revlog: don't consider nullrev when choosing delta base

In the most complex case, we try using the incoming delta base, then
we try both parents, and then we try the previous revlog entry. If
none of these result in a good delta, we natually use the null
revision as base. However, we sometimes consider the nullrev before we
have exhausted our other options. Specifically, when both parents are
null, we use the nullrev as delta base if it produces a good delta
(according to _isgooddelta()), and we fail to try the previous revlog
entry as delta base. After 20a9226bdc8a (addrevision: use general
delta when the incoming base delta is bad, 2015-12-01), it can also
happen for non-merge commits when the incoming delta is not good.

The Firefox repo (from many months back) shrinks a tiny bit with this
patch: from 1.855GB to 1.830GB (1.4%). The hg repo itself shrinks even
less: by less than 0.1%. There may be repos that get larger instead.

This undoes the unexplained test change in 20a9226bdc8a.

mercurial/revlog.py

@@ -1,1731 +1,1731 b''
 # revlog.py - storage back-end for mercurial
 #
 # Copyright 2005-2007 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 """Storage back-end for Mercurial.
 
 This provides efficient delta storage with O(1) retrieve and append
 and O(changes) merge between branches.
 """
 
 # import stuff from node for others to import from revlog
 import collections
 from node import bin, hex, nullid, nullrev
 from i18n import _
 import ancestor, mdiff, parsers, error, util, templatefilters
 import struct, zlib, errno
 
 _pack = struct.pack
 _unpack = struct.unpack
 _compress = zlib.compress
 _decompress = zlib.decompress
 _sha = util.sha1
 
 # revlog header flags
 REVLOGV0 = 0
 REVLOGNG = 1
 REVLOGNGINLINEDATA = (1 << 16)
 REVLOGGENERALDELTA = (1 << 17)
 REVLOG_DEFAULT_FLAGS = REVLOGNGINLINEDATA
 REVLOG_DEFAULT_FORMAT = REVLOGNG
 REVLOG_DEFAULT_VERSION = REVLOG_DEFAULT_FORMAT | REVLOG_DEFAULT_FLAGS
 REVLOGNG_FLAGS = REVLOGNGINLINEDATA | REVLOGGENERALDELTA
 
 # revlog index flags
 REVIDX_ISCENSORED = (1 << 15) # revision has censor metadata, must be verified
 REVIDX_DEFAULT_FLAGS = 0
 REVIDX_KNOWN_FLAGS = REVIDX_ISCENSORED
 
 # max size of revlog with inline data
 _maxinline = 131072
 _chunksize = 1048576
 
 RevlogError = error.RevlogError
 LookupError = error.LookupError
 CensoredNodeError = error.CensoredNodeError
 
 def getoffset(q):
     return int(q >> 16)
 
 def gettype(q):
     return int(q & 0xFFFF)
 
 def offset_type(offset, type):
     return long(long(offset) << 16 | type)
 
 _nullhash = _sha(nullid)
 
 def hash(text, p1, p2):
     """generate a hash from the given text and its parent hashes
 
     This hash combines both the current file contents and its history
     in a manner that makes it easy to distinguish nodes with the same
     content in the revision graph.
     """
     # As of now, if one of the parent node is null, p2 is null
     if p2 == nullid:
         # deep copy of a hash is faster than creating one
         s = _nullhash.copy()
         s.update(p1)
     else:
         # none of the parent nodes are nullid
         l = [p1, p2]
         l.sort()
         s = _sha(l[0])
         s.update(l[1])
     s.update(text)
     return s.digest()
 
 def decompress(bin):
     """ decompress the given input """
     if not bin:
         return bin
     t = bin[0]
     if t == '\0':
         return bin
     if t == 'x':
         try:
             return _decompress(bin)
         except zlib.error as e:
             raise RevlogError(_("revlog decompress error: %s") % str(e))
     if t == 'u':
         return bin[1:]
     raise RevlogError(_("unknown compression type %r") % t)
 
 # index v0:
 #  4 bytes: offset
 #  4 bytes: compressed length
 #  4 bytes: base rev
 #  4 bytes: link rev
 # 20 bytes: parent 1 nodeid
 # 20 bytes: parent 2 nodeid
 # 20 bytes: nodeid
 indexformatv0 = ">4l20s20s20s"
 
 class revlogoldio(object):
     def __init__(self):
         self.size = struct.calcsize(indexformatv0)
 
     def parseindex(self, data, inline):
         s = self.size
         index = []
         nodemap =  {nullid: nullrev}
         n = off = 0
         l = len(data)
         while off + s <= l:
             cur = data[off:off + s]
             off += s
             e = _unpack(indexformatv0, cur)
             # transform to revlogv1 format
             e2 = (offset_type(e[0], 0), e[1], -1, e[2], e[3],
                   nodemap.get(e[4], nullrev), nodemap.get(e[5], nullrev), e[6])
             index.append(e2)
             nodemap[e[6]] = n
             n += 1
 
         # add the magic null revision at -1
         index.append((0, 0, 0, -1, -1, -1, -1, nullid))
 
         return index, nodemap, None
 
     def packentry(self, entry, node, version, rev):
         if gettype(entry[0]):
             raise RevlogError(_("index entry flags need RevlogNG"))
         e2 = (getoffset(entry[0]), entry[1], entry[3], entry[4],
               node(entry[5]), node(entry[6]), entry[7])
         return _pack(indexformatv0, *e2)
 
 # index ng:
 #  6 bytes: offset
 #  2 bytes: flags
 #  4 bytes: compressed length
 #  4 bytes: uncompressed length
 #  4 bytes: base rev
 #  4 bytes: link rev
 #  4 bytes: parent 1 rev
 #  4 bytes: parent 2 rev
 # 32 bytes: nodeid
 indexformatng = ">Qiiiiii20s12x"
 versionformat = ">I"
 
 # corresponds to uncompressed length of indexformatng (2 gigs, 4-byte
 # signed integer)
 _maxentrysize = 0x7fffffff
 
 class revlogio(object):
     def __init__(self):
         self.size = struct.calcsize(indexformatng)
 
     def parseindex(self, data, inline):
         # call the C implementation to parse the index data
         index, cache = parsers.parse_index2(data, inline)
         return index, getattr(index, 'nodemap', None), cache
 
     def packentry(self, entry, node, version, rev):
         p = _pack(indexformatng, *entry)
         if rev == 0:
             p = _pack(versionformat, version) + p[4:]
         return p
 
 class revlog(object):
     """
     the underlying revision storage object
 
     A revlog consists of two parts, an index and the revision data.
 
     The index is a file with a fixed record size containing
     information on each revision, including its nodeid (hash), the
     nodeids of its parents, the position and offset of its data within
     the data file, and the revision it's based on. Finally, each entry
     contains a linkrev entry that can serve as a pointer to external
     data.
 
     The revision data itself is a linear collection of data chunks.
     Each chunk represents a revision and is usually represented as a
     delta against the previous chunk. To bound lookup time, runs of
     deltas are limited to about 2 times the length of the original
     version data. This makes retrieval of a version proportional to
     its size, or O(1) relative to the number of revisions.
 
     Both pieces of the revlog are written to in an append-only
     fashion, which means we never need to rewrite a file to insert or
     remove data, and can use some simple techniques to avoid the need
     for locking while reading.
     """
     def __init__(self, opener, indexfile):
         """
         create a revlog object
 
         opener is a function that abstracts the file opening operation
         and can be used to implement COW semantics or the like.
         """
         self.indexfile = indexfile
         self.datafile = indexfile[:-2] + ".d"
         self.opener = opener
         # 3-tuple of (node, rev, text) for a raw revision.
         self._cache = None
         # 2-tuple of (rev, baserev) defining the base revision the delta chain
         # begins at for a revision.
         self._basecache = None
         # 2-tuple of (offset, data) of raw data from the revlog at an offset.
         self._chunkcache = (0, '')
         # How much data to read and cache into the raw revlog data cache.
         self._chunkcachesize = 65536
         self._maxchainlen = None
         self._aggressivemergedeltas = False
         self.index = []
         # Mapping of partial identifiers to full nodes.
         self._pcache = {}
         # Mapping of revision integer to full node.
         self._nodecache = {nullid: nullrev}
         self._nodepos = None
 
         v = REVLOG_DEFAULT_VERSION
         opts = getattr(opener, 'options', None)
         if opts is not None:
             if 'revlogv1' in opts:
                 if 'generaldelta' in opts:
                     v |= REVLOGGENERALDELTA
             else:
                 v = 0
             if 'chunkcachesize' in opts:
                 self._chunkcachesize = opts['chunkcachesize']
             if 'maxchainlen' in opts:
                 self._maxchainlen = opts['maxchainlen']
             if 'aggressivemergedeltas' in opts:
                 self._aggressivemergedeltas = opts['aggressivemergedeltas']
             self._lazydeltabase = bool(opts.get('lazydeltabase', False))
 
         if self._chunkcachesize <= 0:
             raise RevlogError(_('revlog chunk cache size %r is not greater '
                                 'than 0') % self._chunkcachesize)
         elif self._chunkcachesize & (self._chunkcachesize - 1):
             raise RevlogError(_('revlog chunk cache size %r is not a power '
                                 'of 2') % self._chunkcachesize)
 
         indexdata = ''
         self._initempty = True
         try:
             f = self.opener(self.indexfile)
             indexdata = f.read()
             f.close()
             if len(indexdata) > 0:
                 v = struct.unpack(versionformat, indexdata[:4])[0]
                 self._initempty = False
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
 
         self.version = v
         self._inline = v & REVLOGNGINLINEDATA
         self._generaldelta = v & REVLOGGENERALDELTA
         flags = v & ~0xFFFF
         fmt = v & 0xFFFF
         if fmt == REVLOGV0 and flags:
             raise RevlogError(_("index %s unknown flags %#04x for format v0")
                               % (self.indexfile, flags >> 16))
         elif fmt == REVLOGNG and flags & ~REVLOGNG_FLAGS:
             raise RevlogError(_("index %s unknown flags %#04x for revlogng")
                               % (self.indexfile, flags >> 16))
         elif fmt > REVLOGNG:
             raise RevlogError(_("index %s unknown format %d")
                               % (self.indexfile, fmt))
 
         self._io = revlogio()
         if self.version == REVLOGV0:
             self._io = revlogoldio()
         try:
             d = self._io.parseindex(indexdata, self._inline)
         except (ValueError, IndexError):
             raise RevlogError(_("index %s is corrupted") % (self.indexfile))
         self.index, nodemap, self._chunkcache = d
         if nodemap is not None:
             self.nodemap = self._nodecache = nodemap
         if not self._chunkcache:
             self._chunkclear()
         # revnum -> (chain-length, sum-delta-length)
         self._chaininfocache = {}
 
     def tip(self):
         return self.node(len(self.index) - 2)
     def __contains__(self, rev):
         return 0 <= rev < len(self)
     def __len__(self):
         return len(self.index) - 1
     def __iter__(self):
         return iter(xrange(len(self)))
     def revs(self, start=0, stop=None):
         """iterate over all rev in this revlog (from start to stop)"""
         step = 1
         if stop is not None:
             if start > stop:
                 step = -1
             stop += step
         else:
             stop = len(self)
         return xrange(start, stop, step)
 
