import Darcs.ColorPrinter ( traceDoc )
--import Printer ( greenText )
\end{code} \section{Commuting patches} \subsection{Composite patches} Composite patches are made up of a series of patches intended to be applied sequentially. They are represented by a list of patches, with the first patch in the list being applied first. \newcommand{\commutex}{\longleftrightarrow} \newcommand{\commutes}{\longleftrightarrow} The first way (of only two) to change the context of a patch is by commutation, which is the process of changing the order of two sequential patches. \begin{dfn} The commutation of patches $P_1$ and $P_2$ is represented by \[ P_2 P_1 \commutes {P_1}' {P_2}'. \] Here $P_1'$ is intended to describe the same change as $P_1$, with the only difference being that $P_1'$ is applied after $P_2'$ rather than before $P_2$. \end{dfn} The above definition is obviously rather vague, the reason being that what is the ``same change'' has not been defined, and we simply assume (and hope) that the code's view of what is the ``same change'' will match those of its human users. The `$\commutes$' operator should be read as something like the $==$ operator in C, indicating that the right hand side performs identical changes to the left hand side, but the two patches are in reversed order. When read in this manner, it is clear that commutation must be a reversible process, and indeed this means that commutation \emph{can} fail, and must fail in certain cases. For example, the creation and deletion of the same file cannot be commuted. When two patches fail to commutex, it is said that the second patch depends on the first, meaning that it must have the first patch in its context (remembering that the context of a patch is a set of patches, which is how we represent a tree). \footnote{The fact that commutation can fail makes a huge difference in the whole patch formalism. It may be possible to create a formalism in which commutation always succeeds, with the result of what would otherwise be a commutation that fails being something like a virtual particle (which can violate conservation of energy), and it may be that such a formalism would allow strict mathematical proofs (whereas those used in the current formalism are mostly only hand waving ``physicist'' proofs). However, I'm not sure how you'd deal with a request to delete a file that has not yet been created, for example. Obviously you'd need to create some kind of antifile, which would annihilate with the file when that file finally got created, but I'm not entirely sure how I'd go about doing this. $\ddot\frown$ So I'm sticking with my hand waving formalism.} %I should add that one using the inversion relationship of sequential %patches, one can avoid having to provide redundant definitions of %commutation. % There is another interesting property which is that a commutex's results % can't be affected by commuting another thingamabopper. \begin{code}
data Perhaps a = Unknown | Failed | Succeeded a

instance  Monad Perhaps where
    (Succeeded x) >>= k =  k x
    Failed   >>= _      =  Failed
    Unknown  >>= _      =  Unknown
    Failed   >> _       =  Failed
    (Succeeded _) >> k  =  k
    Unknown  >> k       =  k
    return              =  Succeeded
    fail _              =  Unknown

instance  MonadPlus Perhaps where
    mzero                 = Unknown
    Unknown `mplus` ys    = ys
    Failed  `mplus` _     = Failed
    (Succeeded x) `mplus` _ = Succeeded x

toMaybe :: Perhaps a -> Maybe a
toMaybe (Succeeded x) = Just x
toMaybe _ = Nothing

toPerhaps :: Maybe a -> Perhaps a
toPerhaps (Just x) = Succeeded x
toPerhaps Nothing = Failed

cleverCommute :: CommuteFunction -> CommuteFunction
cleverCommute c (p1:<p2) =
    case c (p1 :< p2) of
    Succeeded x -> Succeeded x
    Failed -> Failed
    Unknown -> case c (invert p2 :< invert p1) of
               Succeeded (p1' :< p2') -> Succeeded (invert p2' :< invert p1')
               Failed -> Failed
               Unknown -> Unknown

speedyCommute :: CommuteFunction
speedyCommute (p1 :< p2) -- Deal with common case quickly!
    | p1_modifies /= Nothing && p2_modifies /= Nothing &&
      p1_modifies /= p2_modifies = Succeeded (unsafeCoerceP p2 :< unsafeCoerceP p1)
    | otherwise = Unknown
    where p1_modifies = isFilepatchMerger p1
          p2_modifies = isFilepatchMerger p2

instance Commute p => Commute (Named p) where
    commute (NamedP n1 d1 p1 :> NamedP n2 d2 p2) =
        if n2 `elem` d1 || n1 `elem` d2
        then Nothing
        else do (p2' :> p1') <- commute (p1 :> p2)
                return (NamedP n2 d2 p2' :> NamedP n1 d1 p1')
    merge (NamedP n1 d1 p1 :\/: NamedP n2 d2 p2)
        = case merge (p1 :\/: p2) of
          (p2' :/\: p1') -> NamedP n2 d2 p2' :/\: NamedP n1 d1 p1'
    listTouchedFiles (NamedP _ _ p) = listTouchedFiles p
    hunkMatches f (NamedP _ _ p) = hunkMatches f p

instance Conflict p => Conflict (Named p) where
    listConflictedFiles (NamedP _ _ p) = listConflictedFiles p
    resolveConflicts (NamedP _ _ p) = resolveConflicts p
    isInconsistent (NamedP _ _ p) = isInconsistent p

