phases: allow registration and boundary advancement with revision sets

The core internals either use revision sets already or can trivially use
them. Use the new interface in cg1unpacker.apply to avoid materializing
the list of all new nodes as it is normally just a revision range. This
avoids about 67 Bytes / changeset on AMD64 in peak RSS.

Differential Revision: https://phab.mercurial-scm.org/D9232

mercurial/changegroup.py

@@ -1,1702 +1,1703 b''
 # changegroup.py - Mercurial changegroup manipulation functions
 #
 #  Copyright 2006 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.
 
 from __future__ import absolute_import
 
 import os
 import struct
 import weakref
 
 from .i18n import _
 from .node import (
     hex,
     nullid,
     nullrev,
     short,
 )
 from .pycompat import open
 
 from . import (
     error,
     match as matchmod,
     mdiff,
     phases,
     pycompat,
     requirements,
     scmutil,
     util,
 )
 
 from .interfaces import repository
 
 _CHANGEGROUPV1_DELTA_HEADER = struct.Struct(b"20s20s20s20s")
 _CHANGEGROUPV2_DELTA_HEADER = struct.Struct(b"20s20s20s20s20s")
 _CHANGEGROUPV3_DELTA_HEADER = struct.Struct(b">20s20s20s20s20sH")
 
 LFS_REQUIREMENT = b'lfs'
 
 readexactly = util.readexactly
 
 
 def getchunk(stream):
     """return the next chunk from stream as a string"""
     d = readexactly(stream, 4)
     l = struct.unpack(b">l", d)[0]
     if l <= 4:
         if l:
             raise error.Abort(_(b"invalid chunk length %d") % l)
         return b""
     return readexactly(stream, l - 4)
 
 
 def chunkheader(length):
     """return a changegroup chunk header (string)"""
     return struct.pack(b">l", length + 4)
 
 
 def closechunk():
     """return a changegroup chunk header (string) for a zero-length chunk"""
     return struct.pack(b">l", 0)
 
 
 def _fileheader(path):
     """Obtain a changegroup chunk header for a named path."""
     return chunkheader(len(path)) + path
 
 
 def writechunks(ui, chunks, filename, vfs=None):
     """Write chunks to a file and return its filename.
 
     The stream is assumed to be a bundle file.
     Existing files will not be overwritten.
     If no filename is specified, a temporary file is created.
     """
     fh = None
     cleanup = None
     try:
         if filename:
             if vfs:
                 fh = vfs.open(filename, b"wb")
             else:
                 # Increase default buffer size because default is usually
                 # small (4k is common on Linux).
                 fh = open(filename, b"wb", 131072)
         else:
             fd, filename = pycompat.mkstemp(prefix=b"hg-bundle-", suffix=b".hg")
             fh = os.fdopen(fd, "wb")
         cleanup = filename
         for c in chunks:
             fh.write(c)
         cleanup = None
         return filename
     finally:
         if fh is not None:
             fh.close()
         if cleanup is not None:
             if filename and vfs:
                 vfs.unlink(cleanup)
             else:
                 os.unlink(cleanup)
 
 
 class cg1unpacker(object):
     """Unpacker for cg1 changegroup streams.
 
     A changegroup unpacker handles the framing of the revision data in
     the wire format. Most consumers will want to use the apply()
     method to add the changes from the changegroup to a repository.
 
     If you're forwarding a changegroup unmodified to another consumer,
     use getchunks(), which returns an iterator of changegroup
     chunks. This is mostly useful for cases where you need to know the
     data stream has ended by observing the end of the changegroup.
 
     deltachunk() is useful only if you're applying delta data. Most
     consumers should prefer apply() instead.
 
     A few other public methods exist. Those are used only for
     bundlerepo and some debug commands - their use is discouraged.
     """
 
     deltaheader = _CHANGEGROUPV1_DELTA_HEADER
     deltaheadersize = deltaheader.size
     version = b'01'
     _grouplistcount = 1  # One list of files after the manifests
 
     def __init__(self, fh, alg, extras=None):
         if alg is None:
             alg = b'UN'
         if alg not in util.compengines.supportedbundletypes:
             raise error.Abort(_(b'unknown stream compression type: %s') % alg)
         if alg == b'BZ':
             alg = b'_truncatedBZ'
 
         compengine = util.compengines.forbundletype(alg)
         self._stream = compengine.decompressorreader(fh)
         self._type = alg
         self.extras = extras or {}
         self.callback = None
 
     # These methods (compressed, read, seek, tell) all appear to only
     # be used by bundlerepo, but it's a little hard to tell.
     def compressed(self):
         return self._type is not None and self._type != b'UN'
 
     def read(self, l):
         return self._stream.read(l)
 
     def seek(self, pos):
         return self._stream.seek(pos)
 
     def tell(self):
         return self._stream.tell()
 
     def close(self):
         return self._stream.close()
 
     def _chunklength(self):
         d = readexactly(self._stream, 4)
         l = struct.unpack(b">l", d)[0]
         if l <= 4:
             if l:
                 raise error.Abort(_(b"invalid chunk length %d") % l)
             return 0
         if self.callback:
             self.callback()
         return l - 4
 
     def changelogheader(self):
         """v10 does not have a changelog header chunk"""
         return {}
 
     def manifestheader(self):
         """v10 does not have a manifest header chunk"""
         return {}
 
     def filelogheader(self):
         """return the header of the filelogs chunk, v10 only has the filename"""
         l = self._chunklength()
         if not l:
             return {}
         fname = readexactly(self._stream, l)
         return {b'filename': fname}
 
     def _deltaheader(self, headertuple, prevnode):
         node, p1, p2, cs = headertuple
         if prevnode is None:
             deltabase = p1
         else:
             deltabase = prevnode
         flags = 0
         return node, p1, p2, deltabase, cs, flags
 
     def deltachunk(self, prevnode):
         l = self._chunklength()
         if not l:
             return {}
         headerdata = readexactly(self._stream, self.deltaheadersize)
         header = self.deltaheader.unpack(headerdata)
         delta = readexactly(self._stream, l - self.deltaheadersize)
         node, p1, p2, deltabase, cs, flags = self._deltaheader(header, prevnode)
         return (node, p1, p2, cs, deltabase, delta, flags)
 
     def getchunks(self):
         """returns all the chunks contains in the bundle
 
         Used when you need to forward the binary stream to a file or another
         network API. To do so, it parse the changegroup data, otherwise it will
         block in case of sshrepo because it don't know the end of the stream.
         """
         # For changegroup 1 and 2, we expect 3 parts: changelog, manifestlog,
         # and a list of filelogs. For changegroup 3, we expect 4 parts:
         # changelog, manifestlog, a list of tree manifestlogs, and a list of
         # filelogs.
         #
         # Changelog and manifestlog parts are terminated with empty chunks. The
         # tree and file parts are a list of entry sections. Each entry section
         # is a series of chunks terminating in an empty chunk. The list of these
         # entry sections is terminated in yet another empty chunk, so we know
         # we've reached the end of the tree/file list when we reach an empty
         # chunk that was proceeded by no non-empty chunks.
 
         parts = 0
         while parts < 2 + self._grouplistcount:
             noentries = True
             while True:
                 chunk = getchunk(self)
                 if not chunk:
                     # The first two empty chunks represent the end of the
                     # changelog and the manifestlog portions. The remaining
                     # empty chunks represent either A) the end of individual
                     # tree or file entries in the file list, or B) the end of
                     # the entire list. It's the end of the entire list if there
                     # were no entries (i.e. noentries is True).
                     if parts < 2:
                         parts += 1
                     elif noentries:
                         parts += 1
                     break
                 noentries = False
                 yield chunkheader(len(chunk))
                 pos = 0
                 while pos < len(chunk):
                     next = pos + 2 ** 20
                     yield chunk[pos:next]
                     pos = next
             yield closechunk()
 
     def _unpackmanifests(self, repo, revmap, trp, prog):
         self.callback = prog.increment
         # no need to check for empty manifest group here:
         # if the result of the merge of 1 and 2 is the same in 3 and 4,
         # no new manifest will be created and the manifest group will
         # be empty during the pull
         self.manifestheader()
         deltas = self.deltaiter()
         repo.manifestlog.getstorage(b'').addgroup(deltas, revmap, trp)
         prog.complete()
         self.callback = None
 
     def apply(
         self,
         repo,
         tr,
         srctype,
         url,
         targetphase=phases.draft,
         expectedtotal=None,
     ):
         """Add the changegroup returned by source.read() to this repo.
         srctype is a string like 'push', 'pull', or 'unbundle'.  url is
         the URL of the repo where this changegroup is coming from.
 