     @util.propertycache
     def nodemap(self):
         self.rev(self.node(0))
         return self._nodecache
 
     def hasnode(self, node):
         try:
             self.rev(node)
             return True
         except KeyError:
             return False
 
     def clearcaches(self):
         try:
             self._nodecache.clearcaches()
         except AttributeError:
             self._nodecache = {nullid: nullrev}
             self._nodepos = None
 
     def rev(self, node):
         try:
             return self._nodecache[node]
         except TypeError:
             raise
         except RevlogError:
             # parsers.c radix tree lookup failed
             raise LookupError(node, self.indexfile, _('no node'))
         except KeyError:
             # pure python cache lookup failed
             n = self._nodecache
             i = self.index
             p = self._nodepos
             if p is None:
                 p = len(i) - 2
             for r in xrange(p, -1, -1):
                 v = i[r][7]
                 n[v] = r
                 if v == node:
                     self._nodepos = r - 1
                     return r
             raise LookupError(node, self.indexfile, _('no node'))
 
     def node(self, rev):
         return self.index[rev][7]
     def linkrev(self, rev):
         return self.index[rev][4]
     def parents(self, node):
         i = self.index
         d = i[self.rev(node)]
         return i[d[5]][7], i[d[6]][7] # map revisions to nodes inline
     def parentrevs(self, rev):
         return self.index[rev][5:7]
     def start(self, rev):
         return int(self.index[rev][0] >> 16)
     def end(self, rev):
         return self.start(rev) + self.length(rev)
     def length(self, rev):
         return self.index[rev][1]
     def chainbase(self, rev):
         index = self.index
         base = index[rev][3]
         while base != rev:
             rev = base
             base = index[rev][3]
         return base
     def chainlen(self, rev):
         return self._chaininfo(rev)[0]
 
     def _chaininfo(self, rev):
         chaininfocache = self._chaininfocache
         if rev in chaininfocache:
             return chaininfocache[rev]
         index = self.index
         generaldelta = self._generaldelta
         iterrev = rev
         e = index[iterrev]
         clen = 0
         compresseddeltalen = 0
         while iterrev != e[3]:
             clen += 1
             compresseddeltalen += e[1]
             if generaldelta:
                 iterrev = e[3]
             else:
                 iterrev -= 1
             if iterrev in chaininfocache:
                 t = chaininfocache[iterrev]
                 clen += t[0]
                 compresseddeltalen += t[1]
                 break
             e = index[iterrev]
         else:
             # Add text length of base since decompressing that also takes
             # work. For cache hits the length is already included.
             compresseddeltalen += e[1]
         r = (clen, compresseddeltalen)
         chaininfocache[rev] = r
         return r
 
     def flags(self, rev):
         return self.index[rev][0] & 0xFFFF
     def rawsize(self, rev):
         """return the length of the uncompressed text for a given revision"""
         l = self.index[rev][2]
         if l >= 0:
             return l
 
         t = self.revision(self.node(rev))
         return len(t)
     size = rawsize
 
     def ancestors(self, revs, stoprev=0, inclusive=False):
         """Generate the ancestors of 'revs' in reverse topological order.
         Does not generate revs lower than stoprev.
 
         See the documentation for ancestor.lazyancestors for more details."""
 
         return ancestor.lazyancestors(self.parentrevs, revs, stoprev=stoprev,
                                       inclusive=inclusive)
 
     def descendants(self, revs):
         """Generate the descendants of 'revs' in revision order.
 
         Yield a sequence of revision numbers starting with a child of
         some rev in revs, i.e., each revision is *not* considered a
         descendant of itself.  Results are ordered by revision number (a
         topological sort)."""
         first = min(revs)
         if first == nullrev:
             for i in self:
                 yield i
             return
 
         seen = set(revs)
         for i in self.revs(start=first + 1):
             for x in self.parentrevs(i):
                 if x != nullrev and x in seen:
                     seen.add(i)
                     yield i
                     break
 
     def findcommonmissing(self, common=None, heads=None):
         """Return a tuple of the ancestors of common and the ancestors of heads
         that are not ancestors of common. In revset terminology, we return the
         tuple:
 
           ::common, (::heads) - (::common)
 
         The list is sorted by revision number, meaning it is
         topologically sorted.
 
         'heads' and 'common' are both lists of node IDs.  If heads is
         not supplied, uses all of the revlog's heads.  If common is not
         supplied, uses nullid."""
         if common is None:
             common = [nullid]
         if heads is None:
             heads = self.heads()
 
         common = [self.rev(n) for n in common]
         heads = [self.rev(n) for n in heads]
 
         # we want the ancestors, but inclusive
         class lazyset(object):
             def __init__(self, lazyvalues):
                 self.addedvalues = set()
                 self.lazyvalues = lazyvalues
 
             def __contains__(self, value):
                 return value in self.addedvalues or value in self.lazyvalues
 
             def __iter__(self):
                 added = self.addedvalues
                 for r in added:
                     yield r
                 for r in self.lazyvalues:
                     if not r in added:
                         yield r
 
             def add(self, value):
                 self.addedvalues.add(value)
 
             def update(self, values):
                 self.addedvalues.update(values)
 
         has = lazyset(self.ancestors(common))
         has.add(nullrev)
         has.update(common)
 
         # take all ancestors from heads that aren't in has
         missing = set()
         visit = collections.deque(r for r in heads if r not in has)
         while visit:
             r = visit.popleft()
             if r in missing:
                 continue
             else:
                 missing.add(r)
                 for p in self.parentrevs(r):
                     if p not in has:
                         visit.append(p)
         missing = list(missing)
         missing.sort()
         return has, [self.node(r) for r in missing]
 
     def incrementalmissingrevs(self, common=None):
         """Return an object that can be used to incrementally compute the
         revision numbers of the ancestors of arbitrary sets that are not
         ancestors of common. This is an ancestor.incrementalmissingancestors
         object.
 
         'common' is a list of revision numbers. If common is not supplied, uses
         nullrev.
         """
         if common is None:
             common = [nullrev]
 
         return ancestor.incrementalmissingancestors(self.parentrevs, common)
 
     def findmissingrevs(self, common=None, heads=None):
         """Return the revision numbers of the ancestors of heads that
         are not ancestors of common.
 
         More specifically, return a list of revision numbers corresponding to
         nodes N such that every N satisfies the following constraints:
 
           1. N is an ancestor of some node in 'heads'
           2. N is not an ancestor of any node in 'common'
 
         The list is sorted by revision number, meaning it is
         topologically sorted.
 
         'heads' and 'common' are both lists of revision numbers.  If heads is
         not supplied, uses all of the revlog's heads.  If common is not
         supplied, uses nullid."""
         if common is None:
             common = [nullrev]
         if heads is None:
             heads = self.headrevs()
 
         inc = self.incrementalmissingrevs(common=common)
         return inc.missingancestors(heads)
 
     def findmissing(self, common=None, heads=None):
         """Return the ancestors of heads that are not ancestors of common.
 
         More specifically, return a list of nodes N such that every N
         satisfies the following constraints:
 
           1. N is an ancestor of some node in 'heads'
           2. N is not an ancestor of any node in 'common'
 
         The list is sorted by revision number, meaning it is
         topologically sorted.
 
         'heads' and 'common' are both lists of node IDs.  If heads is
         not supplied, uses all of the revlog's heads.  If common is not
         supplied, uses nullid."""
         if common is None:
             common = [nullid]
         if heads is None:
             heads = self.heads()
 
         common = [self.rev(n) for n in common]
         heads = [self.rev(n) for n in heads]
 
         inc = self.incrementalmissingrevs(common=common)
         return [self.node(r) for r in inc.missingancestors(heads)]
 
     def nodesbetween(self, roots=None, heads=None):
         """Return a topological path from 'roots' to 'heads'.
 
         Return a tuple (nodes, outroots, outheads) where 'nodes' is a
         topologically sorted list of all nodes N that satisfy both of
         these constraints:
 
           1. N is a descendant of some node in 'roots'
           2. N is an ancestor of some node in 'heads'
 
         Every node is considered to be both a descendant and an ancestor
         of itself, so every reachable node in 'roots' and 'heads' will be
         included in 'nodes'.
 
         'outroots' is the list of reachable nodes in 'roots', i.e., the
         subset of 'roots' that is returned in 'nodes'.  Likewise,
         'outheads' is the subset of 'heads' that is also in 'nodes'.
 
         'roots' and 'heads' are both lists of node IDs.  If 'roots' is
         unspecified, uses nullid as the only root.  If 'heads' is
         unspecified, uses list of all of the revlog's heads."""
         nonodes = ([], [], [])
         if roots is not None:
             roots = list(roots)
             if not roots:
                 return nonodes
             lowestrev = min([self.rev(n) for n in roots])
         else:
             roots = [nullid] # Everybody's a descendant of nullid
             lowestrev = nullrev
         if (lowestrev == nullrev) and (heads is None):
             # We want _all_ the nodes!
             return ([self.node(r) for r in self], [nullid], list(self.heads()))
         if heads is None:
             # All nodes are ancestors, so the latest ancestor is the last
             # node.
             highestrev = len(self) - 1
             # Set ancestors to None to signal that every node is an ancestor.
             ancestors = None
             # Set heads to an empty dictionary for later discovery of heads
             heads = {}
         else:
             heads = list(heads)
             if not heads:
                 return nonodes
             ancestors = set()
             # Turn heads into a dictionary so we can remove 'fake' heads.
             # Also, later we will be using it to filter out the heads we can't
             # find from roots.
             heads = dict.fromkeys(heads, False)
             # Start at the top and keep marking parents until we're done.
             nodestotag = set(heads)
             # Remember where the top was so we can use it as a limit later.
             highestrev = max([self.rev(n) for n in nodestotag])
             while nodestotag:
                 # grab a node to tag
                 n = nodestotag.pop()
                 # Never tag nullid
                 if n == nullid:
                     continue
                 # A node's revision number represents its place in a
                 # topologically sorted list of nodes.
                 r = self.rev(n)
                 if r >= lowestrev:
                     if n not in ancestors:
                         # If we are possibly a descendant of one of the roots
                         # and we haven't already been marked as an ancestor
                         ancestors.add(n) # Mark as ancestor
                         # Add non-nullid parents to list of nodes to tag.
                         nodestotag.update([p for p in self.parents(n) if
                                            p != nullid])
                     elif n in heads: # We've seen it before, is it a fake head?
                         # So it is, real heads should not be the ancestors of
                         # any other heads.
                         heads.pop(n)
             if not ancestors:
                 return nonodes
             # Now that we have our set of ancestors, we want to remove any
             # roots that are not ancestors.
 
             # If one of the roots was nullid, everything is included anyway.
             if lowestrev > nullrev:
                 # But, since we weren't, let's recompute the lowest rev to not
                 # include roots that aren't ancestors.
 