everythingElseCommute :: MaybeCommute -> CommuteFunction
everythingElseCommute c x = eec x
    eec :: CommuteFunction
    eec (PP px :< PP py) = toPerhaps $ do x' :> y' <- commute (py :> px)
                                          return (PP y' :< PP x')
    eec (ComP NilFL :< p1) = Succeeded (unsafeCoerceP p1 :< (ComP NilFL))
    eec (p2 :< ComP NilFL) = Succeeded (ComP NilFL :< unsafeCoerceP p2)
    eec (ComP (p:>:ps) :< p1) = toPerhaps $ do
                              (p1' :< p') <- c (p :< p1)
                              (p1'' :< ComP ps') <- c (ComP ps :< p1')
                              return (p1'' :< ComP (p':>:ps'))
    eec (patch2 :< ComP patches) =
        toPerhaps $ do (patches' :< patch2') <- ccr (patch2 :< reverseFL patches)
                       return (ComP (reverseRL patches') :< patch2')
        where ccr :: FORALL(x y) (Patch :< RL Patch) C(x y) -> Maybe ((RL Patch :< Patch) C(x y))
              ccr (p2 :< NilRL) = seq p2 $ return (NilRL :< p2)
              ccr (p2 :< p:<:ps) = do (p' :< p2') <- c (p2 :< p)
                                      (ps' :< p2'') <- ccr (p2' :< ps)
                                      return (p':<:ps' :< p2'')
    eec _xx =
        msum [
              cleverCommute commuteRecursiveMerger       _xx
             ,cleverCommute otherCommuteRecursiveMerger _xx

Note that it must be true that

commutex (A^-1 A, P) = Just (P, A'^-1 A')


if commutex (A, B) == Just (B', A')
then commutex (B^-1, A^-1) == Just (A'^-1, B'^-1)

unsafeMerger :: String -> Patch C(x y) -> Patch C(x z) -> Patch C(a b)
unsafeMerger x p1 p2 = unsafeCoercePStart $ unsafeUnseal $ merger x p1 p2

mergerCommute :: (Patch :< Patch) C(x y) -> Perhaps ((Patch :< Patch) C(x y))
mergerCommute (Merger _ _ p1 p2 :< pA)
    | unsafeCompare pA p1 = Succeeded (unsafeMerger "0.0" p2 p1 :< unsafeCoercePStart p2)
    | unsafeCompare pA (invert (unsafeMerger "0.0" p2 p1)) = Failed
mergerCommute (Merger _ _
                (Merger _ _ c b)
                (Merger _ _ c' a) :<
                Merger _ _ b' c'')
    | unsafeCompare b' b && unsafeCompare c c' && unsafeCompare c c'' =
        Succeeded (unsafeMerger "0.0" (unsafeMerger "0.0" b (unsafeCoercePStart a)) (unsafeMerger "0.0" b c) :<
                   unsafeMerger "0.0" b (unsafeCoercePStart a))
mergerCommute _ = Unknown

instance Commute Patch where
    merge (y :\/: z) =
        case actualMerge (y:\/:z) of
        Sealed y' -> case commute (z :> y') of
                         Nothing -> bugDoc $ text "merge_patches bug"
                                    $$ showPatch_ y
                                   $$ showPatch_ z
                                   $$ showPatch_ y'
                         Just (_ :> z') -> -- actualMerge returns one arm of a
                                           -- merge result, so commuting then gives
                                           -- us the other arm but we have to assert
                                           -- that the starting context is correct
                                           unsafeCoercePStart z' :/\: y'
    commute x = toMaybe $ msum
                  [toFwdCommute speedyCommute x,
                   toFwdCommute (cleverCommute mergerCommute) x,
                   toFwdCommute (everythingElseCommute (toRevCommute commute)) x
    -- Recurse on everything, these are potentially spoofed patches
    listTouchedFiles (ComP ps) = nubsort $ concat $ mapFL listTouchedFiles ps
    listTouchedFiles (Merger _ _ p1 p2) = nubsort $ listTouchedFiles p1
                                            ++ listTouchedFiles p2
    listTouchedFiles c@(Regrem _ _ _ _) = listTouchedFiles $ invert c
    listTouchedFiles (PP p) = listTouchedFiles p

    hunkMatches f (ComP ps) = or $ mapFL (hunkMatches f) ps
    hunkMatches f (Merger _ _ p1 p2) = hunkMatches f p1 || hunkMatches f p2
    hunkMatches f c@(Regrem _ _ _ _) = hunkMatches f $ invert c
    hunkMatches f (PP p) = hunkMatches f p

commuteNoMerger :: MaybeCommute
commuteNoMerger x =
    toMaybe $ msum [speedyCommute x,
                    everythingElseCommute commuteNoMerger x]