         Return an integer summarizing the change to this repo:
         - nothing changed or no source: 0
         - more heads than before: 1+added heads (2..n)
         - fewer heads than before: -1-removed heads (-2..-n)
         - number of heads stays the same: 1
         """
         repo = repo.unfiltered()
 
         def csmap(x):
             repo.ui.debug(b"add changeset %s\n" % short(x))
             return len(cl)
 
         def revmap(x):
             return cl.rev(x)
 
         try:
             # The transaction may already carry source information. In this
             # case we use the top level data. We overwrite the argument
             # because we need to use the top level value (if they exist)
             # in this function.
             srctype = tr.hookargs.setdefault(b'source', srctype)
             tr.hookargs.setdefault(b'url', url)
             repo.hook(
                 b'prechangegroup', throw=True, **pycompat.strkwargs(tr.hookargs)
             )
 
             # write changelog data to temp files so concurrent readers
             # will not see an inconsistent view
             cl = repo.changelog
             cl.delayupdate(tr)
             oldheads = set(cl.heads())
 
             trp = weakref.proxy(tr)
             # pull off the changeset group
             repo.ui.status(_(b"adding changesets\n"))
             clstart = len(cl)
             progress = repo.ui.makeprogress(
                 _(b'changesets'), unit=_(b'chunks'), total=expectedtotal
             )
             self.callback = progress.increment
 
             efilesset = set()
             cgnodes = []
 
+            def ondupchangelog(cl, node):
+                if cl.rev(node) < clstart:
+                    cgnodes.append(node)
+
             def onchangelog(cl, node):
                 efilesset.update(cl.readfiles(node))
-                cgnodes.append(node)
-
-            def ondupchangelog(cl, node):
-                cgnodes.append(node)
 
             self.changelogheader()
             deltas = self.deltaiter()
             if not cl.addgroup(
                 deltas,
                 csmap,
                 trp,
                 addrevisioncb=onchangelog,
                 duplicaterevisioncb=ondupchangelog,
             ):
                 repo.ui.develwarn(
                     b'applied empty changelog from changegroup',
                     config=b'warn-empty-changegroup',
                 )
             efiles = len(efilesset)
             clend = len(cl)
             changesets = clend - clstart
             progress.complete()
             del deltas
             # TODO Python 2.7 removal
             # del efilesset
             efilesset = None
             self.callback = None
 
             # pull off the manifest group
             repo.ui.status(_(b"adding manifests\n"))
             # We know that we'll never have more manifests than we had
             # changesets.
             progress = repo.ui.makeprogress(
                 _(b'manifests'), unit=_(b'chunks'), total=changesets
             )
             self._unpackmanifests(repo, revmap, trp, progress)
 
             needfiles = {}
             if repo.ui.configbool(b'server', b'validate'):
                 cl = repo.changelog
                 ml = repo.manifestlog
                 # validate incoming csets have their manifests
                 for cset in pycompat.xrange(clstart, clend):
                     mfnode = cl.changelogrevision(cset).manifest
                     mfest = ml[mfnode].readdelta()
-                    # store file cgnodes we must see
+                    # store file nodes we must see
                     for f, n in pycompat.iteritems(mfest):
                         needfiles.setdefault(f, set()).add(n)
 
             # process the files
             repo.ui.status(_(b"adding file changes\n"))
             newrevs, newfiles = _addchangegroupfiles(
                 repo, self, revmap, trp, efiles, needfiles
             )
 
             # making sure the value exists
             tr.changes.setdefault(b'changegroup-count-changesets', 0)
             tr.changes.setdefault(b'changegroup-count-revisions', 0)
             tr.changes.setdefault(b'changegroup-count-files', 0)
             tr.changes.setdefault(b'changegroup-count-heads', 0)
 
             # some code use bundle operation for internal purpose. They usually
             # set `ui.quiet` to do this outside of user sight. Size the report
             # of such operation now happens at the end of the transaction, that
             # ui.quiet has not direct effect on the output.
             #
             # To preserve this intend use an inelegant hack, we fail to report
             # the change if `quiet` is set. We should probably move to
             # something better, but this is a good first step to allow the "end
             # of transaction report" to pass tests.
             if not repo.ui.quiet:
                 tr.changes[b'changegroup-count-changesets'] += changesets
                 tr.changes[b'changegroup-count-revisions'] += newrevs
                 tr.changes[b'changegroup-count-files'] += newfiles
 
             deltaheads = 0
             if oldheads:
                 heads = cl.heads()
                 deltaheads += len(heads) - len(oldheads)
                 for h in heads:
                     if h not in oldheads and repo[h].closesbranch():
                         deltaheads -= 1
 
             # see previous comment about checking ui.quiet
             if not repo.ui.quiet:
                 tr.changes[b'changegroup-count-heads'] += deltaheads
             repo.invalidatevolatilesets()
 
             if changesets > 0:
                 if b'node' not in tr.hookargs:
                     tr.hookargs[b'node'] = hex(cl.node(clstart))
                     tr.hookargs[b'node_last'] = hex(cl.node(clend - 1))
                     hookargs = dict(tr.hookargs)
                 else:
                     hookargs = dict(tr.hookargs)
                     hookargs[b'node'] = hex(cl.node(clstart))
                     hookargs[b'node_last'] = hex(cl.node(clend - 1))
                 repo.hook(
                     b'pretxnchangegroup',
                     throw=True,
                     **pycompat.strkwargs(hookargs)
                 )
 
-            added = [cl.node(r) for r in pycompat.xrange(clstart, clend)]
+            added = pycompat.xrange(clstart, clend)
             phaseall = None
             if srctype in (b'push', b'serve'):
                 # Old servers can not push the boundary themselves.
                 # New servers won't push the boundary if changeset already
                 # exists locally as secret
                 #
                 # We should not use added here but the list of all change in
                 # the bundle
                 if repo.publishing():
                     targetphase = phaseall = phases.public
                 else:
                     # closer target phase computation
 
                     # Those changesets have been pushed from the
                     # outside, their phases are going to be pushed
                     # alongside. Therefor `targetphase` is
                     # ignored.
                     targetphase = phaseall = phases.draft
             if added:
-                phases.registernew(repo, tr, targetphase, added)
+                phases.registernew(repo, tr, targetphase, [], revs=added)
             if phaseall is not None:
-                phases.advanceboundary(repo, tr, phaseall, cgnodes)
+                phases.advanceboundary(repo, tr, phaseall, cgnodes, revs=added)
+                cgnodes = []
 
             if changesets > 0:
 
                 def runhooks(unused_success):
                     # These hooks run when the lock releases, not when the
                     # transaction closes. So it's possible for the changelog
                     # to have changed since we last saw it.
                     if clstart >= len(repo):
                         return
 
                     repo.hook(b"changegroup", **pycompat.strkwargs(hookargs))
 
-                    for n in added:
+                    for rev in added:
                         args = hookargs.copy()
-                        args[b'node'] = hex(n)
+                        args[b'node'] = hex(cl.node(rev))
                         del args[b'node_last']
                         repo.hook(b"incoming", **pycompat.strkwargs(args))
 
                     newheads = [h for h in repo.heads() if h not in oldheads]
                     repo.ui.log(
                         b"incoming",
                         b"%d incoming changes - new heads: %s\n",
                         len(added),
                         b', '.join([hex(c[:6]) for c in newheads]),
                     )
 
                 tr.addpostclose(
                     b'changegroup-runhooks-%020i' % clstart,
                     lambda tr: repo._afterlock(runhooks),
                 )
         finally:
             repo.ui.flush()
         # never return 0 here:
         if deltaheads < 0:
             ret = deltaheads - 1
         else:
             ret = deltaheads + 1
         return ret
 
     def deltaiter(self):
         """
         returns an iterator of the deltas in this changegroup
 
         Useful for passing to the underlying storage system to be stored.
         """
         chain = None
         for chunkdata in iter(lambda: self.deltachunk(chain), {}):
             # Chunkdata: (node, p1, p2, cs, deltabase, delta, flags)
             yield chunkdata
             chain = chunkdata[0]
 
 
 class cg2unpacker(cg1unpacker):
     """Unpacker for cg2 streams.
 
     cg2 streams add support for generaldelta, so the delta header
     format is slightly different. All other features about the data
     remain the same.
     """
 
     deltaheader = _CHANGEGROUPV2_DELTA_HEADER
     deltaheadersize = deltaheader.size
     version = b'02'
 
     def _deltaheader(self, headertuple, prevnode):
         node, p1, p2, deltabase, cs = headertuple
         flags = 0
         return node, p1, p2, deltabase, cs, flags
 
 
 class cg3unpacker(cg2unpacker):
     """Unpacker for cg3 streams.
 
     cg3 streams add support for exchanging treemanifests and revlog
     flags. It adds the revlog flags to the delta header and an empty chunk
     separating manifests and files.
     """
 
     deltaheader = _CHANGEGROUPV3_DELTA_HEADER
     deltaheadersize = deltaheader.size
     version = b'03'
     _grouplistcount = 2  # One list of manifests and one list of files
 
     def _deltaheader(self, headertuple, prevnode):
         node, p1, p2, deltabase, cs, flags = headertuple
         return node, p1, p2, deltabase, cs, flags
 
     def _unpackmanifests(self, repo, revmap, trp, prog):
         super(cg3unpacker, self)._unpackmanifests(repo, revmap, trp, prog)
         for chunkdata in iter(self.filelogheader, {}):
             # If we get here, there are directory manifests in the changegroup
             d = chunkdata[b"filename"]
             repo.ui.debug(b"adding %s revisions\n" % d)
             deltas = self.deltaiter()
             if not repo.manifestlog.getstorage(d).addgroup(deltas, revmap, trp):
                 raise error.Abort(_(b"received dir revlog group is empty"))
 
 
 class headerlessfixup(object):
     def __init__(self, fh, h):
         self._h = h
         self._fh = fh
 
     def read(self, n):
         if self._h:
             d, self._h = self._h[:n], self._h[n:]
             if len(d) < n:
                 d += readexactly(self._fh, n - len(d))
             return d
         return readexactly(self._fh, n)
 
 
 def _revisiondeltatochunks(delta, headerfn):
     """Serialize a revisiondelta to changegroup chunks."""
 