                 # Filter out roots that aren't ancestors of heads
                 roots = [n for n in roots if n in ancestors]
                 # Recompute the lowest revision
                 if roots:
                     lowestrev = min([self.rev(n) for n in roots])
                 else:
                     # No more roots?  Return empty list
                     return nonodes
             else:
                 # We are descending from nullid, and don't need to care about
                 # any other roots.
                 lowestrev = nullrev
                 roots = [nullid]
         # Transform our roots list into a set.
         descendants = set(roots)
         # Also, keep the original roots so we can filter out roots that aren't
         # 'real' roots (i.e. are descended from other roots).
         roots = descendants.copy()
         # Our topologically sorted list of output nodes.
         orderedout = []
         # Don't start at nullid since we don't want nullid in our output list,
         # and if nullid shows up in descendants, empty parents will look like
         # they're descendants.
         for r in self.revs(start=max(lowestrev, 0), stop=highestrev + 1):
             n = self.node(r)
             isdescendant = False
             if lowestrev == nullrev:  # Everybody is a descendant of nullid
                 isdescendant = True
             elif n in descendants:
                 # n is already a descendant
                 isdescendant = True
                 # This check only needs to be done here because all the roots
                 # will start being marked is descendants before the loop.
                 if n in roots:
                     # If n was a root, check if it's a 'real' root.
                     p = tuple(self.parents(n))
                     # If any of its parents are descendants, it's not a root.
                     if (p[0] in descendants) or (p[1] in descendants):
                         roots.remove(n)
             else:
                 p = tuple(self.parents(n))
                 # A node is a descendant if either of its parents are
                 # descendants.  (We seeded the dependents list with the roots
                 # up there, remember?)
                 if (p[0] in descendants) or (p[1] in descendants):
                     descendants.add(n)
                     isdescendant = True
             if isdescendant and ((ancestors is None) or (n in ancestors)):
                 # Only include nodes that are both descendants and ancestors.
                 orderedout.append(n)
                 if (ancestors is not None) and (n in heads):
                     # We're trying to figure out which heads are reachable
                     # from roots.
                     # Mark this head as having been reached
                     heads[n] = True
                 elif ancestors is None:
                     # Otherwise, we're trying to discover the heads.
                     # Assume this is a head because if it isn't, the next step
                     # will eventually remove it.
                     heads[n] = True
                     # But, obviously its parents aren't.
                     for p in self.parents(n):
                         heads.pop(p, None)
         heads = [n for n, flag in heads.iteritems() if flag]
         roots = list(roots)
         assert orderedout
         assert roots
         assert heads
         return (orderedout, roots, heads)
 
     def headrevs(self):
         try:
             return self.index.headrevs()
         except AttributeError:
             return self._headrevs()
 
     def computephases(self, roots):
         return self.index.computephasesmapsets(roots)
 
     def _headrevs(self):
         count = len(self)
         if not count:
             return [nullrev]
         # we won't iter over filtered rev so nobody is a head at start
         ishead = [0] * (count + 1)
         index = self.index
         for r in self:
             ishead[r] = 1  # I may be an head
             e = index[r]
             ishead[e[5]] = ishead[e[6]] = 0  # my parent are not
         return [r for r, val in enumerate(ishead) if val]
 
     def heads(self, start=None, stop=None):
         """return the list of all nodes that have no children
 
         if start is specified, only heads that are descendants of
         start will be returned
         if stop is specified, it will consider all the revs from stop
         as if they had no children
         """
         if start is None and stop is None:
             if not len(self):
                 return [nullid]
             return [self.node(r) for r in self.headrevs()]
 
         if start is None:
             start = nullid
         if stop is None:
             stop = []
         stoprevs = set([self.rev(n) for n in stop])
         startrev = self.rev(start)
         reachable = set((startrev,))
         heads = set((startrev,))
 
         parentrevs = self.parentrevs
         for r in self.revs(start=startrev + 1):
             for p in parentrevs(r):
                 if p in reachable:
                     if r not in stoprevs:
                         reachable.add(r)
                     heads.add(r)
                 if p in heads and p not in stoprevs:
                     heads.remove(p)
 
         return [self.node(r) for r in heads]
 
     def children(self, node):
         """find the children of a given node"""
         c = []
         p = self.rev(node)
         for r in self.revs(start=p + 1):
             prevs = [pr for pr in self.parentrevs(r) if pr != nullrev]
             if prevs:
                 for pr in prevs:
                     if pr == p:
                         c.append(self.node(r))
             elif p == nullrev:
                 c.append(self.node(r))
         return c
 
     def descendant(self, start, end):
         if start == nullrev:
             return True
         for i in self.descendants([start]):
             if i == end:
                 return True
             elif i > end:
                 break
         return False
 
     def commonancestorsheads(self, a, b):
         """calculate all the heads of the common ancestors of nodes a and b"""
         a, b = self.rev(a), self.rev(b)
         try:
             ancs = self.index.commonancestorsheads(a, b)
         except (AttributeError, OverflowError): # C implementation failed
             ancs = ancestor.commonancestorsheads(self.parentrevs, a, b)
         return map(self.node, ancs)
 
     def isancestor(self, a, b):
         """return True if node a is an ancestor of node b
 
         The implementation of this is trivial but the use of
         commonancestorsheads is not."""
         return a in self.commonancestorsheads(a, b)
 
     def ancestor(self, a, b):
         """calculate the "best" common ancestor of nodes a and b"""
 
         a, b = self.rev(a), self.rev(b)
         try:
             ancs = self.index.ancestors(a, b)
         except (AttributeError, OverflowError):
             ancs = ancestor.ancestors(self.parentrevs, a, b)
         if ancs:
             # choose a consistent winner when there's a tie
             return min(map(self.node, ancs))
         return nullid
 
     def _match(self, id):
         if isinstance(id, int):
             # rev
             return self.node(id)
         if len(id) == 20:
             # possibly a binary node
             # odds of a binary node being all hex in ASCII are 1 in 10**25
             try:
                 node = id
                 self.rev(node) # quick search the index
                 return node
             except LookupError:
                 pass # may be partial hex id
         try:
             # str(rev)
             rev = int(id)
             if str(rev) != id:
                 raise ValueError
             if rev < 0:
                 rev = len(self) + rev
             if rev < 0 or rev >= len(self):
                 raise ValueError
             return self.node(rev)
         except (ValueError, OverflowError):
             pass
         if len(id) == 40:
             try:
                 # a full hex nodeid?
                 node = bin(id)
                 self.rev(node)
                 return node
             except (TypeError, LookupError):
                 pass
 
     def _partialmatch(self, id):
         try:
             n = self.index.partialmatch(id)
             if n and self.hasnode(n):
                 return n
             return None
         except RevlogError:
             # parsers.c radix tree lookup gave multiple matches
             # fall through to slow path that filters hidden revisions
             pass
         except (AttributeError, ValueError):
             # we are pure python, or key was too short to search radix tree
             pass
 
         if id in self._pcache:
             return self._pcache[id]
 
         if len(id) < 40:
             try:
                 # hex(node)[:...]
                 l = len(id) // 2  # grab an even number of digits
                 prefix = bin(id[:l * 2])
                 nl = [e[7] for e in self.index if e[7].startswith(prefix)]
                 nl = [n for n in nl if hex(n).startswith(id) and
                       self.hasnode(n)]
                 if len(nl) > 0:
                     if len(nl) == 1:
                         self._pcache[id] = nl[0]
                         return nl[0]
                     raise LookupError(id, self.indexfile,
                                       _('ambiguous identifier'))
                 return None
             except TypeError:
                 pass
 
     def lookup(self, id):
         """locate a node based on:
             - revision number or str(revision number)
             - nodeid or subset of hex nodeid
         """
         n = self._match(id)
         if n is not None:
             return n
         n = self._partialmatch(id)
         if n:
             return n
 
         raise LookupError(id, self.indexfile, _('no match found'))
 
     def cmp(self, node, text):
         """compare text with a given file revision
 
         returns True if text is different than what is stored.
         """
         p1, p2 = self.parents(node)
         return hash(text, p1, p2) != node
 
     def _addchunk(self, offset, data):
         """Add a segment to the revlog cache.
 
         Accepts an absolute offset and the data that is at that location.
         """
         o, d = self._chunkcache
         # try to add to existing cache
         if o + len(d) == offset and len(d) + len(data) < _chunksize:
             self._chunkcache = o, d + data
         else:
             self._chunkcache = offset, data
 
     def _loadchunk(self, offset, length, df=None):
         """Load a segment of raw data from the revlog.
 
         Accepts an absolute offset, length to read, and an optional existing
         file handle to read from.
 
         If an existing file handle is passed, it will be seeked and the
         original seek position will NOT be restored.
 
         Returns a str or buffer of raw byte data.
         """
         if df is not None:
             closehandle = False
         else:
             if self._inline:
                 df = self.opener(self.indexfile)
             else:
                 df = self.opener(self.datafile)
             closehandle = True
 
         # Cache data both forward and backward around the requested
         # data, in a fixed size window. This helps speed up operations
         # involving reading the revlog backwards.
         cachesize = self._chunkcachesize
         realoffset = offset & ~(cachesize - 1)
         reallength = (((offset + length + cachesize) & ~(cachesize - 1))
                       - realoffset)
         df.seek(realoffset)
         d = df.read(reallength)
         if closehandle:
             df.close()
         self._addchunk(realoffset, d)
         if offset != realoffset or reallength != length:
             return util.buffer(d, offset - realoffset, length)
         return d
 
     def _getchunk(self, offset, length, df=None):
         """Obtain a segment of raw data from the revlog.
 
         Accepts an absolute offset, length of bytes to obtain, and an
         optional file handle to the already-opened revlog. If the file
         handle is used, it's original seek position will not be preserved.
 
         Requests for data may be returned from a cache.
 
         Returns a str or a buffer instance of raw byte data.
         """
         o, d = self._chunkcache
         l = len(d)
 
         # is it in the cache?
         cachestart = offset - o
         cacheend = cachestart + length
         if cachestart >= 0 and cacheend <= l:
             if cachestart == 0 and cacheend == l:
                 return d # avoid a copy
             return util.buffer(d, cachestart, cacheend - cachestart)
 
         return self._loadchunk(offset, length, df=df)
 
     def _chunkraw(self, startrev, endrev, df=None):
         """Obtain a segment of raw data corresponding to a range of revisions.
 
         Accepts the start and end revisions and an optional already-open
         file handle to be used for reading. If the file handle is read, its
         seek position will not be preserved.
 
         Requests for data may be satisfied by a cache.
 
         Returns a str or a buffer instance of raw byte data. Callers will
         need to call ``self.start(rev)`` and ``self.length()`` to determine
         where each revision's data begins and ends.
         """
         start = self.start(startrev)
         end = self.end(endrev)
         if self._inline:
             start += (startrev + 1) * self._io.size
             end += (endrev + 1) * self._io.size
         length = end - start
         return self._getchunk(start, length, df=df)
 
     def _chunk(self, rev, df=None):
         """Obtain a single decompressed chunk for a revision.
 
         Accepts an integer revision and an optional already-open file handle
         to be used for reading. If used, the seek position of the file will not
         be preserved.
 
         Returns a str holding uncompressed data for the requested revision.
         """
         return decompress(self._chunkraw(rev, rev, df=df))
 
     def _chunks(self, revs, df=None):
         """Obtain decompressed chunks for the specified revisions.
 
         Accepts an iterable of numeric revisions that are assumed to be in
         ascending order. Also accepts an optional already-open file handle
         to be used for reading. If used, the seek position of the file will
         not be preserved.
 
         This function is similar to calling ``self._chunk()`` multiple times,
         but is faster.
 