isFilepatchMerger :: Patch C(x y) -> Maybe FileName
isFilepatchMerger (PP p) = is_filepatch p
isFilepatchMerger (Merger _ _ p1 p2) = do
     f1 <- isFilepatchMerger p1
     f2 <- isFilepatchMerger p2
     if f1 == f2 then return f1 else Nothing
isFilepatchMerger (Regrem und unw p1 p2)
    = isFilepatchMerger (Merger und unw p1 p2)
isFilepatchMerger (ComP _) = Nothing

commuteRecursiveMerger :: (Patch :< Patch) C(x y) -> Perhaps ((Patch :< Patch) C(x y))
commuteRecursiveMerger (p@(Merger _ _ p1 p2) :< pA) = toPerhaps $
  do (_ :> pA') <- commute (pA :> undo)
     commute (pA' :> invert undo)
     (_ :> pAmid) <- commute (pA :> unsafeCoercePStart (invert p1))
     (p1' :> pAx) <- commute (pAmid :> p1)
     guard (pAx `unsafeCompare` pA)
     (p2' :> _) <- commute (pAmid :> p2)
     (p2o :> _) <- commute (invert pAmid :> p2')
     guard (p2o `unsafeCompare` p2)
     let p' = if unsafeCompare p1' p1 && unsafeCompare p2' p2
              then unsafeCoerceP p
              else unsafeMerger "0.0" p1' p2'
         undo' = mergerUndo p'
     (pAo :> _) <- commute (undo' :> pA')
     guard (pAo `unsafeCompare` pA)
     return (pA' :< p')
    where undo = mergerUndo p
commuteRecursiveMerger _ = Unknown

otherCommuteRecursiveMerger :: (Patch :< Patch) C(x y) -> Perhaps ((Patch :< Patch) C(x y))
otherCommuteRecursiveMerger (pA':< p_old@(Merger _ _ p1' p2')) =
  toPerhaps $
  do (pA :> _) <- commute (mergerUndo p_old :> pA')
     (pAmid :> p1) <- commute (unsafeCoercePEnd p1' :> pA)
     (_ :> pAmido) <- commute (pA :> invert p1)
     guard (pAmido `unsafeCompare` pAmid)
     (p2 :> _) <- commute (invert pAmid :> p2')
     (p2o' :> _) <- commute (pAmid :> p2)
     guard (p2o' `unsafeCompare` p2')
     let p = if p1 `unsafeCompare` p1' && p2 `unsafeCompare` p2'
             then unsafeCoerceP p_old
             else unsafeMerger "0.0" p1 p2
         undo = mergerUndo p
     guard (not $ pA `unsafeCompare` p1) -- special case here...
     (_ :> pAo') <- commute (pA :> undo)
     guard (pAo' `unsafeCompare` pA')
     return (p :< pA)
otherCommuteRecursiveMerger _ = Unknown

type CommuteFunction = FORALL(x y) (Patch :< Patch) C(x y) -> Perhaps ((Patch :< Patch) C(x y))
type MaybeCommute = FORALL(x y) (Patch :< Patch) C(x y) -> Maybe ((Patch :< Patch) C(x y))
\end{code} \paragraph{Merge} \newcommand{\merge}{\Longrightarrow} The second way one can change the context of a patch is by a {\bf merge} operation. A merge is an operation that takes two parallel patches and gives a pair of sequential patches. The merge operation is represented by the arrow ``\( \merge \)''. \begin{dfn}\label{merge_dfn} The result of a merge of two patches, $P_1$ and $P_2$ is one of two patches, $P_1'$ and $P_2'$, which satisfy the relationship: \[ P_2 \parallel P_1 \merge {P_2}' P_1 \commutex {P_1}' P_2. \] \end{dfn} Note that the sequential patches resulting from a merge are \emph{required} to commutex. This is an important consideration, as without it most of the manipulations we would like to perform would not be possible. The other important fact is that a merge \emph{cannot fail}. Naively, those two requirements seem contradictory. In reality, what it means is that the result of a merge may be a patch which is much more complex than any we have yet considered\footnote{Alas, I don't know how to prove that the two constraints even \emph{can} be satisfied. The best I have been able to do is to believe that they can be satisfied, and to be unable to find an case in which my implementation fails to satisfy them. These two requirements are the foundation of the entire theory of patches (have you been counting how many foundations it has?).}. \subsection{How merges are actually performed} The constraint that any two compatible patches (patches which can successfully be applied to the same tree) can be merged is actually quite difficult to apply. The above merge constraints also imply that the result of a series of merges must be independent of the order of the merges. So I'm putting a whole section here for the interested to see what algorithms I use to actually perform the merges (as this is pretty close to being the most difficult part of the code). The first case is that in which the two merges don't actually conflict, but don't trivially merge either (e.g.\ hunk patches on the same file, where the line number has to be shifted as they are merged). This kind of merge can actually be very elegantly dealt with using only commutation and inversion. There is a handy little theorem which is immensely useful when trying to merge two patches. \begin{thm}\label{merge_thm} $ P_2' P_1 \commutex P_1' P_2 $ if and only if $ P_1'^{ -1} P_2' \commutex P_2 P_1^{ -1} $, provided both commutations succeed. If either commutex fails, this theorem does not apply. \end{thm} This can easily be proven by multiplying both sides of the first commutation by $P_1'^{ -1}$ on the left, and by $P_1^{ -1}$ on the right. Besides being used in merging, this theorem is also useful in the recursive commutations of mergers. From Theorem~\ref{merge_thm}, we see that the merge of $P_1$ and $P_2'$ is simply the commutation of $P_2$ with $P_1^{ -1}$ (making sure to do the commutation the right way). Of course, if this commutation fails, the patches conflict. Moreover, one must check that the merged result actually commutes with $P_1$, as the theorem applies only when \emph{both} commutations are successful. \begin{code}