     # The captured revision delta may be encoded as a delta against
     # a base revision or as a full revision. The changegroup format
     # requires that everything on the wire be deltas. So for full
     # revisions, we need to invent a header that says to rewrite
     # data.
 
     if delta.delta is not None:
         prefix, data = b'', delta.delta
     elif delta.basenode == nullid:
         data = delta.revision
         prefix = mdiff.trivialdiffheader(len(data))
     else:
         data = delta.revision
         prefix = mdiff.replacediffheader(delta.baserevisionsize, len(data))
 
     meta = headerfn(delta)
 
     yield chunkheader(len(meta) + len(prefix) + len(data))
     yield meta
     if prefix:
         yield prefix
     yield data
 
 
 def _sortnodesellipsis(store, nodes, cl, lookup):
     """Sort nodes for changegroup generation."""
     # Ellipses serving mode.
     #
     # In a perfect world, we'd generate better ellipsis-ified graphs
     # for non-changelog revlogs. In practice, we haven't started doing
     # that yet, so the resulting DAGs for the manifestlog and filelogs
     # are actually full of bogus parentage on all the ellipsis
     # nodes. This has the side effect that, while the contents are
     # correct, the individual DAGs might be completely out of whack in
     # a case like 882681bc3166 and its ancestors (back about 10
     # revisions or so) in the main hg repo.
     #
     # The one invariant we *know* holds is that the new (potentially
     # bogus) DAG shape will be valid if we order the nodes in the
     # order that they're introduced in dramatis personae by the
     # changelog, so what we do is we sort the non-changelog histories
     # by the order in which they are used by the changelog.
     key = lambda n: cl.rev(lookup(n))
     return sorted(nodes, key=key)
 
 
 def _resolvenarrowrevisioninfo(
     cl,
     store,
     ischangelog,
     rev,
     linkrev,
     linknode,
     clrevtolocalrev,
     fullclnodes,
     precomputedellipsis,
 ):
     linkparents = precomputedellipsis[linkrev]
 
     def local(clrev):
         """Turn a changelog revnum into a local revnum.
 
         The ellipsis dag is stored as revnums on the changelog,
         but when we're producing ellipsis entries for
         non-changelog revlogs, we need to turn those numbers into
         something local. This does that for us, and during the
         changelog sending phase will also expand the stored
         mappings as needed.
         """
         if clrev == nullrev:
             return nullrev
 
         if ischangelog:
             return clrev
 
         # Walk the ellipsis-ized changelog breadth-first looking for a
         # change that has been linked from the current revlog.
         #
         # For a flat manifest revlog only a single step should be necessary
         # as all relevant changelog entries are relevant to the flat
         # manifest.
         #
         # For a filelog or tree manifest dirlog however not every changelog
         # entry will have been relevant, so we need to skip some changelog
         # nodes even after ellipsis-izing.
         walk = [clrev]
         while walk:
             p = walk[0]
             walk = walk[1:]
             if p in clrevtolocalrev:
                 return clrevtolocalrev[p]
             elif p in fullclnodes:
                 walk.extend([pp for pp in cl.parentrevs(p) if pp != nullrev])
             elif p in precomputedellipsis:
                 walk.extend(
                     [pp for pp in precomputedellipsis[p] if pp != nullrev]
                 )
             else:
                 # In this case, we've got an ellipsis with parents
                 # outside the current bundle (likely an
                 # incremental pull). We "know" that we can use the
                 # value of this same revlog at whatever revision
                 # is pointed to by linknode. "Know" is in scare
                 # quotes because I haven't done enough examination
                 # of edge cases to convince myself this is really
                 # a fact - it works for all the (admittedly
                 # thorough) cases in our testsuite, but I would be
                 # somewhat unsurprised to find a case in the wild
                 # where this breaks down a bit. That said, I don't
                 # know if it would hurt anything.
                 for i in pycompat.xrange(rev, 0, -1):
                     if store.linkrev(i) == clrev:
                         return i
                 # We failed to resolve a parent for this node, so
                 # we crash the changegroup construction.
                 raise error.Abort(
                     b'unable to resolve parent while packing %r %r'
                     b' for changeset %r' % (store.indexfile, rev, clrev)
                 )
 
         return nullrev
 
     if not linkparents or (store.parentrevs(rev) == (nullrev, nullrev)):
         p1, p2 = nullrev, nullrev
     elif len(linkparents) == 1:
         (p1,) = sorted(local(p) for p in linkparents)
         p2 = nullrev
     else:
         p1, p2 = sorted(local(p) for p in linkparents)
 
     p1node, p2node = store.node(p1), store.node(p2)
 
     return p1node, p2node, linknode
 
 
 def deltagroup(
     repo,
     store,
     nodes,
     ischangelog,
     lookup,
     forcedeltaparentprev,
     topic=None,
     ellipses=False,
     clrevtolocalrev=None,
     fullclnodes=None,
     precomputedellipsis=None,
 ):
     """Calculate deltas for a set of revisions.
 
     Is a generator of ``revisiondelta`` instances.
 
     If topic is not None, progress detail will be generated using this
     topic name (e.g. changesets, manifests, etc).
     """
     if not nodes:
         return
 
     cl = repo.changelog
 
     if ischangelog:
         # `hg log` shows changesets in storage order. To preserve order
         # across clones, send out changesets in storage order.
         nodesorder = b'storage'
     elif ellipses:
         nodes = _sortnodesellipsis(store, nodes, cl, lookup)
         nodesorder = b'nodes'
     else:
         nodesorder = None
 
     # Perform ellipses filtering and revision massaging. We do this before
     # emitrevisions() because a) filtering out revisions creates less work
     # for emitrevisions() b) dropping revisions would break emitrevisions()'s
     # assumptions about delta choices and we would possibly send a delta
     # referencing a missing base revision.
     #
     # Also, calling lookup() has side-effects with regards to populating
     # data structures. If we don't call lookup() for each node or if we call
     # lookup() after the first pass through each node, things can break -
     # possibly intermittently depending on the python hash seed! For that
     # reason, we store a mapping of all linknodes during the initial node
     # pass rather than use lookup() on the output side.
     if ellipses:
         filtered = []
         adjustedparents = {}
         linknodes = {}
 
         for node in nodes:
             rev = store.rev(node)
             linknode = lookup(node)
             linkrev = cl.rev(linknode)
             clrevtolocalrev[linkrev] = rev
 
             # If linknode is in fullclnodes, it means the corresponding
             # changeset was a full changeset and is being sent unaltered.
             if linknode in fullclnodes:
                 linknodes[node] = linknode
 