         Returns a list with decompressed data for each requested revision.
         """
         if not revs:
             return []
         start = self.start
         length = self.length
         inline = self._inline
         iosize = self._io.size
         buffer = util.buffer
 
         l = []
         ladd = l.append
 
         # preload the cache
         try:
             while True:
                 # ensure that the cache doesn't change out from under us
                 _cache = self._chunkcache
                 self._chunkraw(revs[0], revs[-1], df=df)
                 if _cache == self._chunkcache:
                     break
             offset, data = _cache
         except OverflowError:
             # issue4215 - we can't cache a run of chunks greater than
             # 2G on Windows
             return [self._chunk(rev, df=df) for rev in revs]
 
         for rev in revs:
             chunkstart = start(rev)
             if inline:
                 chunkstart += (rev + 1) * iosize
             chunklength = length(rev)
             ladd(decompress(buffer(data, chunkstart - offset, chunklength)))
 
         return l
 
     def _chunkclear(self):
         """Clear the raw chunk cache."""
         self._chunkcache = (0, '')
 
     def deltaparent(self, rev):
         """return deltaparent of the given revision"""
         base = self.index[rev][3]
         if base == rev:
             return nullrev
         elif self._generaldelta:
             return base
         else:
             return rev - 1
 
     def revdiff(self, rev1, rev2):
         """return or calculate a delta between two revisions"""
         if rev1 != nullrev and self.deltaparent(rev2) == rev1:
             return str(self._chunk(rev2))
 
         return mdiff.textdiff(self.revision(rev1),
                               self.revision(rev2))
 
     def revision(self, nodeorrev, _df=None):
         """return an uncompressed revision of a given node or revision
         number.
 
         _df is an existing file handle to read from. It is meant to only be
         used internally.
         """
         if isinstance(nodeorrev, int):
             rev = nodeorrev
             node = self.node(rev)
         else:
             node = nodeorrev
             rev = None
 
         cachedrev = None
         if node == nullid:
             return ""
         if self._cache:
             if self._cache[0] == node:
                 return self._cache[2]
             cachedrev = self._cache[1]
 
         # look up what we need to read
         text = None
         if rev is None:
             rev = self.rev(node)
 
         # check rev flags
         if self.flags(rev) & ~REVIDX_KNOWN_FLAGS:
             raise RevlogError(_('incompatible revision flag %x') %
                               (self.flags(rev) & ~REVIDX_KNOWN_FLAGS))
 
         # build delta chain
         chain = []
         index = self.index # for performance
         generaldelta = self._generaldelta
         iterrev = rev
         e = index[iterrev]
         while iterrev != e[3] and iterrev != cachedrev:
             chain.append(iterrev)
             if generaldelta:
                 iterrev = e[3]
             else:
                 iterrev -= 1
             e = index[iterrev]
 
         if iterrev == cachedrev:
             # cache hit
             text = self._cache[2]
         else:
             chain.append(iterrev)
         chain.reverse()
 
         # drop cache to save memory
         self._cache = None
 
         bins = self._chunks(chain, df=_df)
         if text is None:
             text = str(bins[0])
             bins = bins[1:]
 
         text = mdiff.patches(text, bins)
 
         text = self._checkhash(text, node, rev)
 
         self._cache = (node, rev, text)
         return text
 
     def hash(self, text, p1, p2):
         """Compute a node hash.
 
         Available as a function so that subclasses can replace the hash
         as needed.
         """
         return hash(text, p1, p2)
 
     def _checkhash(self, text, node, rev):
         p1, p2 = self.parents(node)
         self.checkhash(text, p1, p2, node, rev)
         return text
 
     def checkhash(self, text, p1, p2, node, rev=None):
         if node != self.hash(text, p1, p2):
             revornode = rev
             if revornode is None:
                 revornode = templatefilters.short(hex(node))
             raise RevlogError(_("integrity check failed on %s:%s")
                 % (self.indexfile, revornode))
 
     def checkinlinesize(self, tr, fp=None):
         """Check if the revlog is too big for inline and convert if so.
 
         This should be called after revisions are added to the revlog. If the
         revlog has grown too large to be an inline revlog, it will convert it
         to use multiple index and data files.
         """
         if not self._inline or (self.start(-2) + self.length(-2)) < _maxinline:
             return
 
         trinfo = tr.find(self.indexfile)
         if trinfo is None:
             raise RevlogError(_("%s not found in the transaction")
                               % self.indexfile)
 
         trindex = trinfo[2]
         if trindex is not None:
             dataoff = self.start(trindex)
         else:
             # revlog was stripped at start of transaction, use all leftover data
             trindex = len(self) - 1
             dataoff = self.end(-2)
 
         tr.add(self.datafile, dataoff)
 
         if fp:
             fp.flush()
             fp.close()
 
         df = self.opener(self.datafile, 'w')
         try:
             for r in self:
                 df.write(self._chunkraw(r, r))
         finally:
             df.close()
 
         fp = self.opener(self.indexfile, 'w', atomictemp=True)
         self.version &= ~(REVLOGNGINLINEDATA)
         self._inline = False
         for i in self:
             e = self._io.packentry(self.index[i], self.node, self.version, i)
             fp.write(e)
 
         # if we don't call close, the temp file will never replace the
         # real index
         fp.close()
 
         tr.replace(self.indexfile, trindex * self._io.size)
         self._chunkclear()
 
     def addrevision(self, text, transaction, link, p1, p2, cachedelta=None,
                     node=None):
         """add a revision to the log
 
         text - the revision data to add
         transaction - the transaction object used for rollback
         link - the linkrev data to add
         p1, p2 - the parent nodeids of the revision
         cachedelta - an optional precomputed delta
         node - nodeid of revision; typically node is not specified, and it is
             computed by default as hash(text, p1, p2), however subclasses might
             use different hashing method (and override checkhash() in such case)
         """
         if link == nullrev:
             raise RevlogError(_("attempted to add linkrev -1 to %s")
                               % self.indexfile)
 
         if len(text) > _maxentrysize:
             raise RevlogError(
                 _("%s: size of %d bytes exceeds maximum revlog storage of 2GiB")
                 % (self.indexfile, len(text)))
 
         node = node or self.hash(text, p1, p2)
         if node in self.nodemap:
             return node
 
         dfh = None
         if not self._inline:
             dfh = self.opener(self.datafile, "a+")
         ifh = self.opener(self.indexfile, "a+")
         try:
             return self._addrevision(node, text, transaction, link, p1, p2,
                                      REVIDX_DEFAULT_FLAGS, cachedelta, ifh, dfh)
         finally:
             if dfh:
                 dfh.close()
             ifh.close()
 
     def compress(self, text):
         """ generate a possibly-compressed representation of text """
         if not text:
             return ("", text)
         l = len(text)
         bin = None
         if l < 44:
             pass
         elif l > 1000000:
             # zlib makes an internal copy, thus doubling memory usage for
             # large files, so lets do this in pieces
             z = zlib.compressobj()
             p = []
             pos = 0
             while pos < l:
                 pos2 = pos + 2**20
                 p.append(z.compress(text[pos:pos2]))
                 pos = pos2
             p.append(z.flush())
             if sum(map(len, p)) < l:
                 bin = "".join(p)
         else:
             bin = _compress(text)
         if bin is None or len(bin) > l:
             if text[0] == '\0':
                 return ("", text)
             return ('u', text)
         return ("", bin)
 
     def _isgooddelta(self, d, textlen):
         """Returns True if the given delta is good. Good means that it is within
         the disk span, disk size, and chain length bounds that we know to be
         performant."""
         if d is None:
             return False
 
         # - 'dist' is the distance from the base revision -- bounding it limits
         #   the amount of I/O we need to do.
         # - 'compresseddeltalen' is the sum of the total size of deltas we need
         #   to apply -- bounding it limits the amount of CPU we consume.
         dist, l, data, base, chainbase, chainlen, compresseddeltalen = d
         if (dist > textlen * 4 or l > textlen or
             compresseddeltalen > textlen * 2 or
             (self._maxchainlen and chainlen > self._maxchainlen)):
             return False
 
         return True
 
     def _addrevision(self, node, text, transaction, link, p1, p2, flags,
                      cachedelta, ifh, dfh, alwayscache=False):
         """internal function to add revisions to the log
 
         see addrevision for argument descriptions.
         invariants:
         - text is optional (can be None); if not set, cachedelta must be set.
           if both are set, they must correspond to each other.
         """
         btext = [text]
         def buildtext():
             if btext[0] is not None:
                 return btext[0]
             baserev = cachedelta[0]
             delta = cachedelta[1]
             # special case deltas which replace entire base; no need to decode
             # base revision. this neatly avoids censored bases, which throw when
             # they're decoded.
             hlen = struct.calcsize(">lll")
             if delta[:hlen] == mdiff.replacediffheader(self.rawsize(baserev),
                                                        len(delta) - hlen):
                 btext[0] = delta[hlen:]
             else:
                 if self._inline:
                     fh = ifh
                 else:
                     fh = dfh
                 basetext = self.revision(self.node(baserev), _df=fh)
                 btext[0] = mdiff.patch(basetext, delta)
             try:
                 self.checkhash(btext[0], p1, p2, node)
                 if flags & REVIDX_ISCENSORED:
                     raise RevlogError(_('node %s is not censored') % node)
             except CensoredNodeError:
                 # must pass the censored index flag to add censored revisions
                 if not flags & REVIDX_ISCENSORED:
                     raise
             return btext[0]
 
         def builddelta(rev):
             # can we use the cached delta?
             if cachedelta and cachedelta[0] == rev:
                 delta = cachedelta[1]
             else:
                 t = buildtext()
                 if self.iscensored(rev):
                     # deltas based on a censored revision must replace the
                     # full content in one patch, so delta works everywhere
                     header = mdiff.replacediffheader(self.rawsize(rev), len(t))
                     delta = header + t
                 else:
                     if self._inline:
                         fh = ifh
                     else:
                         fh = dfh
                     ptext = self.revision(self.node(rev), _df=fh)
                     delta = mdiff.textdiff(ptext, t)
             data = self.compress(delta)
             l = len(data[1]) + len(data[0])
             if basecache[0] == rev:
                 chainbase = basecache[1]
             else:
                 chainbase = self.chainbase(rev)
             dist = l + offset - self.start(chainbase)
             if self._generaldelta:
                 base = rev
             else:
                 base = chainbase
             chainlen, compresseddeltalen = self._chaininfo(rev)
             chainlen += 1
             compresseddeltalen += l
             return dist, l, data, base, chainbase, chainlen, compresseddeltalen
 
         curr = len(self)
         prev = curr - 1
         base = chainbase = curr
         offset = self.end(prev)
         delta = None
         if self._basecache is None:
             self._basecache = (prev, self.chainbase(prev))
         basecache = self._basecache
         p1r, p2r = self.rev(p1), self.rev(p2)
 
         # full versions are inserted when the needed deltas
         # become comparable to the uncompressed text
         if text is None:
             textlen = mdiff.patchedsize(self.rawsize(cachedelta[0]),
                                         cachedelta[1])
         else:
             textlen = len(text)
 