elegantMerge :: (Patch :\/: Patch) C(x y)
              -> Maybe ((Patch :/\: Patch) C(x y))
elegantMerge (p1 :\/: p2) = do
  p1' :> ip2' <- commute (invert p2 :> p1)
  p1o :> _    <- commute (p2 :> p1')
  guard $ unsafeCompare p1o p1 -- should be a redundant check
  return $ invert ip2' :/\: p1'

\end{code} Of course, there are patches that actually conflict, meaning a merge where the two patches truly cannot both be applied (e.g.\ trying to create a file and a directory with the same name). We deal with this case by creating a special kind of patch to support the merge, which we will call a ``merger''. Basically, a merger is a patch that contains the two patches that conflicted, and instructs darcs basically to resolve the conflict. By construction a merger will satisfy the commutation property (see Definition~\ref{merge_dfn}) that characterizes all merges. Moreover the merger's properties are what makes the order of merges unimportant (which is a rather critical property for darcs as a whole). The job of a merger is basically to undo the two conflicting patches, and then apply some sort of a ``resolution'' of the two instead. In the case of two conflicting hunks, this will look much like what CVS does, where it inserts both versions into the file. In general, of course, the two conflicting patches may both be mergers themselves, in which case the situation is considerably more complicated. \begin{code}
A note about mergers and type witnesses

The merger code predates the introduction of type witnesses, and because
of its complexity has proved the hardest part of the codebase to retrofit.
Attempting to do this has exposed various places where the code behaves
oddly (e.g. 'putBefore' below); these are likely to be bugs but fixing
them would be potentially disruptive and dangerous as it might change
the existing merge behaviour and thus break existing repositories.

As a result the addition of witnesses to this code has required the
liberal use of unsafe operators. In effect, witnesses bring no safety
in this area; the sole purpose of adding them here was to allow this
code to run as part of a codebase that uses witnesses everywhere else.

A key problem point is the type of the 'Merger' and 'Regrem' constructors
of Patch, where the witnesses seem odd. It is likely that some or many
of the unsafe operations could be removed by finding a better type for
these constructors.

actualMerge :: (Patch :\/: Patch) C(x y) -> Sealed (Patch C(y))

actualMerge (ComP the_p1s :\/: ComP the_p2s) =
    mapSeal joinPatchesFL $ mc (the_p1s :\/: the_p2s)
    where mc :: (FL Patch :\/: FL Patch) C(x y) -> Sealed (FL Patch C(y))
          mc (NilFL :\/: (_:>:_)) = Sealed NilFL
          mc (p1s :\/: NilFL) = Sealed p1s
          mc (p1s :\/: (p2:>:p2s)) = case mergePatchesAfterPatch (p1s:\/:p2) of
                                       Sealed x -> mc (x:\/:p2s)
actualMerge (ComP p1s :\/: p2) = seq p2 $
                              mapSeal joinPatchesFL $ mergePatchesAfterPatch (p1s:\/:p2)
actualMerge (p1 :\/: ComP p2s) = seq p1 $ mergePatchAfterPatches (p1:\/:p2s)

actualMerge (p1 :\/: p2) = case elegantMerge (p1:\/:p2) of
                             Just (_ :/\: p1') -> Sealed p1'
                             Nothing -> merger "0.0" p2 p1

mergePatchAfterPatches :: (Patch :\/: FL Patch) C(x y) -> Sealed (Patch C(y))
mergePatchAfterPatches (p :\/: (p1:>:p1s)) =
    case actualMerge (p:\/:p1) of
     Sealed x -> mergePatchAfterPatches (x :\/: p1s)
mergePatchAfterPatches (p :\/: NilFL) = Sealed p