             # If the corresponding changeset wasn't in the set computed
             # as relevant to us, it should be dropped outright.
             elif linkrev not in precomputedellipsis:
                 continue
 
             else:
                 # We could probably do this later and avoid the dict
                 # holding state. But it likely doesn't matter.
                 p1node, p2node, linknode = _resolvenarrowrevisioninfo(
                     cl,
                     store,
                     ischangelog,
                     rev,
                     linkrev,
                     linknode,
                     clrevtolocalrev,
                     fullclnodes,
                     precomputedellipsis,
                 )
 
                 adjustedparents[node] = (p1node, p2node)
                 linknodes[node] = linknode
 
             filtered.append(node)
 
         nodes = filtered
 
     # We expect the first pass to be fast, so we only engage the progress
     # meter for constructing the revision deltas.
     progress = None
     if topic is not None:
         progress = repo.ui.makeprogress(
             topic, unit=_(b'chunks'), total=len(nodes)
         )
 
     configtarget = repo.ui.config(b'devel', b'bundle.delta')
     if configtarget not in (b'', b'p1', b'full'):
         msg = _("""config "devel.bundle.delta" as unknown value: %s""")
         repo.ui.warn(msg % configtarget)
 
     deltamode = repository.CG_DELTAMODE_STD
     if forcedeltaparentprev:
         deltamode = repository.CG_DELTAMODE_PREV
     elif configtarget == b'p1':
         deltamode = repository.CG_DELTAMODE_P1
     elif configtarget == b'full':
         deltamode = repository.CG_DELTAMODE_FULL
 
     revisions = store.emitrevisions(
         nodes,
         nodesorder=nodesorder,
         revisiondata=True,
         assumehaveparentrevisions=not ellipses,
         deltamode=deltamode,
     )
 
     for i, revision in enumerate(revisions):
         if progress:
             progress.update(i + 1)
 
         if ellipses:
             linknode = linknodes[revision.node]
 
             if revision.node in adjustedparents:
                 p1node, p2node = adjustedparents[revision.node]
                 revision.p1node = p1node
                 revision.p2node = p2node
                 revision.flags |= repository.REVISION_FLAG_ELLIPSIS
 
         else:
             linknode = lookup(revision.node)
 
         revision.linknode = linknode
         yield revision
 
     if progress:
         progress.complete()
 
 
 class cgpacker(object):
     def __init__(
         self,
         repo,
         oldmatcher,
         matcher,
         version,
         builddeltaheader,
         manifestsend,
         forcedeltaparentprev=False,
         bundlecaps=None,
         ellipses=False,
         shallow=False,
         ellipsisroots=None,
         fullnodes=None,
     ):
         """Given a source repo, construct a bundler.
 
         oldmatcher is a matcher that matches on files the client already has.
         These will not be included in the changegroup.
 
         matcher is a matcher that matches on files to include in the
         changegroup. Used to facilitate sparse changegroups.
 
         forcedeltaparentprev indicates whether delta parents must be against
         the previous revision in a delta group. This should only be used for
         compatibility with changegroup version 1.
 
         builddeltaheader is a callable that constructs the header for a group
         delta.
 
         manifestsend is a chunk to send after manifests have been fully emitted.
 
         ellipses indicates whether ellipsis serving mode is enabled.
 
         bundlecaps is optional and can be used to specify the set of
         capabilities which can be used to build the bundle. While bundlecaps is
         unused in core Mercurial, extensions rely on this feature to communicate
         capabilities to customize the changegroup packer.
 
         shallow indicates whether shallow data might be sent. The packer may
         need to pack file contents not introduced by the changes being packed.
 
         fullnodes is the set of changelog nodes which should not be ellipsis
         nodes. We store this rather than the set of nodes that should be
         ellipsis because for very large histories we expect this to be
         significantly smaller.
         """
         assert oldmatcher
         assert matcher
         self._oldmatcher = oldmatcher
         self._matcher = matcher
 
         self.version = version
         self._forcedeltaparentprev = forcedeltaparentprev
         self._builddeltaheader = builddeltaheader
         self._manifestsend = manifestsend
         self._ellipses = ellipses
 
         # Set of capabilities we can use to build the bundle.
         if bundlecaps is None:
             bundlecaps = set()
         self._bundlecaps = bundlecaps
         self._isshallow = shallow
         self._fullclnodes = fullnodes
 
         # Maps ellipsis revs to their roots at the changelog level.
         self._precomputedellipsis = ellipsisroots
 
         self._repo = repo
 
         if self._repo.ui.verbose and not self._repo.ui.debugflag:
             self._verbosenote = self._repo.ui.note
         else:
             self._verbosenote = lambda s: None
 
     def generate(
         self, commonrevs, clnodes, fastpathlinkrev, source, changelog=True
     ):
         """Yield a sequence of changegroup byte chunks.
         If changelog is False, changelog data won't be added to changegroup
         """
 
         repo = self._repo
         cl = repo.changelog
 
         self._verbosenote(_(b'uncompressed size of bundle content:\n'))
         size = 0
 
         clstate, deltas = self._generatechangelog(
             cl, clnodes, generate=changelog
         )
         for delta in deltas:
             for chunk in _revisiondeltatochunks(delta, self._builddeltaheader):
                 size += len(chunk)
                 yield chunk
 
         close = closechunk()
         size += len(close)
         yield closechunk()
 
         self._verbosenote(_(b'%8.i (changelog)\n') % size)
 
         clrevorder = clstate[b'clrevorder']
         manifests = clstate[b'manifests']
         changedfiles = clstate[b'changedfiles']
 
         # We need to make sure that the linkrev in the changegroup refers to
         # the first changeset that introduced the manifest or file revision.
         # The fastpath is usually safer than the slowpath, because the filelogs
         # are walked in revlog order.
         #
         # When taking the slowpath when the manifest revlog uses generaldelta,
         # the manifest may be walked in the "wrong" order. Without 'clrevorder',
         # we would get an incorrect linkrev (see fix in cc0ff93d0c0c).
         #
         # When taking the fastpath, we are only vulnerable to reordering
         # of the changelog itself. The changelog never uses generaldelta and is
         # never reordered. To handle this case, we simply take the slowpath,
         # which already has the 'clrevorder' logic. This was also fixed in
         # cc0ff93d0c0c.
 
         # Treemanifests don't work correctly with fastpathlinkrev
         # either, because we don't discover which directory nodes to
         # send along with files. This could probably be fixed.
         fastpathlinkrev = fastpathlinkrev and not scmutil.istreemanifest(repo)
 
         fnodes = {}  # needed file nodes
 
         size = 0
         it = self.generatemanifests(
             commonrevs,
             clrevorder,
             fastpathlinkrev,
             manifests,
             fnodes,
             source,
             clstate[b'clrevtomanifestrev'],
         )
 
         for tree, deltas in it:
             if tree:
                 assert self.version == b'03'
                 chunk = _fileheader(tree)
                 size += len(chunk)
                 yield chunk
 
             for delta in deltas:
                 chunks = _revisiondeltatochunks(delta, self._builddeltaheader)
                 for chunk in chunks:
                     size += len(chunk)
                     yield chunk
 
             close = closechunk()
             size += len(close)
             yield close
 
         self._verbosenote(_(b'%8.i (manifests)\n') % size)
         yield self._manifestsend
 
         mfdicts = None
         if self._ellipses and self._isshallow:
             mfdicts = [
                 (self._repo.manifestlog[n].read(), lr)
                 for (n, lr) in pycompat.iteritems(manifests)
             ]
 
         manifests.clear()
         clrevs = {cl.rev(x) for x in clnodes}
 
         it = self.generatefiles(
             changedfiles,
             commonrevs,
             source,
             mfdicts,
             fastpathlinkrev,
             fnodes,
             clrevs,
         )
 
         for path, deltas in it:
             h = _fileheader(path)
             size = len(h)
             yield h
 
             for delta in deltas:
                 chunks = _revisiondeltatochunks(delta, self._builddeltaheader)
                 for chunk in chunks:
                     size += len(chunk)
                     yield chunk
 
             close = closechunk()
             size += len(close)
             yield close
 
             self._verbosenote(_(b'%8.i  %s\n') % (size, path))
 
         yield closechunk()
 
         if clnodes:
             repo.hook(b'outgoing', node=hex(clnodes[0]), source=source)
 
     def _generatechangelog(self, cl, nodes, generate=True):
         """Generate data for changelog chunks.
 
         Returns a 2-tuple of a dict containing state and an iterable of
         byte chunks. The state will not be fully populated until the
         chunk stream has been fully consumed.
 
         if generate is False, the state will be fully populated and no chunk
         stream will be yielded
         """
         clrevorder = {}
         manifests = {}
         mfl = self._repo.manifestlog
         changedfiles = set()
         clrevtomanifestrev = {}
 
         state = {
             b'clrevorder': clrevorder,
             b'manifests': manifests,
             b'changedfiles': changedfiles,
             b'clrevtomanifestrev': clrevtomanifestrev,
         }
 
         if not (generate or self._ellipses):
             # sort the nodes in storage order
             nodes = sorted(nodes, key=cl.rev)
             for node in nodes:
                 c = cl.changelogrevision(node)
                 clrevorder[node] = len(clrevorder)
                 # record the first changeset introducing this manifest version
                 manifests.setdefault(c.manifest, node)
                 # Record a complete list of potentially-changed files in
                 # this manifest.
                 changedfiles.update(c.files)
 
             return state, ()
 