         # should we try to build a delta?
         if prev != nullrev:
             tested = set()
             if cachedelta and self._generaldelta and self._lazydeltabase:
                 # Assume what we received from the server is a good choice
                 # build delta will reuse the cache
                 candidatedelta = builddelta(cachedelta[0])
                 tested.add(cachedelta[0])
                 if self._isgooddelta(candidatedelta, textlen):
                     delta = candidatedelta
             if delta is None and self._generaldelta:
-                parents = [p1r, p2r]
                 # exclude already lazy tested base if any
-                parents = [p for p in parents if p not in tested]
+                parents = [p for p in (p1r, p2r)
+                           if p != nullrev and p not in tested]
                 if parents and not self._aggressivemergedeltas:
                     # Pick whichever parent is closer to us (to minimize the
                     # chance of having to build a fulltext).
                     parents = [max(parents)]
                 tested.update(parents)
                 pdeltas = []
                 for p in parents:
                     pd = builddelta(p)
                     if self._isgooddelta(pd, textlen):
                         pdeltas.append(pd)
                 if pdeltas:
                     delta = min(pdeltas, key=lambda x: x[1])
             if delta is None and prev not in tested:
                 # other approach failed try against prev to hopefully save us a
                 # fulltext.
                 candidatedelta = builddelta(prev)
                 if self._isgooddelta(candidatedelta, textlen):
                     delta = candidatedelta
         if delta is not None:
             dist, l, data, base, chainbase, chainlen, compresseddeltalen = delta
         else:
             text = buildtext()
             data = self.compress(text)
             l = len(data[1]) + len(data[0])
             base = chainbase = curr
 
         e = (offset_type(offset, flags), l, textlen,
              base, link, p1r, p2r, node)
         self.index.insert(-1, e)
         self.nodemap[node] = curr
 
         entry = self._io.packentry(e, self.node, self.version, curr)
         self._writeentry(transaction, ifh, dfh, entry, data, link, offset)
 
         if alwayscache and text is None:
             text = buildtext()
 
         if type(text) == str: # only accept immutable objects
             self._cache = (node, curr, text)
         self._basecache = (curr, chainbase)
         return node
 
     def _writeentry(self, transaction, ifh, dfh, entry, data, link, offset):
         curr = len(self) - 1
         if not self._inline:
             transaction.add(self.datafile, offset)
             transaction.add(self.indexfile, curr * len(entry))
             if data[0]:
                 dfh.write(data[0])
             dfh.write(data[1])
             ifh.write(entry)
         else:
             offset += curr * self._io.size
             transaction.add(self.indexfile, offset, curr)
             ifh.write(entry)
             ifh.write(data[0])
             ifh.write(data[1])
             self.checkinlinesize(transaction, ifh)
 
     def addgroup(self, cg, linkmapper, transaction, addrevisioncb=None):
         """
         add a delta group
 
         given a set of deltas, add them to the revision log. the
         first delta is against its parent, which should be in our
         log, the rest are against the previous delta.
 
         If ``addrevisioncb`` is defined, it will be called with arguments of
         this revlog and the node that was added.
         """
 
         # track the base of the current delta log
         content = []
         node = None
 
         r = len(self)
         end = 0
         if r:
             end = self.end(r - 1)
         ifh = self.opener(self.indexfile, "a+")
         isize = r * self._io.size
         if self._inline:
             transaction.add(self.indexfile, end + isize, r)
             dfh = None
         else:
             transaction.add(self.indexfile, isize, r)
             transaction.add(self.datafile, end)
             dfh = self.opener(self.datafile, "a+")
         def flush():
             if dfh:
                 dfh.flush()
             ifh.flush()
         try:
             # loop through our set of deltas
             chain = None
             while True:
                 chunkdata = cg.deltachunk(chain)
                 if not chunkdata:
                     break
                 node = chunkdata['node']
                 p1 = chunkdata['p1']
                 p2 = chunkdata['p2']
                 cs = chunkdata['cs']
                 deltabase = chunkdata['deltabase']
                 delta = chunkdata['delta']
 
                 content.append(node)
 
                 link = linkmapper(cs)
                 if node in self.nodemap:
                     # this can happen if two branches make the same change
                     chain = node
                     continue
 
                 for p in (p1, p2):
                     if p not in self.nodemap:
                         raise LookupError(p, self.indexfile,
                                           _('unknown parent'))
 
                 if deltabase not in self.nodemap:
                     raise LookupError(deltabase, self.indexfile,
                                       _('unknown delta base'))
 
                 baserev = self.rev(deltabase)
 
                 if baserev != nullrev and self.iscensored(baserev):
                     # if base is censored, delta must be full replacement in a
                     # single patch operation
                     hlen = struct.calcsize(">lll")
                     oldlen = self.rawsize(baserev)
                     newlen = len(delta) - hlen
                     if delta[:hlen] != mdiff.replacediffheader(oldlen, newlen):
                         raise error.CensoredBaseError(self.indexfile,
                                                       self.node(baserev))
 
                 flags = REVIDX_DEFAULT_FLAGS
                 if self._peek_iscensored(baserev, delta, flush):
                     flags |= REVIDX_ISCENSORED
 
                 # We assume consumers of addrevisioncb will want to retrieve
                 # the added revision, which will require a call to
                 # revision(). revision() will fast path if there is a cache
                 # hit. So, we tell _addrevision() to always cache in this case.
                 chain = self._addrevision(node, None, transaction, link,
                                           p1, p2, flags, (baserev, delta),
                                           ifh, dfh,
                                           alwayscache=bool(addrevisioncb))
 
                 if addrevisioncb:
                     addrevisioncb(self, chain)
 
                 if not dfh and not self._inline:
                     # addrevision switched from inline to conventional
                     # reopen the index
                     ifh.close()
                     dfh = self.opener(self.datafile, "a+")
                     ifh = self.opener(self.indexfile, "a+")
         finally:
             if dfh:
                 dfh.close()
             ifh.close()
 
         return content
 
     def iscensored(self, rev):
         """Check if a file revision is censored."""
         return False
 
     def _peek_iscensored(self, baserev, delta, flush):
         """Quickly check if a delta produces a censored revision."""
         return False
 
     def getstrippoint(self, minlink):
         """find the minimum rev that must be stripped to strip the linkrev
 
         Returns a tuple containing the minimum rev and a set of all revs that
         have linkrevs that will be broken by this strip.
         """
         brokenrevs = set()
         strippoint = len(self)
 
         heads = {}
         futurelargelinkrevs = set()
         for head in self.headrevs():
             headlinkrev = self.linkrev(head)
             heads[head] = headlinkrev
             if headlinkrev >= minlink:
                 futurelargelinkrevs.add(headlinkrev)
 
         # This algorithm involves walking down the rev graph, starting at the
         # heads. Since the revs are topologically sorted according to linkrev,
         # once all head linkrevs are below the minlink, we know there are
         # no more revs that could have a linkrev greater than minlink.
         # So we can stop walking.
         while futurelargelinkrevs:
             strippoint -= 1
             linkrev = heads.pop(strippoint)
 
             if linkrev < minlink:
                 brokenrevs.add(strippoint)
             else:
                 futurelargelinkrevs.remove(linkrev)
 
             for p in self.parentrevs(strippoint):
                 if p != nullrev:
                     plinkrev = self.linkrev(p)
                     heads[p] = plinkrev
                     if plinkrev >= minlink:
                         futurelargelinkrevs.add(plinkrev)
 
         return strippoint, brokenrevs
 
     def strip(self, minlink, transaction):
         """truncate the revlog on the first revision with a linkrev >= minlink
 
         This function is called when we're stripping revision minlink and
         its descendants from the repository.
 
         We have to remove all revisions with linkrev >= minlink, because
         the equivalent changelog revisions will be renumbered after the
         strip.
 
         So we truncate the revlog on the first of these revisions, and
         trust that the caller has saved the revisions that shouldn't be
         removed and that it'll re-add them after this truncation.
         """
         if len(self) == 0:
             return
 
         rev, _ = self.getstrippoint(minlink)
         if rev == len(self):
             return
 
         # first truncate the files on disk
         end = self.start(rev)
         if not self._inline:
             transaction.add(self.datafile, end)
             end = rev * self._io.size
         else:
             end += rev * self._io.size
 
         transaction.add(self.indexfile, end)
 
         # then reset internal state in memory to forget those revisions
         self._cache = None
         self._chaininfocache = {}
         self._chunkclear()
         for x in xrange(rev, len(self)):
             del self.nodemap[self.node(x)]
 
         del self.index[rev:-1]
 
     def checksize(self):
         expected = 0
         if len(self):
             expected = max(0, self.end(len(self) - 1))
 
         try:
             f = self.opener(self.datafile)
             f.seek(0, 2)
             actual = f.tell()
             f.close()
             dd = actual - expected
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
             dd = 0
 
         try:
             f = self.opener(self.indexfile)
             f.seek(0, 2)
             actual = f.tell()
             f.close()
             s = self._io.size
             i = max(0, actual // s)
             di = actual - (i * s)
             if self._inline:
                 databytes = 0
                 for r in self:
                     databytes += max(0, self.length(r))
                 dd = 0
                 di = actual - len(self) * s - databytes
         except IOError as inst:
             if inst.errno != errno.ENOENT:
                 raise
             di = 0
 
         return (dd, di)
 
     def files(self):
         res = [self.indexfile]
         if not self._inline:
             res.append(self.datafile)
         return res

tests/test-bundle.t

@@ -1,725 +1,725 b''
 
   $ cat << EOF >> $HGRCPATH
   > [format]
   > usegeneraldelta=yes
   > EOF
 
 Setting up test
 
   $ hg init test
   $ cd test
   $ echo 0 > afile
   $ hg add afile
   $ hg commit -m "0.0"
   $ echo 1 >> afile
   $ hg commit -m "0.1"
   $ echo 2 >> afile
   $ hg commit -m "0.2"
   $ echo 3 >> afile
   $ hg commit -m "0.3"
   $ hg update -C 0
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 1 >> afile
   $ hg commit -m "1.1"
   created new head
   $ echo 2 >> afile
   $ hg commit -m "1.2"
   $ echo "a line" > fred
   $ echo 3 >> afile
   $ hg add fred
   $ hg commit -m "1.3"
   $ hg mv afile adifferentfile
   $ hg commit -m "1.3m"
   $ hg update -C 3
   1 files updated, 0 files merged, 2 files removed, 0 files unresolved
   $ hg mv afile anotherfile
   $ hg commit -m "0.3m"
   $ hg verify
   checking changesets
   checking manifests
   crosschecking files in changesets and manifests
   checking files
   4 files, 9 changesets, 7 total revisions
   $ cd ..
   $ hg init empty
 
 Bundle and phase
 
   $ hg -R test phase --force --secret 0
   $ hg -R test bundle phase.hg empty
   searching for changes
   no changes found (ignored 9 secret changesets)
   [1]
   $ hg -R test phase --draft -r 'head()'
 
 Bundle --all
 
   $ hg -R test bundle --all all.hg
   9 changesets found
 
 Bundle test to full.hg
 
   $ hg -R test bundle full.hg empty
   searching for changes
   9 changesets found
 
 Unbundle full.hg in test
 
   $ hg -R test unbundle full.hg
   adding changesets
   adding manifests
   adding file changes
   added 0 changesets with 0 changes to 4 files
   (run 'hg update' to get a working copy)
 
 Verify empty
 
   $ hg -R empty heads
   [1]
   $ hg -R empty verify
   checking changesets
   checking manifests
   crosschecking files in changesets and manifests
   checking files
   0 files, 0 changesets, 0 total revisions
 
 Pull full.hg into test (using --cwd)
 
   $ hg --cwd test pull ../full.hg
   pulling from ../full.hg
   searching for changes
   no changes found
 
 Verify that there are no leaked temporary files after pull (issue2797)
 
   $ ls test/.hg | grep .hg10un
   [1]
 
 Pull full.hg into empty (using --cwd)
 
   $ hg --cwd empty pull ../full.hg
   pulling from ../full.hg
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 7 changes to 4 files (+1 heads)
   (run 'hg heads' to see heads, 'hg merge' to merge)
 
 Rollback empty
 
   $ hg -R empty rollback
   repository tip rolled back to revision -1 (undo pull)
 
 Pull full.hg into empty again (using --cwd)
 
   $ hg --cwd empty pull ../full.hg
   pulling from ../full.hg
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 7 changes to 4 files (+1 heads)
   (run 'hg heads' to see heads, 'hg merge' to merge)
 
 Pull full.hg into test (using -R)
 
   $ hg -R test pull full.hg
   pulling from full.hg
   searching for changes
   no changes found
 