mergePatchesAfterPatch :: (FL Patch :\/: Patch) C(x y) -> Sealed (FL Patch C(y))
mergePatchesAfterPatch (p2s :\/: p) =
    case mergePatchAfterPatches (p :\/: p2s) of
     Sealed x -> case commute (joinPatchesFL p2s :> x) of
                  Just (_ :> ComP p2s') -> Sealed (unsafeCoercePStart p2s')
                  _ -> impossible
\end{code} Much of the merger code depends on a routine which recreates from a single merger the entire sequence of patches which led up to that merger (this is, of course, assuming that this is the complicated general case of a merger of mergers of mergers). This ``unwind'' procedure is rather complicated, but absolutely critical to the merger code, as without it we wouldn't even be able to undo the effects of the patches involved in the merger, since we wouldn't know what patches were all involved in it. Basically, unwind takes a merger such as \begin{verbatim} M( M(A,B), M(A,M(C,D))) \end{verbatim} From which it recreates a merge history: \begin{verbatim} C A M(A,B) M( M(A,B), M(A,M(C,D))) \end{verbatim} (For the curious, yes I can easily unwind this merger in my head [and on paper can unwind insanely more complex mergers]---that's what comes of working for a few months on an algorithm.) Let's start with a simple unwinding. The merger \verb!M(A,B)! simply means that two patches (\verb!A! and \verb!B!) conflicted, and of the two of them \verb!A! is first in the history. The last two patches in the unwinding of any merger are always just this easy. So this unwinds to: \begin{verbatim} A M(A,B) \end{verbatim} What about a merger of mergers? How about \verb!M(A,M(C,D))!. In this case we know the two most recent patches are: \begin{verbatim} A M(A,M(C,D)) \end{verbatim} But obviously the unwinding isn't complete, since we don't yet see where \verb!C! and \verb!D! came from. In this case we take the unwinding of \verb!M(C,D)! and drop its latest patch (which is \verb!M(C,D)! itself) and place that at the beginning of our patch train: \begin{verbatim} C A M(A,M(C,D)) \end{verbatim} As we look at \verb!M( M(A,B), M(A,M(C,D)))!, we consider the unwindings of each of its subpatches: \begin{verbatim} C A A M(A,B) M(A,M(C,D)) \end{verbatim} As we did with \verb!M(A,M(C,D))!, we'll drop the first patch on the right and insert the first patch on the left. That moves us up to the two \verb!A!'s. Since these agree, we can use just one of them (they ``should'' agree). That leaves us with the \verb!C! which goes first. The catch is that things don't always turn out this easily. There is no guarantee that the two \verb!A!'s would come out at the same time, and if they didn't, we'd have to rearrange things until they did. Or if there was no way to rearrange things so that they would agree, we have to go on to plan B, which I will explain now. Consider the case of \verb!M( M(A,B), M(C,D))!. We can easily unwind the two subpatches \begin{verbatim} A C M(A,B) M(C,D) \end{verbatim} Now we need to reconcile the \verb!A! and \verb!C!. How do we do this? Well, as usual, the solution is to use the most wonderful Theorem~\ref{merge_thm}. In this case we have to use it in the reverse of how we used it when merging, since we know that \verb!A! and \verb!C! could either one be the \emph{last} patch applied before \verb!M(A,B)! or \verb!M(C,D)!. So we can find \verb!C'! using \[ A^{ -1} C \commutex C' A'^{ -1} \] Giving an unwinding of \begin{verbatim} C' A M(A,B) M( M(A,B), M(C,D) ) \end{verbatim} There is a bit more complexity to the unwinding process (mostly having to do with cases where you have deeper nesting), but I think the general principles that are followed are pretty much included in the above discussion. \begin{code}
unwind :: Patch C(x y) -> Sealed (RL Patch C(x)) -- Recreates a patch history in reverse.
unwind (Merger _ unwindings _ _) = Sealed unwindings
unwind p = Sealed (p :<: NilRL)

trueUnwind :: Patch C(x y) -> Sealed (RL Patch C(x)) -- Recreates a patch history in reverse.
trueUnwind p@(Merger _ _ p1 p2) =
    case (unwind p1, unwind p2) of
    (Sealed (_:<:p1s),Sealed (_:<:p2s)) ->
         Sealed (p :<: unsafeCoerceP p1 :<: unsafeUnsealFlipped (reconcileUnwindings p p1s (unsafeCoercePEnd p2s)))
    _ -> impossible
trueUnwind _ = impossible

reconcileUnwindings :: Patch C(a b) -> RL Patch C(x z) -> RL Patch C(y z) -> FlippedSeal (RL Patch) C(z)
reconcileUnwindings _ NilRL p2s = FlippedSeal p2s
reconcileUnwindings _ p1s NilRL = FlippedSeal p1s
reconcileUnwindings p (p1:<:p1s) p2s@(p2:<:tp2s) =
    case [(p1s', p2s')|
          p1s'@(hp1s':<:_) <- headPermutationsRL (p1:<:p1s),
          p2s'@(hp2s':<:_) <- headPermutationsRL p2s,
          hp1s' `unsafeCompare` hp2s'] of
    ((p1':<:p1s', _:<:p2s'):_) ->
        mapFlipped (p1' :<:) $ reconcileUnwindings p p1s' (unsafeCoercePEnd p2s')
    [] -> case reverseFL `fmap` putBefore p1 (reverseRL p2s) of
          Just p2s' -> mapFlipped (p1 :<:) $ reconcileUnwindings p p1s p2s'
          Nothing ->
              case fmap reverseFL $ putBefore p2 $
                   reverseRL (p1:<:p1s) of
              Just p1s' -> mapFlipped (p2 :<:) $
                           reconcileUnwindings p p1s' tp2s
              Nothing ->
                  bugDoc $ text "in function reconcileUnwindings"
                        $$ text "Original patch:"
                        $$ showPatch_ p
    _ -> bug "in reconcileUnwindings"