         # Callback for the changelog, used to collect changed files and
         # manifest nodes.
         # Returns the linkrev node (identity in the changelog case).
         def lookupcl(x):
             c = cl.changelogrevision(x)
             clrevorder[x] = len(clrevorder)
 
             if self._ellipses:
                 # Only update manifests if x is going to be sent. Otherwise we
                 # end up with bogus linkrevs specified for manifests and
                 # we skip some manifest nodes that we should otherwise
                 # have sent.
                 if (
                     x in self._fullclnodes
                     or cl.rev(x) in self._precomputedellipsis
                 ):
 
                     manifestnode = c.manifest
                     # Record the first changeset introducing this manifest
                     # version.
                     manifests.setdefault(manifestnode, x)
                     # Set this narrow-specific dict so we have the lowest
                     # manifest revnum to look up for this cl revnum. (Part of
                     # mapping changelog ellipsis parents to manifest ellipsis
                     # parents)
                     clrevtomanifestrev.setdefault(
                         cl.rev(x), mfl.rev(manifestnode)
                     )
                 # We can't trust the changed files list in the changeset if the
                 # client requested a shallow clone.
                 if self._isshallow:
                     changedfiles.update(mfl[c.manifest].read().keys())
                 else:
                     changedfiles.update(c.files)
             else:
                 # record the first changeset introducing this manifest version
                 manifests.setdefault(c.manifest, x)
                 # Record a complete list of potentially-changed files in
                 # this manifest.
                 changedfiles.update(c.files)
 
             return x
 
         gen = deltagroup(
             self._repo,
             cl,
             nodes,
             True,
             lookupcl,
             self._forcedeltaparentprev,
             ellipses=self._ellipses,
             topic=_(b'changesets'),
             clrevtolocalrev={},
             fullclnodes=self._fullclnodes,
             precomputedellipsis=self._precomputedellipsis,
         )
 
         return state, gen
 
     def generatemanifests(
         self,
         commonrevs,
         clrevorder,
         fastpathlinkrev,
         manifests,
         fnodes,
         source,
         clrevtolocalrev,
     ):
         """Returns an iterator of changegroup chunks containing manifests.
 
         `source` is unused here, but is used by extensions like remotefilelog to
         change what is sent based in pulls vs pushes, etc.
         """
         repo = self._repo
         mfl = repo.manifestlog
         tmfnodes = {b'': manifests}
 
         # Callback for the manifest, used to collect linkrevs for filelog
         # revisions.
         # Returns the linkrev node (collected in lookupcl).
         def makelookupmflinknode(tree, nodes):
             if fastpathlinkrev:
                 assert not tree
                 return (
                     manifests.__getitem__
                 )  # pytype: disable=unsupported-operands
 
             def lookupmflinknode(x):
                 """Callback for looking up the linknode for manifests.
 
                 Returns the linkrev node for the specified manifest.
 
                 SIDE EFFECT:
 
                 1) fclnodes gets populated with the list of relevant
                    file nodes if we're not using fastpathlinkrev
                 2) When treemanifests are in use, collects treemanifest nodes
                    to send
 
                 Note that this means manifests must be completely sent to
                 the client before you can trust the list of files and
                 treemanifests to send.
                 """
                 clnode = nodes[x]
                 mdata = mfl.get(tree, x).readfast(shallow=True)
                 for p, n, fl in mdata.iterentries():
                     if fl == b't':  # subdirectory manifest
                         subtree = tree + p + b'/'
                         tmfclnodes = tmfnodes.setdefault(subtree, {})
                         tmfclnode = tmfclnodes.setdefault(n, clnode)
                         if clrevorder[clnode] < clrevorder[tmfclnode]:
                             tmfclnodes[n] = clnode
                     else:
                         f = tree + p
                         fclnodes = fnodes.setdefault(f, {})
                         fclnode = fclnodes.setdefault(n, clnode)
                         if clrevorder[clnode] < clrevorder[fclnode]:
                             fclnodes[n] = clnode
                 return clnode
 
             return lookupmflinknode
 
         while tmfnodes:
             tree, nodes = tmfnodes.popitem()
 
             should_visit = self._matcher.visitdir(tree[:-1])
             if tree and not should_visit:
                 continue
 
             store = mfl.getstorage(tree)
 
             if not should_visit:
                 # No nodes to send because this directory is out of
                 # the client's view of the repository (probably
                 # because of narrow clones). Do this even for the root
                 # directory (tree=='')
                 prunednodes = []
             else:
                 # Avoid sending any manifest nodes we can prove the
                 # client already has by checking linkrevs. See the
                 # related comment in generatefiles().
                 prunednodes = self._prunemanifests(store, nodes, commonrevs)
 
             if tree and not prunednodes:
                 continue
 
             lookupfn = makelookupmflinknode(tree, nodes)
 
             deltas = deltagroup(
                 self._repo,
                 store,
                 prunednodes,
                 False,
                 lookupfn,
                 self._forcedeltaparentprev,
                 ellipses=self._ellipses,
                 topic=_(b'manifests'),
                 clrevtolocalrev=clrevtolocalrev,
                 fullclnodes=self._fullclnodes,
                 precomputedellipsis=self._precomputedellipsis,
             )
 
             if not self._oldmatcher.visitdir(store.tree[:-1]):
                 yield tree, deltas
             else:
                 # 'deltas' is a generator and we need to consume it even if
                 # we are not going to send it because a side-effect is that
                 # it updates tmdnodes (via lookupfn)
                 for d in deltas:
                     pass
                 if not tree:
                     yield tree, []
 
     def _prunemanifests(self, store, nodes, commonrevs):
         if not self._ellipses:
             # In non-ellipses case and large repositories, it is better to
             # prevent calling of store.rev and store.linkrev on a lot of
             # nodes as compared to sending some extra data
             return nodes.copy()
         # This is split out as a separate method to allow filtering
         # commonrevs in extension code.
         #
         # TODO(augie): this shouldn't be required, instead we should
         # make filtering of revisions to send delegated to the store
         # layer.
         frev, flr = store.rev, store.linkrev
         return [n for n in nodes if flr(frev(n)) not in commonrevs]
 
     # The 'source' parameter is useful for extensions
     def generatefiles(
         self,
         changedfiles,
         commonrevs,
         source,
         mfdicts,
         fastpathlinkrev,
         fnodes,
         clrevs,
     ):
         changedfiles = [
             f
             for f in changedfiles
             if self._matcher(f) and not self._oldmatcher(f)
         ]
 
         if not fastpathlinkrev:
 
             def normallinknodes(unused, fname):
                 return fnodes.get(fname, {})
 
         else:
             cln = self._repo.changelog.node
 
             def normallinknodes(store, fname):
                 flinkrev = store.linkrev
                 fnode = store.node
                 revs = ((r, flinkrev(r)) for r in store)
                 return {fnode(r): cln(lr) for r, lr in revs if lr in clrevs}
 
         clrevtolocalrev = {}
 
         if self._isshallow:
             # In a shallow clone, the linknodes callback needs to also include
             # those file nodes that are in the manifests we sent but weren't
             # introduced by those manifests.
             commonctxs = [self._repo[c] for c in commonrevs]
             clrev = self._repo.changelog.rev
 
             def linknodes(flog, fname):
                 for c in commonctxs:
                     try:
                         fnode = c.filenode(fname)
                         clrevtolocalrev[c.rev()] = flog.rev(fnode)
                     except error.ManifestLookupError:
                         pass
                 links = normallinknodes(flog, fname)
                 if len(links) != len(mfdicts):
                     for mf, lr in mfdicts:
                         fnode = mf.get(fname, None)
                         if fnode in links:
                             links[fnode] = min(links[fnode], lr, key=clrev)
                         elif fnode:
                             links[fnode] = lr
                 return links
 
         else:
             linknodes = normallinknodes
 
         repo = self._repo
         progress = repo.ui.makeprogress(
             _(b'files'), unit=_(b'files'), total=len(changedfiles)
         )
         for i, fname in enumerate(sorted(changedfiles)):
             filerevlog = repo.file(fname)
             if not filerevlog:
                 raise error.Abort(
                     _(b"empty or missing file data for %s") % fname
                 )
 
             clrevtolocalrev.clear()
 
             linkrevnodes = linknodes(filerevlog, fname)
             # Lookup for filenodes, we collected the linkrev nodes above in the
             # fastpath case and with lookupmf in the slowpath case.
             def lookupfilelog(x):
                 return linkrevnodes[x]
 
             frev, flr = filerevlog.rev, filerevlog.linkrev
             # Skip sending any filenode we know the client already
             # has. This avoids over-sending files relatively
             # inexpensively, so it's not a problem if we under-filter
             # here.
             filenodes = [
                 n for n in linkrevnodes if flr(frev(n)) not in commonrevs
             ]
 
             if not filenodes:
                 continue
 
             progress.update(i + 1, item=fname)
 
             deltas = deltagroup(
                 self._repo,
                 filerevlog,
                 filenodes,
                 False,
                 lookupfilelog,
                 self._forcedeltaparentprev,
                 ellipses=self._ellipses,
                 clrevtolocalrev=clrevtolocalrev,
                 fullclnodes=self._fullclnodes,
                 precomputedellipsis=self._precomputedellipsis,
             )
 
             yield fname, deltas
 
         progress.complete()
 