 Pull full.hg into empty (using -R)
 
   $ hg -R empty pull full.hg
   pulling from full.hg
   searching for changes
   no changes found
 
 Rollback empty
 
   $ hg -R empty rollback
   repository tip rolled back to revision -1 (undo pull)
 
 Pull full.hg into empty again (using -R)
 
   $ hg -R empty pull full.hg
   pulling from full.hg
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 7 changes to 4 files (+1 heads)
   (run 'hg heads' to see heads, 'hg merge' to merge)
 
 Log -R full.hg in fresh empty
 
   $ rm -r empty
   $ hg init empty
   $ cd empty
   $ hg -R bundle://../full.hg log
   changeset:   8:aa35859c02ea
   tag:         tip
   parent:      3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3m
   
   changeset:   7:a6a34bfa0076
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3m
   
   changeset:   6:7373c1169842
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3
   
   changeset:   5:1bb50a9436a7
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.2
   
   changeset:   4:095197eb4973
   parent:      0:f9ee2f85a263
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.1
   
   changeset:   3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3
   
   changeset:   2:e38ba6f5b7e0
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.2
   
   changeset:   1:34c2bf6b0626
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.1
   
   changeset:   0:f9ee2f85a263
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.0
   
 Make sure bundlerepo doesn't leak tempfiles (issue2491)
 
   $ ls .hg
   00changelog.i
   cache
   requires
   store
 
 Pull ../full.hg into empty (with hook)
 
   $ echo "[hooks]" >> .hg/hgrc
   $ echo "changegroup = printenv.py changegroup" >> .hg/hgrc
 
 doesn't work (yet ?)
 
 hg -R bundle://../full.hg verify
 
   $ hg pull bundle://../full.hg
   pulling from bundle:../full.hg
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 7 changes to 4 files (+1 heads)
   changegroup hook: HG_NODE=f9ee2f85a263049e9ae6d37a0e67e96194ffb735 HG_SOURCE=pull HG_TXNID=TXN:* HG_URL=bundle:../full.hg (glob)
   (run 'hg heads' to see heads, 'hg merge' to merge)
 
 Rollback empty
 
   $ hg rollback
   repository tip rolled back to revision -1 (undo pull)
   $ cd ..
 
 Log -R bundle:empty+full.hg
 
   $ hg -R bundle:empty+full.hg log --template="{rev} "; echo ""
   8 7 6 5 4 3 2 1 0 
 
 Pull full.hg into empty again (using -R; with hook)
 
   $ hg -R empty pull full.hg
   pulling from full.hg
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 7 changes to 4 files (+1 heads)
   changegroup hook: HG_NODE=f9ee2f85a263049e9ae6d37a0e67e96194ffb735 HG_SOURCE=pull HG_TXNID=TXN:* HG_URL=bundle:empty+full.hg (glob)
   (run 'hg heads' to see heads, 'hg merge' to merge)
 
 Cannot produce streaming clone bundles with "hg bundle"
 
   $ hg -R test bundle -t packed1 packed.hg
   abort: packed bundles cannot be produced by "hg bundle"
   (use "hg debugcreatestreamclonebundle")
   [255]
 
 packed1 is produced properly
 
   $ hg -R test debugcreatestreamclonebundle packed.hg
-  writing 2667 bytes for 6 files
+  writing 2663 bytes for 6 files
   bundle requirements: generaldelta, revlogv1
 
   $ f -B 64 --size --sha1 --hexdump packed.hg
-  packed.hg: size=2830, sha1=c28255110a88ffa52ddc44985cad295b1ab349bc
+  packed.hg: size=2826, sha1=e139f97692a142b19cdcff64a69697d5307ce6d4
   0000: 48 47 53 31 55 4e 00 00 00 00 00 00 00 06 00 00 |HGS1UN..........|
-  0010: 00 00 00 00 0a 6b 00 16 67 65 6e 65 72 61 6c 64 |.....k..generald|
+  0010: 00 00 00 00 0a 67 00 16 67 65 6e 65 72 61 6c 64 |.....g..generald|
   0020: 65 6c 74 61 2c 72 65 76 6c 6f 67 76 31 00 64 61 |elta,revlogv1.da|
   0030: 74 61 2f 61 64 69 66 66 65 72 65 6e 74 66 69 6c |ta/adifferentfil|
 
 generaldelta requirement is listed in stream clone bundles
 
   $ hg --config format.generaldelta=true init testgd
   $ cd testgd
   $ touch foo
   $ hg -q commit -A -m initial
   $ cd ..
   $ hg -R testgd debugcreatestreamclonebundle packedgd.hg
   writing 301 bytes for 3 files
   bundle requirements: generaldelta, revlogv1
 
   $ f -B 64 --size --sha1 --hexdump packedgd.hg
   packedgd.hg: size=396, sha1=981f9e589799335304a5a9a44caa3623a48d2a9f
   0000: 48 47 53 31 55 4e 00 00 00 00 00 00 00 03 00 00 |HGS1UN..........|
   0010: 00 00 00 00 01 2d 00 16 67 65 6e 65 72 61 6c 64 |.....-..generald|
   0020: 65 6c 74 61 2c 72 65 76 6c 6f 67 76 31 00 64 61 |elta,revlogv1.da|
   0030: 74 61 2f 66 6f 6f 2e 69 00 36 34 0a 00 03 00 01 |ta/foo.i.64.....|
 
 Unpacking packed1 bundles with "hg unbundle" isn't allowed
 
   $ hg init packed
   $ hg -R packed unbundle packed.hg
   abort: packed bundles cannot be applied with "hg unbundle"
   (use "hg debugapplystreamclonebundle")
   [255]
 
 packed1 can be consumed from debug command
 
   $ hg -R packed debugapplystreamclonebundle packed.hg
   6 files to transfer, 2.60 KB of data
   transferred 2.60 KB in *.* seconds (* */sec) (glob)
 
 Does not work on non-empty repo
 
   $ hg -R packed debugapplystreamclonebundle packed.hg
   abort: cannot apply stream clone bundle on non-empty repo
   [255]
 
 Create partial clones
 
   $ rm -r empty
   $ hg init empty
   $ hg clone -r 3 test partial
   adding changesets
   adding manifests
   adding file changes
   added 4 changesets with 4 changes to 1 files
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg clone partial partial2
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ cd partial
 
 Log -R full.hg in partial
 
   $ hg -R bundle://../full.hg log -T phases
   changeset:   8:aa35859c02ea
   tag:         tip
   phase:       draft
   parent:      3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3m
   
   changeset:   7:a6a34bfa0076
   phase:       draft
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3m
   
   changeset:   6:7373c1169842
   phase:       draft
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3
   
   changeset:   5:1bb50a9436a7
   phase:       draft
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.2
   
   changeset:   4:095197eb4973
   phase:       draft
   parent:      0:f9ee2f85a263
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.1
   
   changeset:   3:eebf5a27f8ca
   phase:       public
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3
   
   changeset:   2:e38ba6f5b7e0
   phase:       public
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.2
   
   changeset:   1:34c2bf6b0626
   phase:       public
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.1
   
   changeset:   0:f9ee2f85a263
   phase:       public
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.0
   
 
 Incoming full.hg in partial
 
   $ hg incoming bundle://../full.hg
   comparing with bundle:../full.hg
   searching for changes
   changeset:   4:095197eb4973
   parent:      0:f9ee2f85a263
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.1
   
   changeset:   5:1bb50a9436a7
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.2
   
   changeset:   6:7373c1169842
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3
   
   changeset:   7:a6a34bfa0076
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3m
   
   changeset:   8:aa35859c02ea
   tag:         tip
   parent:      3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3m
   
 
 Outgoing -R full.hg vs partial2 in partial
 
   $ hg -R bundle://../full.hg outgoing ../partial2
   comparing with ../partial2
   searching for changes
   changeset:   4:095197eb4973
   parent:      0:f9ee2f85a263
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.1
   
   changeset:   5:1bb50a9436a7
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.2
   
   changeset:   6:7373c1169842
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3
   
   changeset:   7:a6a34bfa0076
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3m
   
   changeset:   8:aa35859c02ea
   tag:         tip
   parent:      3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3m
   
 
 Outgoing -R does-not-exist.hg vs partial2 in partial
 
   $ hg -R bundle://../does-not-exist.hg outgoing ../partial2
   abort: *../does-not-exist.hg* (glob)
   [255]
   $ cd ..
 
 hide outer repo
   $ hg init
 
 Direct clone from bundle (all-history)
 
   $ hg clone full.hg full-clone
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 7 changes to 4 files (+1 heads)
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg -R full-clone heads
   changeset:   8:aa35859c02ea
   tag:         tip
   parent:      3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3m
   
   changeset:   7:a6a34bfa0076
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     1.3m
   
   $ rm -r full-clone
 
 When cloning from a non-copiable repository into '', do not
 recurse infinitely (issue2528)
 
   $ hg clone full.hg ''
   abort: empty destination path is not valid
   [255]
 
 test for https://bz.mercurial-scm.org/216
 
 Unbundle incremental bundles into fresh empty in one go
 
   $ rm -r empty
   $ hg init empty
   $ hg -R test bundle --base null -r 0 ../0.hg
   1 changesets found
   $ hg -R test bundle --base 0    -r 1 ../1.hg
   1 changesets found
   $ hg -R empty unbundle -u ../0.hg ../1.hg
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 1 changes to 1 files
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 1 changes to 1 files
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
 
 View full contents of the bundle
   $ hg -R test bundle --base null -r 3  ../partial.hg
   4 changesets found
   $ cd test
   $ hg -R ../../partial.hg log -r "bundle()"
   changeset:   0:f9ee2f85a263
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.0
   
   changeset:   1:34c2bf6b0626
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.1
   
   changeset:   2:e38ba6f5b7e0
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.2
   
   changeset:   3:eebf5a27f8ca
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     0.3
   
   $ cd ..
 
 test for 540d1059c802
 
 test for 540d1059c802
 
   $ hg init orig
   $ cd orig
   $ echo foo > foo
   $ hg add foo
   $ hg ci -m 'add foo'
 
   $ hg clone . ../copy
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg tag foo
 
   $ cd ../copy
   $ echo >> foo
   $ hg ci -m 'change foo'
   $ hg bundle ../bundle.hg ../orig
   searching for changes
   1 changesets found
 
   $ cd ../orig
   $ hg incoming ../bundle.hg
   comparing with ../bundle.hg
   searching for changes
   changeset:   2:ed1b79f46b9a
   tag:         tip
   parent:      0:bbd179dfa0a7
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     change foo
   
   $ cd ..
 
 test bundle with # in the filename (issue2154):
 
   $ cp bundle.hg 'test#bundle.hg'
   $ cd orig
   $ hg incoming '../test#bundle.hg'
   comparing with ../test
   abort: unknown revision 'bundle.hg'!
   [255]
 
 note that percent encoding is not handled:
 
   $ hg incoming ../test%23bundle.hg
   abort: repository ../test%23bundle.hg not found!
   [255]
   $ cd ..
 
 test to bundle revisions on the newly created branch (issue3828):
 
   $ hg -q clone -U test test-clone
   $ cd test
 
   $ hg -q branch foo
   $ hg commit -m "create foo branch"
   $ hg -q outgoing ../test-clone
   9:b4f5acb1ee27
   $ hg -q bundle --branch foo foo.hg ../test-clone
   $ hg -R foo.hg -q log -r "bundle()"
   9:b4f5acb1ee27
 
   $ cd ..
 