-- This code seems wrong, shouldn't the commute be invert p1 :> p2 ? And why isn't p1' re-inverted?
-- it seems to have been this way forever:
-- Fri May 23 10:27:04 BST 2003  droundy@abridgegame.org
--    * fix bug in unwind and add docs on unwind algorithm.
putBefore :: Patch C(y z) -> FL Patch C(x z) -> Maybe (FL Patch C(y w))
putBefore p1 (p2:>:p2s) =
    do p1' :> p2' <- commute (unsafeCoerceP p2 :> invert p1)
       commute (p2' :> p1)
       (unsafeCoerceP p2' :>:) `fmap` putBefore p1' (unsafeCoerceP p2s)
putBefore _ NilFL = Just (unsafeCoerceP NilFL)
\end{code} \section{Conflicts} There are a couple of simple constraints on the routine which determines how to resolve two conflicting patches (which is called `glump'). These must be satisfied in order that the result of a series of merges is always independent of their order. Firstly, the output of glump cannot change when the order of the two conflicting patches is switched. If it did, then commuting the merger could change the resulting patch, which would be bad. Secondly, the result of the merge of three (or more) conflicting patches cannot depend on the order in which the merges are performed. The conflict resolution code (glump) begins by ``unravelling'' the merger into a set of sequences of patches. Each sequence of patches corresponds to one non-conflicted patch that got merged together with the others. The result of the unravelling of a series of merges must obviously be independent of the order in which those merges are performed. This unravelling code (which uses the unwind code mentioned above) uses probably the second most complicated algorithm. Fortunately, if we can successfully unravel the merger, almost any function of the unravelled merger satisfies the two constraints mentioned above that the conflict resolution code must satisfy. \begin{code}
instance Conflict Patch where
  commuteNoConflicts (x:>y) =   do x' :< y' <- commuteNoMerger (y :< x)
                                   return (y':>x')
  resolveConflicts patch = rcs NilFL $ reverseFL $ flattenFL patch
    where rcs :: FL Patch C(y w) -> RL Patch C(x y) -> [[Sealed (FL Prim C(w))]]
          rcs _ NilRL = []
          rcs passedby (p@(Merger _ _ _ _):<:ps) =
              case commuteNoMerger (joinPatchesFL passedby:<p) of
              Just (p'@(Merger _ _ p1 p2):<_) ->
                  (map Sealed $ nubBy unsafeCompare $
                        effect (unsafeCoercePStart $ unsafeUnseal (glump09 p1 p2)) : map (unsafeCoercePStart . unsafeUnseal) (unravel p'))
                  : rcs (p :>: passedby) ps
              Nothing -> rcs (p :>: passedby) ps
              _ -> impossible
          rcs passedby (p:<:ps) = seq passedby $
                                  rcs (p :>: passedby) ps

-- This type seems wrong - the most natural type for the result would seem to be
-- [Sealed (FL Prim C(x))], given the type of unwind.
-- However downstream code in darcs convert assumes the C(y) type, and I was unable
-- to figure out whether this could/should reasonably be changed -- Ganesh 13/4/10
publicUnravel :: Patch C(x y) -> [Sealed (FL Prim C(y))]
publicUnravel = map (mapSeal unsafeCoercePStart) . unravel

unravel :: Patch C(x y) -> [Sealed (FL Prim C(x))]
unravel p = nub $ map (mapSeal (sortCoalesceFL . concatFL . mapFL_FL effect)) $
            getSupers $ map (mapSeal reverseRL) $ unseal (newUr p) $ unwind p

getSupers :: [Sealed (FL Patch C(x))] -> [Sealed (FL Patch C(x))]
getSupers (x:xs) =
    case filter (not.(x `isSuperpatchOf`)) xs of
    xs' -> if or $ map (`isSuperpatchOf` x) xs'
           then getSupers xs'
           else x : getSupers xs'
getSupers [] = []

isSuperpatchOf :: Sealed (FL Patch C(x)) -> Sealed (FL Patch C(x)) -> Bool
Sealed x `isSuperpatchOf` Sealed y | lengthFL y > lengthFL x = False -- should be just an optimisation
Sealed x `isSuperpatchOf` Sealed y = x `iso` y
    where iso :: FL Patch C(x y) -> FL Patch C(x z) -> Bool
          _ `iso` NilFL = True
          NilFL `iso` _ = False
          a `iso` (b:>:bs) =
              head $ ([as `iso` bs | (ah :>: as) <- simpleHeadPermutationsFL a, IsEq <- [ah =\/= b]] :: [Bool]) ++ [False]