 
 def _makecg1packer(
     repo,
     oldmatcher,
     matcher,
     bundlecaps,
     ellipses=False,
     shallow=False,
     ellipsisroots=None,
     fullnodes=None,
 ):
     builddeltaheader = lambda d: _CHANGEGROUPV1_DELTA_HEADER.pack(
         d.node, d.p1node, d.p2node, d.linknode
     )
 
     return cgpacker(
         repo,
         oldmatcher,
         matcher,
         b'01',
         builddeltaheader=builddeltaheader,
         manifestsend=b'',
         forcedeltaparentprev=True,
         bundlecaps=bundlecaps,
         ellipses=ellipses,
         shallow=shallow,
         ellipsisroots=ellipsisroots,
         fullnodes=fullnodes,
     )
 
 
 def _makecg2packer(
     repo,
     oldmatcher,
     matcher,
     bundlecaps,
     ellipses=False,
     shallow=False,
     ellipsisroots=None,
     fullnodes=None,
 ):
     builddeltaheader = lambda d: _CHANGEGROUPV2_DELTA_HEADER.pack(
         d.node, d.p1node, d.p2node, d.basenode, d.linknode
     )
 
     return cgpacker(
         repo,
         oldmatcher,
         matcher,
         b'02',
         builddeltaheader=builddeltaheader,
         manifestsend=b'',
         bundlecaps=bundlecaps,
         ellipses=ellipses,
         shallow=shallow,
         ellipsisroots=ellipsisroots,
         fullnodes=fullnodes,
     )
 
 
 def _makecg3packer(
     repo,
     oldmatcher,
     matcher,
     bundlecaps,
     ellipses=False,
     shallow=False,
     ellipsisroots=None,
     fullnodes=None,
 ):
     builddeltaheader = lambda d: _CHANGEGROUPV3_DELTA_HEADER.pack(
         d.node, d.p1node, d.p2node, d.basenode, d.linknode, d.flags
     )
 
     return cgpacker(
         repo,
         oldmatcher,
         matcher,
         b'03',
         builddeltaheader=builddeltaheader,
         manifestsend=closechunk(),
         bundlecaps=bundlecaps,
         ellipses=ellipses,
         shallow=shallow,
         ellipsisroots=ellipsisroots,
         fullnodes=fullnodes,
     )
 
 
 _packermap = {
     b'01': (_makecg1packer, cg1unpacker),
     # cg2 adds support for exchanging generaldelta
     b'02': (_makecg2packer, cg2unpacker),
     # cg3 adds support for exchanging revlog flags and treemanifests
     b'03': (_makecg3packer, cg3unpacker),
 }
 
 
 def allsupportedversions(repo):
     versions = set(_packermap.keys())
     needv03 = False
     if (
         repo.ui.configbool(b'experimental', b'changegroup3')
         or repo.ui.configbool(b'experimental', b'treemanifest')
         or scmutil.istreemanifest(repo)
     ):
         # we keep version 03 because we need to to exchange treemanifest data
         #
         # we also keep vresion 01 and 02, because it is possible for repo to
         # contains both normal and tree manifest at the same time. so using
         # older version to pull data is viable
         #
         # (or even to push subset of history)
         needv03 = True
     if b'exp-sidedata-flag' in repo.requirements:
         needv03 = True
         # don't attempt to use 01/02 until we do sidedata cleaning
         versions.discard(b'01')
         versions.discard(b'02')
     if not needv03:
         versions.discard(b'03')
     return versions
 
 
 # Changegroup versions that can be applied to the repo
 def supportedincomingversions(repo):
     return allsupportedversions(repo)
 
 
 # Changegroup versions that can be created from the repo
 def supportedoutgoingversions(repo):
     versions = allsupportedversions(repo)
     if scmutil.istreemanifest(repo):
         # Versions 01 and 02 support only flat manifests and it's just too
         # expensive to convert between the flat manifest and tree manifest on
         # the fly. Since tree manifests are hashed differently, all of history
         # would have to be converted. Instead, we simply don't even pretend to
         # support versions 01 and 02.
         versions.discard(b'01')
         versions.discard(b'02')
     if requirements.NARROW_REQUIREMENT in repo.requirements:
         # Versions 01 and 02 don't support revlog flags, and we need to
         # support that for stripping and unbundling to work.
         versions.discard(b'01')
         versions.discard(b'02')
     if LFS_REQUIREMENT in repo.requirements:
         # Versions 01 and 02 don't support revlog flags, and we need to
         # mark LFS entries with REVIDX_EXTSTORED.
         versions.discard(b'01')
         versions.discard(b'02')
 
     return versions
 
 
 def localversion(repo):
     # Finds the best version to use for bundles that are meant to be used
     # locally, such as those from strip and shelve, and temporary bundles.
     return max(supportedoutgoingversions(repo))
 
 
 def safeversion(repo):
     # Finds the smallest version that it's safe to assume clients of the repo
     # will support. For example, all hg versions that support generaldelta also
     # support changegroup 02.
     versions = supportedoutgoingversions(repo)
     if b'generaldelta' in repo.requirements:
         versions.discard(b'01')
     assert versions
     return min(versions)
 
 
 def getbundler(
     version,
     repo,
     bundlecaps=None,
     oldmatcher=None,
     matcher=None,
     ellipses=False,
     shallow=False,
     ellipsisroots=None,
     fullnodes=None,
 ):
     assert version in supportedoutgoingversions(repo)
 
     if matcher is None:
         matcher = matchmod.always()
     if oldmatcher is None:
         oldmatcher = matchmod.never()
 
     if version == b'01' and not matcher.always():
         raise error.ProgrammingError(
             b'version 01 changegroups do not support sparse file matchers'
         )
 
     if ellipses and version in (b'01', b'02'):
         raise error.Abort(
             _(
                 b'ellipsis nodes require at least cg3 on client and server, '
                 b'but negotiated version %s'
             )
             % version
         )
 
     # Requested files could include files not in the local store. So
     # filter those out.
     matcher = repo.narrowmatch(matcher)
 
     fn = _packermap[version][0]
     return fn(
         repo,
         oldmatcher,
         matcher,
         bundlecaps,
         ellipses=ellipses,
         shallow=shallow,
         ellipsisroots=ellipsisroots,
         fullnodes=fullnodes,
     )
 
 
 def getunbundler(version, fh, alg, extras=None):
     return _packermap[version][1](fh, alg, extras=extras)
 
 
 def _changegroupinfo(repo, nodes, source):
     if repo.ui.verbose or source == b'bundle':
         repo.ui.status(_(b"%d changesets found\n") % len(nodes))
     if repo.ui.debugflag:
         repo.ui.debug(b"list of changesets:\n")
         for node in nodes:
             repo.ui.debug(b"%s\n" % hex(node))
 
 
 def makechangegroup(
     repo, outgoing, version, source, fastpath=False, bundlecaps=None
 ):
     cgstream = makestream(
         repo,
         outgoing,
         version,
         source,
         fastpath=fastpath,
         bundlecaps=bundlecaps,
     )
     return getunbundler(
         version,
         util.chunkbuffer(cgstream),
         None,
         {b'clcount': len(outgoing.missing)},
     )
 
 
 def makestream(
     repo,
     outgoing,
     version,
     source,
     fastpath=False,
     bundlecaps=None,
     matcher=None,
 ):
     bundler = getbundler(version, repo, bundlecaps=bundlecaps, matcher=matcher)
 
     repo = repo.unfiltered()
     commonrevs = outgoing.common
     csets = outgoing.missing
     heads = outgoing.ancestorsof
     # We go through the fast path if we get told to, or if all (unfiltered
     # heads have been requested (since we then know there all linkrevs will
     # be pulled by the client).
     heads.sort()
     fastpathlinkrev = fastpath or (
         repo.filtername is None and heads == sorted(repo.heads())
     )
 
     repo.hook(b'preoutgoing', throw=True, source=source)
     _changegroupinfo(repo, csets, source)
     return bundler.generate(commonrevs, csets, fastpathlinkrev, source)
 
 
 def _addchangegroupfiles(repo, source, revmap, trp, expectedfiles, needfiles):
     revisions = 0
     files = 0
     progress = repo.ui.makeprogress(
         _(b'files'), unit=_(b'files'), total=expectedfiles
     )
     for chunkdata in iter(source.filelogheader, {}):
         files += 1
         f = chunkdata[b"filename"]
         repo.ui.debug(b"adding %s revisions\n" % f)
         progress.increment()
         fl = repo.file(f)
         o = len(fl)
         try:
             deltas = source.deltaiter()
             if not fl.addgroup(deltas, revmap, trp):
                 raise error.Abort(_(b"received file revlog group is empty"))
         except error.CensoredBaseError as e:
             raise error.Abort(_(b"received delta base is censored: %s") % e)
         revisions += len(fl) - o
         if f in needfiles:
             needs = needfiles[f]
             for new in pycompat.xrange(o, len(fl)):
                 n = fl.node(new)
                 if n in needs:
                     needs.remove(n)
                 else:
                     raise error.Abort(_(b"received spurious file revlog entry"))
             if not needs:
                 del needfiles[f]
     progress.complete()
 
     for f, needs in pycompat.iteritems(needfiles):
         fl = repo.file(f)
         for n in needs:
             try:
                 fl.rev(n)
             except error.LookupError:
                 raise error.Abort(
                     _(b'missing file data for %s:%s - run hg verify')
                     % (f, hex(n))
                 )
 
     return revisions, files

mercurial/phases.py

@@ -1,932 +1,943 b''
 """ Mercurial phases support code
 
     ---
 
     Copyright 2011 Pierre-Yves David <[e-mail]>
                    Logilab SA        <[e-mail]>
                    Augie Fackler     <[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.
 