 test for https://bz.mercurial-scm.org/1144
 
 test that verify bundle does not traceback
 
 partial history bundle, fails w/ unknown parent
 
   $ hg -R bundle.hg verify
   abort: 00changelog.i@bbd179dfa0a7: unknown parent!
   [255]
 
 full history bundle, refuses to verify non-local repo
 
   $ hg -R all.hg verify
   abort: cannot verify bundle or remote repos
   [255]
 
 but, regular verify must continue to work
 
   $ hg -R orig verify
   checking changesets
   checking manifests
   crosschecking files in changesets and manifests
   checking files
   2 files, 2 changesets, 2 total revisions
 
 diff against bundle
 
   $ hg init b
   $ cd b
   $ hg -R ../all.hg diff -r tip
   diff -r aa35859c02ea anotherfile
   --- a/anotherfile	Thu Jan 01 00:00:00 1970 +0000
   +++ /dev/null	Thu Jan 01 00:00:00 1970 +0000
   @@ -1,4 +0,0 @@
   -0
   -1
   -2
   -3
   $ cd ..
 
 bundle single branch
 
   $ hg init branchy
   $ cd branchy
   $ echo a >a
   $ echo x >x
   $ hg ci -Ama
   adding a
   adding x
   $ echo c >c
   $ echo xx >x
   $ hg ci -Amc
   adding c
   $ echo c1 >c1
   $ hg ci -Amc1
   adding c1
   $ hg up 0
   1 files updated, 0 files merged, 2 files removed, 0 files unresolved
   $ echo b >b
   $ hg ci -Amb
   adding b
   created new head
   $ echo b1 >b1
   $ echo xx >x
   $ hg ci -Amb1
   adding b1
   $ hg clone -q -r2 . part
 
 == bundling via incoming
 
   $ hg in -R part --bundle incoming.hg --template "{node}\n" .
   comparing with .
   searching for changes
   1a38c1b849e8b70c756d2d80b0b9a3ac0b7ea11a
   057f4db07f61970e1c11e83be79e9d08adc4dc31
 
 == bundling
 
   $ hg bundle bundle.hg part --debug --config progress.debug=true
   query 1; heads
   searching for changes
   all remote heads known locally
   2 changesets found
   list of changesets:
   1a38c1b849e8b70c756d2d80b0b9a3ac0b7ea11a
   057f4db07f61970e1c11e83be79e9d08adc4dc31
   bundle2-output-bundle: "HG20", (1 params) 1 parts total
   bundle2-output-part: "changegroup" (params: 1 mandatory) streamed payload
   bundling: 1/2 changesets (50.00%)
   bundling: 2/2 changesets (100.00%)
   bundling: 1/2 manifests (50.00%)
   bundling: 2/2 manifests (100.00%)
   bundling: b 1/3 files (33.33%)
   bundling: b1 2/3 files (66.67%)
   bundling: x 3/3 files (100.00%)
 
 == Test for issue3441
 
   $ hg clone -q -r0 . part2
   $ hg -q -R part2 pull bundle.hg
   $ hg -R part2 verify
   checking changesets
   checking manifests
   crosschecking files in changesets and manifests
   checking files
   4 files, 3 changesets, 5 total revisions
 
   $ cd ..

tests/test-generaldelta.t

@@ -1,160 +1,160 b''
 Check whether size of generaldelta revlog is not bigger than its
 regular equivalent. Test would fail if generaldelta was naive
 implementation of parentdelta: third manifest revision would be fully
 inserted due to big distance from its paren revision (zero).
 
   $ hg init repo --config format.generaldelta=no --config format.usegeneraldelta=no
   $ cd repo
   $ echo foo > foo
   $ echo bar > bar
   $ echo baz > baz
   $ hg commit -q -Am boo
   $ hg clone --pull . ../gdrepo -q --config format.generaldelta=yes
   $ for r in 1 2 3; do
   >   echo $r > foo
   >   hg commit -q -m $r
   >   hg up -q -r 0
   >   hg pull . -q -r $r -R ../gdrepo
   > done
 
   $ cd ..
   >>> import os
   >>> regsize = os.stat("repo/.hg/store/00manifest.i").st_size
   >>> gdsize = os.stat("gdrepo/.hg/store/00manifest.i").st_size
   >>> if regsize < gdsize:
   ...     print 'generaldata increased size of manifest'
 
 Verify rev reordering doesnt create invalid bundles (issue4462)
 This requires a commit tree that when pulled will reorder manifest revs such
 that the second manifest to create a file rev will be ordered before the first
 manifest to create that file rev. We also need to do a partial pull to ensure
 reordering happens. At the end we verify the linkrev points at the earliest
 commit.
 
   $ hg init server --config format.generaldelta=True
   $ cd server
   $ touch a
   $ hg commit -Aqm a
   $ echo x > x
   $ echo y > y
   $ hg commit -Aqm xy
   $ hg up -q '.^'
   $ echo x > x
   $ echo z > z
   $ hg commit -Aqm xz
   $ hg up -q 1
   $ echo b > b
   $ hg commit -Aqm b
   $ hg merge -q 2
   $ hg commit -Aqm merge
   $ echo c > c
   $ hg commit -Aqm c
   $ hg log -G -T '{rev} {shortest(node)} {desc}'
   @  5 ebb8 c
   |
   o    4 baf7 merge
   |\
   | o  3 a129 b
   | |
   o |  2 958c xz
   | |
   | o  1 f00c xy
   |/
   o  0 3903 a
   
   $ cd ..
   $ hg init client --config format.generaldelta=false --config format.usegeneraldelta=false
   $ cd client
   $ hg pull -q ../server -r 4
   $ hg debugindex x
      rev    offset  length   base linkrev nodeid       p1           p2
        0         0       3      0       1 1406e7411862 000000000000 000000000000
 
   $ cd ..
 
 Test "usegeneraldelta" config
 (repo are general delta, but incoming bundle are not re-deltified)
 
 delta coming from the server base delta server are not recompressed.
 (also include the aggressive version for comparison)
 
   $ hg clone repo --pull --config format.usegeneraldelta=1 usegd
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 4 changesets with 6 changes to 3 files (+2 heads)
   updating to branch default
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg clone repo --pull --config format.generaldelta=1 full
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 4 changesets with 6 changes to 3 files (+2 heads)
   updating to branch default
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg -R repo debugindex -m
      rev    offset  length   base linkrev nodeid       p1           p2
        0         0     104      0       0 cef96823c800 000000000000 000000000000
        1       104      57      0       1 58ab9a8d541d cef96823c800 000000000000
        2       161      57      0       2 134fdc6fd680 cef96823c800 000000000000
        3       218     104      3       3 723508934dad cef96823c800 000000000000
   $ hg -R usegd debugindex -m
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0     104     -1       0 cef96823c800 000000000000 000000000000
        1       104      57      0       1 58ab9a8d541d cef96823c800 000000000000
        2       161      57      1       2 134fdc6fd680 cef96823c800 000000000000
        3       218      57      0       3 723508934dad cef96823c800 000000000000
   $ hg -R full debugindex -m
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0     104     -1       0 cef96823c800 000000000000 000000000000
        1       104      57      0       1 58ab9a8d541d cef96823c800 000000000000
        2       161      57      0       2 134fdc6fd680 cef96823c800 000000000000
        3       218      57      0       3 723508934dad cef96823c800 000000000000
 
 Test format.aggressivemergedeltas
 
   $ hg init --config format.generaldelta=1 aggressive
   $ cd aggressive
   $ cat << EOF >> .hg/hgrc
   > [format]
   > generaldelta = 1
   > EOF
   $ touch a b c d e
   $ hg commit -Aqm side1
   $ hg up -q null
   $ touch x y
   $ hg commit -Aqm side2
 
 - Verify non-aggressive merge uses p1 (commit 1) as delta parent
   $ hg merge -q 0
   $ hg commit -q -m merge
   $ hg debugindex -m
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0      59     -1       0 8dde941edb6e 000000000000 000000000000
-       1        59      59     -1       1 315c023f341d 000000000000 000000000000
-       2       118      65      1       2 2ab389a983eb 315c023f341d 8dde941edb6e
+       1        59      61      0       1 315c023f341d 000000000000 000000000000
+       2       120      65      1       2 2ab389a983eb 315c023f341d 8dde941edb6e
 
   $ hg strip -q -r . --config extensions.strip=
 
 - Verify aggressive merge uses p2 (commit 0) as delta parent
   $ hg up -q -C 1
   $ hg merge -q 0
   $ hg commit -q -m merge --config format.aggressivemergedeltas=True
   $ hg debugindex -m
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0      59     -1       0 8dde941edb6e 000000000000 000000000000
-       1        59      59     -1       1 315c023f341d 000000000000 000000000000
-       2       118      62      0       2 2ab389a983eb 315c023f341d 8dde941edb6e
+       1        59      61      0       1 315c023f341d 000000000000 000000000000
+       2       120      62      0       2 2ab389a983eb 315c023f341d 8dde941edb6e
 
 Test that strip bundle use bundle2
   $ hg --config extensions.strip= strip .
   0 files updated, 0 files merged, 5 files removed, 0 files unresolved
   saved backup bundle to $TESTTMP/aggressive/.hg/strip-backup/1c5d4dc9a8b8-6c68e60c-backup.hg (glob)
   $ hg debugbundle .hg/strip-backup/*
   Stream params: {'Compression': 'BZ'}
   changegroup -- "{'version': '02'}"
       1c5d4dc9a8b8d6e1750966d343e94db665e7a1e9
 
   $ cd ..

tests/test-treemanifest.t

@@ -1,390 +1,390 b''
   $ cat << EOF >> $HGRCPATH
   > [format]
   > usegeneraldelta=yes
   > EOF
 
 Set up repo
 
   $ hg --config experimental.treemanifest=True init repo
   $ cd repo
 
 Requirements get set on init
 
   $ grep treemanifest .hg/requires
   treemanifest
 
 Without directories, looks like any other repo
 
   $ echo 0 > a
   $ echo 0 > b
   $ hg ci -Aqm initial
   $ hg debugdata -m 0
   a\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
   b\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
 
 Submanifest is stored in separate revlog
 
   $ mkdir dir1
   $ echo 1 > dir1/a
   $ echo 1 > dir1/b
   $ echo 1 > e
   $ hg ci -Aqm 'add dir1'
   $ hg debugdata -m 1
   a\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
   b\x00362fef284ce2ca02aecc8de6d5e8a1c3af0556fe (esc)
   dir1\x008b3ffd73f901e83304c83d33132c8e774ceac44ed (esc)
   e\x00b8e02f6433738021a065f94175c7cd23db5f05be (esc)
   $ hg debugdata --dir dir1 0
   a\x00b8e02f6433738021a065f94175c7cd23db5f05be (esc)
   b\x00b8e02f6433738021a065f94175c7cd23db5f05be (esc)
 
 Can add nested directories
 
   $ mkdir dir1/dir1
   $ echo 2 > dir1/dir1/a
   $ echo 2 > dir1/dir1/b
   $ mkdir dir1/dir2
   $ echo 2 > dir1/dir2/a
   $ echo 2 > dir1/dir2/b
   $ hg ci -Aqm 'add dir1/dir1'
   $ hg files -r .
   a
   b
   dir1/a (glob)
   dir1/b (glob)
   dir1/dir1/a (glob)
   dir1/dir1/b (glob)
   dir1/dir2/a (glob)
   dir1/dir2/b (glob)
   e
 