merger :: String -> Patch C(x y) -> Patch C(x z) -> Sealed (Patch C(y))
merger "0.0" p1 p2 = Sealed $ Merger undoit unwindings p1 p2
    where fake_p = Merger identity NilRL p1 p2
          unwindings = unsafeUnseal (trueUnwind fake_p)
          p = Merger identity unwindings p1 p2
          undoit =
              case (isMerger p1, isMerger p2) of
              (True ,True ) -> case unwind p of
                                 Sealed (_:<:t) -> unsafeCoerceP $ joinPatchesFL $ invertRL t
                                 _ -> impossible
              (False,False) -> unsafeCoerceP $ invert p1
              (True ,False) -> unsafeCoerceP $ joinPatchesFL NilFL
              (False,True ) -> unsafeCoerceP $ joinPatchesFL (invert p1 :>: mergerUndo p2 :>: NilFL)
merger g _ _ =
    error $ "Cannot handle mergers other than version 0.0\n"++g
    ++ "\nPlease use darcs optimize --modernize with an older darcs."

glump09 :: Patch C(x y) -> Patch C(x z) -> Sealed (Patch C(y))
glump09 p1 p2 = mapSeal fromPrims $ mangleUnravelled $ unseal unravel $ merger "0.0" p1 p2

mangleUnravelled :: [Sealed (FL Prim C(x))] -> Sealed (FL Prim C(x))
mangleUnravelled pss = if onlyHunks pss
                        then (:>: NilFL) `mapSeal` mangleUnravelledHunks pss
                        else head pss

onlyHunks :: [Sealed (FL Prim C(x))] -> Bool
onlyHunks [] = False
onlyHunks pss = fn2fp f /= "" && all oh pss
    where f = getAFilename pss
          oh :: Sealed (FL Prim C(x)) -> Bool
          oh (Sealed (p:>:ps)) = primIsHunk p &&
                                 [fn2fp f] == listTouchedFiles p &&
                                 oh (Sealed ps)
          oh (Sealed NilFL) = True

applyHunks :: [Maybe B.ByteString] -> FL Prim C(x y) -> [Maybe B.ByteString]
applyHunks ms (FP _ (Hunk l o n):>:ps) = applyHunks (rls l ms) ps
    where rls 1 mls = map Just n ++ drop (length o) mls
          rls i (ml:mls) = ml : rls (i-1) mls
          rls _ [] = bug "rls in applyHunks"
applyHunks ms NilFL = ms
applyHunks _ (_:>:_) = impossible

getOld :: [Maybe B.ByteString] -> [Sealed (FL Prim C(x))] -> [Maybe B.ByteString]
getOld mls (ps:pss) = getOld (getHunksOld mls ps) pss
getOld mls [] = mls

getAFilename :: [Sealed (FL Prim C(x))] -> FileName
getAFilename ((Sealed (FP f _:>:_)):_) = f
getAFilename _ = fp2fn ""

getHunksOld :: [Maybe B.ByteString] -> Sealed (FL Prim C(x))
              -> [Maybe B.ByteString]
getHunksOld mls (Sealed ps) =
    applyHunks (applyHunks mls ps) (invert ps)

getHunksNew :: [Maybe B.ByteString] -> Sealed (FL Prim C(x))
              -> [Maybe B.ByteString]
getHunksNew mls (Sealed ps) = applyHunks mls ps

getHunkline :: [[Maybe B.ByteString]] -> Int
getHunkline = ghl 1
    where ghl :: Int -> [[Maybe B.ByteString]] -> Int
          ghl n pps =
            if any (isJust . head) pps
            then n
            else ghl (n+1) $ map tail pps

makeChunk :: Int -> [Maybe B.ByteString] -> [B.ByteString]
makeChunk n mls = pull_chunk $ drop (n-1) mls
    where pull_chunk (Just l:mls') = l : pull_chunk mls'
          pull_chunk (Nothing:_) = []
          pull_chunk [] = bug "should this be [] in pull_chunk?"

mangleUnravelledHunks :: [Sealed (FL Prim C(x))] -> Sealed (Prim C(x))
--mangleUnravelledHunks [[h1],[h2]] = Deal with simple cases handily?
mangleUnravelledHunks pss =
        if null nchs then bug "mangleUnravelledHunks"
                     else Sealed (FP filename (Hunk l old new))
    where oldf = getOld (repeat Nothing) pss
          newfs = map (getHunksNew oldf) pss
          l = getHunkline $ oldf : newfs
          nchs = sort $ map (makeChunk l) newfs
          filename = getAFilename pss
          old = makeChunk l oldf
          new = [top] ++ concat (intersperse [middle] nchs) ++ [bottom]
          top    = BC.pack $ "v v v v v v v" ++ eol_c
          middle = BC.pack $ "*************" ++ eol_c
          bottom = BC.pack $ "^ ^ ^ ^ ^ ^ ^" ++ eol_c
          eol_c  = if any (\ps -> not (B.null ps) && BC.last ps == '\r') old
                   then "\r"
                   else ""

instance Effect Patch where
    effect p@(Merger _ _ _ _) = sortCoalesceFL $ effect $ mergerUndo p
    effect p@(Regrem _ _ _ _) = invert $ effect $ invert p
    effect (ComP ps) = concatFL $ mapFL_FL effect ps
    effect (PP p) = effect p
    isHunk p = do PP p' <- return p
                  isHunk p'