     ---
 
 This module implements most phase logic in mercurial.
 
 
 Basic Concept
 =============
 
 A 'changeset phase' is an indicator that tells us how a changeset is
 manipulated and communicated. The details of each phase is described
 below, here we describe the properties they have in common.
 
 Like bookmarks, phases are not stored in history and thus are not
 permanent and leave no audit trail.
 
 First, no changeset can be in two phases at once. Phases are ordered,
 so they can be considered from lowest to highest. The default, lowest
 phase is 'public' - this is the normal phase of existing changesets. A
 child changeset can not be in a lower phase than its parents.
 
 These phases share a hierarchy of traits:
 
             immutable shared
     public:     X        X
     draft:               X
     secret:
 
 Local commits are draft by default.
 
 Phase Movement and Exchange
 ===========================
 
 Phase data is exchanged by pushkey on pull and push. Some servers have
 a publish option set, we call such a server a "publishing server".
 Pushing a draft changeset to a publishing server changes the phase to
 public.
 
 A small list of fact/rules define the exchange of phase:
 
 * old client never changes server states
 * pull never changes server states
 * publish and old server changesets are seen as public by client
 * any secret changeset seen in another repository is lowered to at
   least draft
 
 Here is the final table summing up the 49 possible use cases of phase
 exchange:
 
                            server
                   old     publish      non-publish
                  N   X    N   D   P    N   D   P
     old client
     pull
      N           -   X/X  -   X/D X/P  -   X/D X/P
      X           -   X/X  -   X/D X/P  -   X/D X/P
     push
      X           X/X X/X  X/P X/P X/P  X/D X/D X/P
     new client
     pull
      N           -   P/X  -   P/D P/P  -   D/D P/P
      D           -   P/X  -   P/D P/P  -   D/D P/P
      P           -   P/X  -   P/D P/P  -   P/D P/P
     push
      D           P/X P/X  P/P P/P P/P  D/D D/D P/P
      P           P/X P/X  P/P P/P P/P  P/P P/P P/P
 
 Legend:
 
     A/B = final state on client / state on server
 
     * N = new/not present,
     * P = public,
     * D = draft,
     * X = not tracked (i.e., the old client or server has no internal
           way of recording the phase.)
 
     passive = only pushes
 
 
     A cell here can be read like this:
 
     "When a new client pushes a draft changeset (D) to a publishing
     server where it's not present (N), it's marked public on both
     sides (P/P)."
 
 Note: old client behave as a publishing server with draft only content
 - other people see it as public
 - content is pushed as draft
 
 """
 
 from __future__ import absolute_import
 
 import errno
 import struct
 
 from .i18n import _
 from .node import (
     bin,
     hex,
     nullid,
     nullrev,
     short,
     wdirrev,
 )
 from .pycompat import (
     getattr,
     setattr,
 )
 from . import (
     error,
     pycompat,
     requirements,
     smartset,
     txnutil,
     util,
 )
 
 _fphasesentry = struct.Struct(b'>i20s')
 
 # record phase index
 public, draft, secret = range(3)
 archived = 32  # non-continuous for compatibility
 internal = 96  # non-continuous for compatibility
 allphases = (public, draft, secret, archived, internal)
 trackedphases = (draft, secret, archived, internal)
 # record phase names
 cmdphasenames = [b'public', b'draft', b'secret']  # known to `hg phase` command
 phasenames = dict(enumerate(cmdphasenames))
 phasenames[archived] = b'archived'
 phasenames[internal] = b'internal'
 # map phase name to phase number
 phasenumber = {name: phase for phase, name in phasenames.items()}
 # like phasenumber, but also include maps for the numeric and binary
 # phase number to the phase number
 phasenumber2 = phasenumber.copy()
 phasenumber2.update({phase: phase for phase in phasenames})
 phasenumber2.update({b'%i' % phase: phase for phase in phasenames})
 # record phase property
 mutablephases = (draft, secret, archived, internal)
 remotehiddenphases = (secret, archived, internal)
 localhiddenphases = (internal, archived)
 
 
 def supportinternal(repo):
     """True if the internal phase can be used on a repository"""
     return requirements.INTERNAL_PHASE_REQUIREMENT in repo.requirements
 
 
 def _readroots(repo, phasedefaults=None):
     """Read phase roots from disk
 
     phasedefaults is a list of fn(repo, roots) callable, which are
     executed if the phase roots file does not exist. When phases are
     being initialized on an existing repository, this could be used to
     set selected changesets phase to something else than public.
 
     Return (roots, dirty) where dirty is true if roots differ from
     what is being stored.
     """
     repo = repo.unfiltered()
     dirty = False
     roots = {i: set() for i in allphases}
     try:
         f, pending = txnutil.trypending(repo.root, repo.svfs, b'phaseroots')
         try:
             for line in f:
                 phase, nh = line.split()
                 roots[int(phase)].add(bin(nh))
         finally:
             f.close()
     except IOError as inst:
         if inst.errno != errno.ENOENT:
             raise
         if phasedefaults:
             for f in phasedefaults:
                 roots = f(repo, roots)
         dirty = True
     return roots, dirty
 
 
 def binaryencode(phasemapping):
     """encode a 'phase -> nodes' mapping into a binary stream
 
     The revision lists are encoded as (phase, root) pairs.
     """
     binarydata = []
     for phase, nodes in pycompat.iteritems(phasemapping):
         for head in nodes:
             binarydata.append(_fphasesentry.pack(phase, head))
     return b''.join(binarydata)
 
 
 def binarydecode(stream):
     """decode a binary stream into a 'phase -> nodes' mapping
 
     The (phase, root) pairs are turned back into a dictionary with
     the phase as index and the aggregated roots of that phase as value."""
     headsbyphase = {i: [] for i in allphases}
     entrysize = _fphasesentry.size
     while True:
         entry = stream.read(entrysize)
         if len(entry) < entrysize:
             if entry:
                 raise error.Abort(_(b'bad phase-heads stream'))
             break
         phase, node = _fphasesentry.unpack(entry)
         headsbyphase[phase].append(node)
     return headsbyphase
 
 
 def _sortedrange_insert(data, idx, rev, t):
     merge_before = False
     if idx:
         r1, t1 = data[idx - 1]
         merge_before = r1[-1] + 1 == rev and t1 == t
     merge_after = False
     if idx < len(data):
         r2, t2 = data[idx]
         merge_after = r2[0] == rev + 1 and t2 == t
 
     if merge_before and merge_after:
         data[idx - 1] = (pycompat.xrange(r1[0], r2[-1] + 1), t)
         data.pop(idx)
     elif merge_before:
         data[idx - 1] = (pycompat.xrange(r1[0], rev + 1), t)
     elif merge_after:
         data[idx] = (pycompat.xrange(rev, r2[-1] + 1), t)
     else:
         data.insert(idx, (pycompat.xrange(rev, rev + 1), t))
 
 
 def _sortedrange_split(data, idx, rev, t):
     r1, t1 = data[idx]
     if t == t1:
         return
     t = (t1[0], t[1])
     if len(r1) == 1:
         data.pop(idx)
         _sortedrange_insert(data, idx, rev, t)
     elif r1[0] == rev:
         data[idx] = (pycompat.xrange(rev + 1, r1[-1] + 1), t1)
         _sortedrange_insert(data, idx, rev, t)
     elif r1[-1] == rev:
         data[idx] = (pycompat.xrange(r1[0], rev), t1)
         _sortedrange_insert(data, idx + 1, rev, t)
     else:
         data[idx : idx + 1] = [
             (pycompat.xrange(r1[0], rev), t1),
             (pycompat.xrange(rev, rev + 1), t),
             (pycompat.xrange(rev + 1, r1[-1] + 1), t1),
         ]
 
 
 def _trackphasechange(data, rev, old, new):
     """add a phase move to the <data> list of ranges
 