 Revision is not created for unchanged directory
 
   $ mkdir dir2
   $ echo 3 > dir2/a
   $ hg add dir2
   adding dir2/a (glob)
   $ hg debugindex --dir dir1 > before
   $ hg ci -qm 'add dir2'
   $ hg debugindex --dir dir1 > after
   $ diff before after
   $ rm before after
 
 Removing directory does not create an revlog entry
 
   $ hg rm dir1/dir1
   removing dir1/dir1/a (glob)
   removing dir1/dir1/b (glob)
   $ hg debugindex --dir dir1/dir1 > before
   $ hg ci -qm 'remove dir1/dir1'
   $ hg debugindex --dir dir1/dir1 > after
   $ diff before after
   $ rm before after
 
 Check that hg files (calls treemanifest.walk()) works
 without loading all directory revlogs
 
   $ hg co 'desc("add dir2")'
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ mv .hg/store/meta/dir2 .hg/store/meta/dir2-backup
   $ hg files -r . dir1
   dir1/a (glob)
   dir1/b (glob)
   dir1/dir1/a (glob)
   dir1/dir1/b (glob)
   dir1/dir2/a (glob)
   dir1/dir2/b (glob)
 
 Check that status between revisions works (calls treemanifest.matches())
 without loading all directory revlogs
 
   $ hg status --rev 'desc("add dir1")' --rev . dir1
   A dir1/dir1/a
   A dir1/dir1/b
   A dir1/dir2/a
   A dir1/dir2/b
   $ mv .hg/store/meta/dir2-backup .hg/store/meta/dir2
 
 Merge creates 2-parent revision of directory revlog
 
   $ echo 5 > dir1/a
   $ hg ci -Aqm 'modify dir1/a'
   $ hg co '.^'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 6 > dir1/b
   $ hg ci -Aqm 'modify dir1/b'
   $ hg merge 'desc("modify dir1/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -m 'conflict-free merge involving dir1/'
   $ cat dir1/a
   5
   $ cat dir1/b
   6
   $ hg debugindex --dir dir1
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0      54     -1       1 8b3ffd73f901 000000000000 000000000000
        1        54      68      0       2 b66d046c644f 8b3ffd73f901 000000000000
        2       122      12      1       4 b87265673c8a b66d046c644f 000000000000
        3       134      55      1       5 aa5d3adcec72 b66d046c644f 000000000000
        4       189      55      1       6 e29b066b91ad b66d046c644f 000000000000
        5       244      55      4       7 a120ce2b83f5 e29b066b91ad aa5d3adcec72
 
 Merge keeping directory from parent 1 does not create revlog entry. (Note that
 dir1's manifest does change, but only because dir1/a's filelog changes.)
 
   $ hg co 'desc("add dir2")'
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 8 > dir2/a
   $ hg ci -m 'modify dir2/a'
   created new head
 
   $ hg debugindex --dir dir2 > before
   $ hg merge 'desc("modify dir1/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg revert -r 'desc("modify dir2/a")' .
   reverting dir1/a (glob)
   $ hg ci -m 'merge, keeping parent 1'
   $ hg debugindex --dir dir2 > after
   $ diff before after
   $ rm before after
 
 Merge keeping directory from parent 2 does not create revlog entry. (Note that
 dir2's manifest does change, but only because dir2/a's filelog changes.)
 
   $ hg co 'desc("modify dir2/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg debugindex --dir dir1 > before
   $ hg merge 'desc("modify dir1/a")'
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg revert -r 'desc("modify dir1/a")' .
   reverting dir2/a (glob)
   $ hg ci -m 'merge, keeping parent 2'
   created new head
   $ hg debugindex --dir dir1 > after
   $ diff before after
   $ rm before after
 
 Create flat source repo for tests with mixed flat/tree manifests
 
   $ cd ..
   $ hg init repo-flat
   $ cd repo-flat
 
 Create a few commits with flat manifest
 
   $ echo 0 > a
   $ echo 0 > b
   $ echo 0 > e
   $ for d in dir1 dir1/dir1 dir1/dir2 dir2
   > do
   >   mkdir $d
   >   echo 0 > $d/a
   >   echo 0 > $d/b
   > done
   $ hg ci -Aqm initial
 
   $ echo 1 > a
   $ echo 1 > dir1/a
   $ echo 1 > dir1/dir1/a
   $ hg ci -Aqm 'modify on branch 1'
 
   $ hg co 0
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 2 > b
   $ echo 2 > dir1/b
   $ echo 2 > dir1/dir1/b
   $ hg ci -Aqm 'modify on branch 2'
 
   $ hg merge 1
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -m 'merge of flat manifests to new flat manifest'
 
 Create clone with tree manifests enabled
 
   $ cd ..
   $ hg clone --pull --config experimental.treemanifest=1 repo-flat repo-mixed
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 4 changesets with 17 changes to 11 files
   updating to branch default
   11 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ cd repo-mixed
   $ test -f .hg/store/meta
   [1]
   $ grep treemanifest .hg/requires
   treemanifest
 
 Commit should store revlog per directory
 
   $ hg co 1
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ echo 3 > a
   $ echo 3 > dir1/a
   $ echo 3 > dir1/dir1/a
   $ hg ci -m 'first tree'
   created new head
   $ find .hg/store/meta | sort
   .hg/store/meta
   .hg/store/meta/dir1
   .hg/store/meta/dir1/00manifest.i
   .hg/store/meta/dir1/dir1
   .hg/store/meta/dir1/dir1/00manifest.i
   .hg/store/meta/dir1/dir2
   .hg/store/meta/dir1/dir2/00manifest.i
   .hg/store/meta/dir2
   .hg/store/meta/dir2/00manifest.i
 
 Merge of two trees
 
   $ hg co 2
   6 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg merge 1
   3 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -m 'merge of flat manifests to new tree manifest'
   created new head
   $ hg diff -r 3
 
 Parent of tree root manifest should be flat manifest, and two for merge
 
   $ hg debugindex -m
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0      80     -1       0 40536115ed9e 000000000000 000000000000
        1        80      83      0       1 f3376063c255 40536115ed9e 000000000000
        2       163      89      0       2 5d9b9da231a2 40536115ed9e 000000000000
        3       252      83      2       3 d17d663cbd8a 5d9b9da231a2 f3376063c255
        4       335     124      1       4 c05a51345f86 f3376063c255 000000000000
        5       459     124      2       5 82594b1f557d 5d9b9da231a2 f3376063c255
 
 
 Status across flat/tree boundary should work
 
   $ hg status --rev '.^' --rev .
   M a
   M dir1/a
   M dir1/dir1/a
 
 
 Turning off treemanifest config has no effect
 
   $ hg debugindex .hg/store/meta/dir1/00manifest.i
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0     125     -1       4 63c9c0557d24 000000000000 000000000000
-       1       125     131     -1       5 23d12a1f6e0e 000000000000 000000000000
+       1       125     109      0       5 23d12a1f6e0e 000000000000 000000000000
   $ echo 2 > dir1/a
   $ hg --config experimental.treemanifest=False ci -qm 'modify dir1/a'
   $ hg debugindex .hg/store/meta/dir1/00manifest.i
      rev    offset  length  delta linkrev nodeid       p1           p2
        0         0     125     -1       4 63c9c0557d24 000000000000 000000000000
-       1       125     131     -1       5 23d12a1f6e0e 000000000000 000000000000
-       2       256      55      1       6 3cb2d87b4250 23d12a1f6e0e 000000000000
+       1       125     109      0       5 23d12a1f6e0e 000000000000 000000000000
+       2       234      55      1       6 3cb2d87b4250 23d12a1f6e0e 000000000000
 
 Create deeper repo with tree manifests.
 
   $ cd ..
   $ hg --config experimental.treemanifest=True init deeprepo
   $ cd deeprepo
 
   $ mkdir a
   $ mkdir b
   $ mkdir b/bar
   $ mkdir b/bar/orange
   $ mkdir b/bar/orange/fly
   $ mkdir b/foo
   $ mkdir b/foo/apple
   $ mkdir b/foo/apple/bees
 
   $ touch a/one.txt
   $ touch a/two.txt
   $ touch b/bar/fruits.txt
   $ touch b/bar/orange/fly/gnat.py
   $ touch b/bar/orange/fly/housefly.txt
   $ touch b/foo/apple/bees/flower.py
   $ touch c.txt
   $ touch d.py
 
   $ hg ci -Aqm 'initial'
 
 We'll see that visitdir works by removing some treemanifest revlogs and running
 the files command with various parameters.
 
 Test files from the root.
 
   $ hg files -r .
   a/one.txt (glob)
   a/two.txt (glob)
   b/bar/fruits.txt (glob)
   b/bar/orange/fly/gnat.py (glob)
   b/bar/orange/fly/housefly.txt (glob)
   b/foo/apple/bees/flower.py (glob)
   c.txt
   d.py
 
 Excludes with a glob should not exclude everything from the glob's root
 
   $ hg files -r . -X 'b/fo?' b
   b/bar/fruits.txt (glob)
   b/bar/orange/fly/gnat.py (glob)
   b/bar/orange/fly/housefly.txt (glob)
 
 Test files for a subdirectory.
 
   $ mv .hg/store/meta/a oldmf
   $ hg files -r . b
   b/bar/fruits.txt (glob)
   b/bar/orange/fly/gnat.py (glob)
   b/bar/orange/fly/housefly.txt (glob)
   b/foo/apple/bees/flower.py (glob)
   $ mv oldmf .hg/store/meta/a
 
 Test files with just includes and excludes.
 
   $ mv .hg/store/meta/a oldmf
   $ mv .hg/store/meta/b/bar/orange/fly oldmf2
   $ mv .hg/store/meta/b/foo/apple/bees oldmf3
   $ hg files -r . -I path:b/bar -X path:b/bar/orange/fly -I path:b/foo -X path:b/foo/apple/bees
   b/bar/fruits.txt (glob)
   $ mv oldmf .hg/store/meta/a
   $ mv oldmf2 .hg/store/meta/b/bar/orange/fly
   $ mv oldmf3 .hg/store/meta/b/foo/apple/bees
 
 Test files for a subdirectory, excluding a directory within it.
 
   $ mv .hg/store/meta/a oldmf
   $ mv .hg/store/meta/b/foo oldmf2
   $ hg files -r . -X path:b/foo b
   b/bar/fruits.txt (glob)
   b/bar/orange/fly/gnat.py (glob)
   b/bar/orange/fly/housefly.txt (glob)
   $ mv oldmf .hg/store/meta/a
   $ mv oldmf2 .hg/store/meta/b/foo
 
 Test files for a sub directory, including only a directory within it, and
 including an unrelated directory.
 
   $ mv .hg/store/meta/a oldmf
   $ mv .hg/store/meta/b/foo oldmf2
   $ hg files -r . -I path:b/bar/orange -I path:a b
   b/bar/orange/fly/gnat.py (glob)
   b/bar/orange/fly/housefly.txt (glob)
   $ mv oldmf .hg/store/meta/a
   $ mv oldmf2 .hg/store/meta/b/foo
 
 Test files for a pattern, including a directory, and excluding a directory
 within that.
 
   $ mv .hg/store/meta/a oldmf
   $ mv .hg/store/meta/b/foo oldmf2
   $ mv .hg/store/meta/b/bar/orange oldmf3
   $ hg files -r . glob:**.txt -I path:b/bar -X path:b/bar/orange
   b/bar/fruits.txt (glob)
   $ mv oldmf .hg/store/meta/a
   $ mv oldmf2 .hg/store/meta/b/foo
   $ mv oldmf3 .hg/store/meta/b/bar/orange