-- |@modernizePatch@ is used during conversion to Darcs 2 format.
-- It does the following:
--   * removes mergers by linearising them, thus removing the ability
--     to commute them
--   * drops mv a b ; add b which was introduced by an error in earlier
--     versions of darcs (TODO: check this; identify the versions)
modernizePatch :: Patch C(x y) -> Patch C(x y)
modernizePatch p@(Merger _ _ _ _) = fromPrims $ effect p
modernizePatch p@(Regrem _ _ _ _) = fromPrims $ effect p
modernizePatch (ComP ps) = ComP $ filtermv $ mapFL_FL modernizePatch ps
    where filtermv :: FL Patch C(x y) -> FL Patch C(x y)
          filtermv (PP (Move _ b :: Prim C(x z)):>:xs)
            | IsEq <- hasadd xs = filtermv xs
              where hasadd :: FL Patch C(a b) -> EqCheck C(x z)
                    hasadd (PP (FP b' AddFile):>:_) | b' == b = unsafeCoerceP IsEq
                    hasadd (PP (DP b' AddDir):>:_) | b' == b = unsafeCoerceP IsEq
                    hasadd (PP (FP b' RmFile):>:_) | b' == b = NotEq
                    hasadd (PP (DP b' RmDir):>:_) | b' == b = NotEq
                    hasadd (_:>:z) = hasadd z
                    hasadd NilFL = NotEq
          filtermv (x:>:xs) = x :>: filtermv xs
          filtermv NilFL = NilFL

modernizePatch (PP p) = fromPrims $ modernizePrim p

instance FromPrims Patch where
    fromPrims (p :>: NilFL) = PP p
    fromPrims ps = joinPatchesFL $ mapFL_FL PP ps
    joinPatches = joinPatchesFL

newUr :: Patch C(a b) -> RL Patch C(x y) -> [Sealed (RL Patch C(x))]
newUr p (Merger _ _ p1 p2 :<: ps) =
   case filter (\(pp:<:_) -> pp `unsafeCompare` p1) $ headPermutationsRL ps of
   ((_:<:ps'):_) -> newUr p (unsafeCoercePStart p1:<:ps') ++ newUr p (unsafeCoercePStart p2:<:ps')
   _ -> bugDoc $ text "in function newUr"
              $$ text "Original patch:"
              $$ showPatch_ p
              $$ text "Unwound:"
              $$ vcat (unseal (mapRL showPatch_) $ unwind p)

newUr op ps =
    case filter (\(p:<:_) -> isMerger p) $ headPermutationsRL ps of
    [] -> [Sealed ps]
    (ps':_) -> newUr op ps'

instance Invert p => Invert (Named p) where
    invert (NamedP n d p)  = NamedP (invertName n) (map invertName d) (invert p)
    identity = NamedP idpatchinfo [] identity

instance Invert Patch where
    invert (Merger undo unwindings p1 p2)
        = Regrem undo unwindings p1 p2
    invert (Regrem undo unwindings p1 p2)
        = Merger undo unwindings p1 p2
    invert (PP p) = PP (invert p)
    invert (ComP ps)  = ComP $ invert ps
    identity = ComP NilFL

instance MyEq Patch where
    unsafeCompare = eqPatches
instance MyEq p => MyEq (Named p) where
    unsafeCompare (NamedP n1 d1 p1) (NamedP n2 d2 p2) =
        n1 == n2 && d1 == d2 && unsafeCompare p1 p2

eqPatches :: Patch C(x y) -> Patch C(w z) -> Bool
eqPatches (PP p1) (PP p2) = unsafeCompare p1 p2
eqPatches (ComP ps1) (ComP ps2)
 = eqFL eqPatches ps1 ps2
eqPatches (ComP NilFL) (PP Identity) = True
eqPatches (PP Identity) (ComP NilFL) = True
eqPatches (Merger _ _ p1a p1b) (Merger _ _ p2a p2b)
 = eqPatches p1a p2a &&
   eqPatches p1b p2b
eqPatches (Regrem _ _ p1a p1b) (Regrem _ _ p2a p2b)
 = eqPatches p1a p2a &&
   eqPatches p1b p2b
eqPatches _ _ = False

eqFL :: (FORALL(b c d e) a C(b c) -> a C(d e) -> Bool)
      -> FL a C(x y) -> FL a C(w z) -> Bool
eqFL _ NilFL NilFL = True
eqFL f (x:>:xs) (y:>:ys) = f x y && eqFL f xs ys
eqFL _ _ _ = False