[PATCH 0/2] A new merge algorithm, take 3

Hi,

Here is the new version of the merge algorithm patch.

The major changes compared to the previous patch are:

* No more messing around with merge-cache. git-ls-files used to get
  the unmerged entries instead.
* The python code is now contained in two files, git-merge-script and
  gitMergeCommon.py.
* The user interface is identical to the interface provided by
  git-resolve-script
* In the non-clean case the unmerged cache entries will not be
  removed from the cache.

I have also attached a test script which can redo every merge in a repository with both git-resolve-script and git-merge-script. It will report any non-clean merges and non-identical results for clean merges. Do _not_ use this script in repositories you care about. It calls 'git reset --hard' repeatedly and will probably not leave the repository in its original state when it's done.

Of the 500 merge commits that currently exists in the kernel repository 19 produces non-clean merges with git-merge-script. The four merge cases listed in <20050827014009.GB18880@c165.ib.student.liu.se> are cleanly merged by git-merge-script. Every merge commit which is cleanly merged by git-resolve-script is also cleanly merged by git-merge-script, furthermore the results are identical. There are currently two merges in the kernel repository which are not cleanly merged by git-resolve-script but are cleanly merged by git-merge-script.

I guess the need for this has decreased with Daniel's new read-tree code. Is there any chance of getting this code merged into mainline git?

- Fredrik

#!/usr/bin/env python

import sys, math, random, os, re, signal, tempfile, time from heapq import heappush, heappop from sets import Set from gitMergeCommon import *

def mergeMerge(a, b):
    print 'Running merge-script HEAD', b.sha, '...'
    [out, code] = runProgram(['git-merge-script', 'HEAD', b.sha, 'merge message'],
                             returnCode=True, pipeOutput=False)
    if code == 0:
        return True
    else:
        return False
    
def gitResolveMerge(a, b):
    print 'Running git resolve HEAD', b.sha, '...'
    [out, code] = runProgram(['git', 'resolve', 'HEAD', b.sha, 'merge message'],
                             returnCode=True, pipeOutput=False)

    if code == 0:
        return True
    else:
        return False

def doWork(graph, commits):
    print 'commits:', repr(commits)
    result = []
    totalMergeTime = 0
    totalResolveTime = 0
    numCommits = 0
    try:
        for commit in graph.commits:
            if len(commits) > 0 and not (commit.sha in commits):
                continue

            if len(commit.parents) > 1:
                res = commit.sha + ' : '
                if len(commit.parents) == 2:
                    numCommits += 1
                    print '---------------------------------------'
                    print 'Testing commit', commit.sha, '(tree)', commit.tree()
                    a = commit.parents[0]
                    b = commit.parents[1]

                    runProgram(['git-reset-script', '--hard', a.sha])
                    print 'Running git resolve...'
                    stdout.flush()
                    startTime = time.time()
                    resResolve = gitResolveMerge(a, b)
                    timeResolve = time.time() - startTime
                    totalResolveTime += timeResolve
                    
                    if resResolve:
                        resolveHead = Commit(runProgram(['git-rev-parse', '--verify', 'HEAD']).rstrip(), [a, b])

                    runProgram(['git-reset-script', '--hard', a.sha])
                    print 'Running merge...'
                    stdout.flush()
                    startTime = time.time()
                    resMerge = mergeMerge(a, b)
                    timeMerge = time.time() - startTime
                    totalMergeTime += timeMerge
                    
                    if resMerge:
                        mergeHead = Commit(runProgram(['git-rev-parse', '--verify', 'HEAD']).rstrip(), [a, b])

                    res += 'time r: ' + str(int(timeResolve)) + ' m: ' + str(int(timeMerge)) + '\t'
                    if resResolve and resMerge:
                        if resolveHead.tree() == mergeHead.tree():
                            res += 'Identical result'
                        else:
                            res += 'Non-identical results! resolve: ' + resolveHead.sha + \
                                   ' merge: ' + mergeHead.sha
                    else:
                        if resResolve:
                            res += 'resolve succeeded (' + resolveHead.sha + '), '
                        else:
                            res += 'resolve failed, '

                        if resMerge:
                            res += 'merge succeeded (' + mergeHead.sha + ')'
                        else:
                            res += 'merge failed'
                else:
                    res += 'Ignoring octupus merge'

                print res
                result.append(res)
                stdout.flush()
    finally:
        print '\n\n\nResults:'
        for r in result:
            print r
        print 'Avg resolve time:', float(totalResolveTime) / numCommits
        print 'Avg merge time:', float(totalMergeTime) / numCommits

def writeHead(head, sha):
    if sha[-1] != '\n':
        sha += '\n'

    try:
        f = open(os.environ['GIT_DIR'] + '/' + head, 'w')
        f.write(sha)
        f.close()
    except IOError, e:
        print 'Failed to write to', os.environ['GIT_DIR'] + '/' + head + ':', e.strerror
        sys.exit(1)
    return True

stdout = sys.stdout setupEnvironment() repoValid()

head = runProgram(['git-rev-parse', '--verify', 'HEAD^0']).rstrip()
print 'Building graph...'
stdout.flush()
graph = buildGraph([head])
print 'Graph building done.'
stdout.flush()
print 'Processing', len(graph.commits), 'commits (' + \
      str(len([x for x in graph.commits if len(x.parents) > 1])) + ' merge commits)...'

originalHead = open('.git/HEAD', 'r').read() print 'Original head:', originalHead stdout.flush()

writeHead('original-head', originalHead)

try:
    doWork(graph, sys.argv[1:])
finally:
    writeHead('HEAD', originalHead)