     If data is None, nothing happens.
     """
     if data is None:
         return
 
     # If data is empty, create a one-revision range and done
     if not data:
         data.insert(0, (pycompat.xrange(rev, rev + 1), (old, new)))
         return
 
     low = 0
     high = len(data)
     t = (old, new)
     while low < high:
         mid = (low + high) // 2
         revs = data[mid][0]
         revs_low = revs[0]
         revs_high = revs[-1]
 
         if rev >= revs_low and rev <= revs_high:
             _sortedrange_split(data, mid, rev, t)
             return
 
         if revs_low == rev + 1:
             if mid and data[mid - 1][0][-1] == rev:
                 _sortedrange_split(data, mid - 1, rev, t)
             else:
                 _sortedrange_insert(data, mid, rev, t)
             return
 
         if revs_high == rev - 1:
             if mid + 1 < len(data) and data[mid + 1][0][0] == rev:
                 _sortedrange_split(data, mid + 1, rev, t)
             else:
                 _sortedrange_insert(data, mid + 1, rev, t)
             return
 
         if revs_low > rev:
             high = mid
         else:
             low = mid + 1
 
     if low == len(data):
         data.append((pycompat.xrange(rev, rev + 1), t))
         return
 
     r1, t1 = data[low]
     if r1[0] > rev:
         data.insert(low, (pycompat.xrange(rev, rev + 1), t))
     else:
         data.insert(low + 1, (pycompat.xrange(rev, rev + 1), t))
 
 
 class phasecache(object):
     def __init__(self, repo, phasedefaults, _load=True):
         if _load:
             # Cheap trick to allow shallow-copy without copy module
             self.phaseroots, self.dirty = _readroots(repo, phasedefaults)
             self._loadedrevslen = 0
             self._phasesets = None
             self.filterunknown(repo)
             self.opener = repo.svfs
 
     def hasnonpublicphases(self, repo):
         """detect if there are revisions with non-public phase"""
         repo = repo.unfiltered()
         cl = repo.changelog
         if len(cl) >= self._loadedrevslen:
             self.invalidate()
             self.loadphaserevs(repo)
         return any(
             revs
             for phase, revs in pycompat.iteritems(self.phaseroots)
             if phase != public
         )
 
     def nonpublicphaseroots(self, repo):
         """returns the roots of all non-public phases
 
         The roots are not minimized, so if the secret revisions are
         descendants of draft revisions, their roots will still be present.
         """
         repo = repo.unfiltered()
         cl = repo.changelog
         if len(cl) >= self._loadedrevslen:
             self.invalidate()
             self.loadphaserevs(repo)
         return set().union(
             *[
                 revs
                 for phase, revs in pycompat.iteritems(self.phaseroots)
                 if phase != public
             ]
         )
 
     def getrevset(self, repo, phases, subset=None):
         """return a smartset for the given phases"""
         self.loadphaserevs(repo)  # ensure phase's sets are loaded
         phases = set(phases)
         publicphase = public in phases
 
         if publicphase:
             # In this case, phases keeps all the *other* phases.
             phases = set(allphases).difference(phases)
             if not phases:
                 return smartset.fullreposet(repo)
 
         # fast path: _phasesets contains the interesting sets,
         # might only need a union and post-filtering.
         revsneedscopy = False
         if len(phases) == 1:
             [p] = phases
             revs = self._phasesets[p]
             revsneedscopy = True  # Don't modify _phasesets
         else:
             # revs has the revisions in all *other* phases.
             revs = set.union(*[self._phasesets[p] for p in phases])
 
         def _addwdir(wdirsubset, wdirrevs):
             if wdirrev in wdirsubset and repo[None].phase() in phases:
                 if revsneedscopy:
                     wdirrevs = wdirrevs.copy()
                 # The working dir would never be in the # cache, but it was in
                 # the subset being filtered for its phase (or filtered out,
                 # depending on publicphase), so add it to the output to be
                 # included (or filtered out).
                 wdirrevs.add(wdirrev)
             return wdirrevs
 
         if not publicphase:
             if repo.changelog.filteredrevs:
                 revs = revs - repo.changelog.filteredrevs
 
             if subset is None:
                 return smartset.baseset(revs)
             else:
                 revs = _addwdir(subset, revs)
                 return subset & smartset.baseset(revs)
         else:
             if subset is None:
                 subset = smartset.fullreposet(repo)
 
             revs = _addwdir(subset, revs)
 
             if not revs:
                 return subset
             return subset.filter(lambda r: r not in revs)
 
     def copy(self):
         # Shallow copy meant to ensure isolation in
         # advance/retractboundary(), nothing more.
         ph = self.__class__(None, None, _load=False)
         ph.phaseroots = self.phaseroots.copy()
         ph.dirty = self.dirty
         ph.opener = self.opener
         ph._loadedrevslen = self._loadedrevslen
         ph._phasesets = self._phasesets
         return ph
 
     def replace(self, phcache):
         """replace all values in 'self' with content of phcache"""
         for a in (
             b'phaseroots',
             b'dirty',
             b'opener',
             b'_loadedrevslen',
             b'_phasesets',
         ):
             setattr(self, a, getattr(phcache, a))
 
     def _getphaserevsnative(self, repo):
         repo = repo.unfiltered()
         return repo.changelog.computephases(self.phaseroots)
 
     def _computephaserevspure(self, repo):
         repo = repo.unfiltered()
         cl = repo.changelog
         self._phasesets = {phase: set() for phase in allphases}
         lowerroots = set()
         for phase in reversed(trackedphases):
             roots = pycompat.maplist(cl.rev, self.phaseroots[phase])
             if roots:
                 ps = set(cl.descendants(roots))
                 for root in roots:
                     ps.add(root)
                 ps.difference_update(lowerroots)
                 lowerroots.update(ps)
                 self._phasesets[phase] = ps
         self._loadedrevslen = len(cl)
 
     def loadphaserevs(self, repo):
         """ensure phase information is loaded in the object"""
         if self._phasesets is None:
             try:
                 res = self._getphaserevsnative(repo)
                 self._loadedrevslen, self._phasesets = res
             except AttributeError:
                 self._computephaserevspure(repo)
 
     def invalidate(self):
         self._loadedrevslen = 0
         self._phasesets = None
 
     def phase(self, repo, rev):
         # We need a repo argument here to be able to build _phasesets
         # if necessary. The repository instance is not stored in
         # phasecache to avoid reference cycles. The changelog instance
         # is not stored because it is a filecache() property and can
         # be replaced without us being notified.
         if rev == nullrev:
             return public
         if rev < nullrev:
             raise ValueError(_(b'cannot lookup negative revision'))
         if rev >= self._loadedrevslen:
             self.invalidate()
             self.loadphaserevs(repo)
         for phase in trackedphases:
             if rev in self._phasesets[phase]:
                 return phase
         return public
 
     def write(self):
         if not self.dirty:
             return
         f = self.opener(b'phaseroots', b'w', atomictemp=True, checkambig=True)
         try:
             self._write(f)
         finally:
             f.close()
 
     def _write(self, fp):
         for phase, roots in pycompat.iteritems(self.phaseroots):
             for h in sorted(roots):
                 fp.write(b'%i %s\n' % (phase, hex(h)))
         self.dirty = False
 
     def _updateroots(self, phase, newroots, tr):
         self.phaseroots[phase] = newroots
         self.invalidate()
         self.dirty = True
 
         tr.addfilegenerator(b'phase', (b'phaseroots',), self._write)
         tr.hookargs[b'phases_moved'] = b'1'
 
-    def registernew(self, repo, tr, targetphase, nodes):
+    def registernew(self, repo, tr, targetphase, nodes, revs=None):
+        if revs is None:
+            revs = []
         repo = repo.unfiltered()
-        self._retractboundary(repo, tr, targetphase, nodes)
+        self._retractboundary(repo, tr, targetphase, nodes, revs=revs)
         if tr is not None and b'phases' in tr.changes:
             phasetracking = tr.changes[b'phases']
             torev = repo.changelog.rev
             phase = self.phase
-            revs = [torev(node) for node in nodes]
+            revs = [torev(node) for node in nodes] + sorted(revs)
             revs.sort()
             for rev in revs:
                 revphase = phase(repo, rev)
                 _trackphasechange(phasetracking, rev, None, revphase)
         repo.invalidatevolatilesets()
 
-    def advanceboundary(self, repo, tr, targetphase, nodes, dryrun=None):
+    def advanceboundary(
+        self, repo, tr, targetphase, nodes, revs=None, dryrun=None
+    ):
         """Set all 'nodes' to phase 'targetphase'
 
         Nodes with a phase lower than 'targetphase' are not affected.
 
         If dryrun is True, no actions will be performed
 
         Returns a set of revs whose phase is changed or should be changed
         """
         # Be careful to preserve shallow-copied values: do not update
         # phaseroots values, replace them.
+        if revs is None:
+            revs = []
         if tr is None:
             phasetracking = None
         else:
             phasetracking = tr.changes.get(b'phases')
 
         repo = repo.unfiltered()
+        revs = [repo[n].rev() for n in nodes] + [r for r in revs]
 
         changes = set()  # set of revisions to be changed
         delroots = []  # set of root deleted by this path
         for phase in (phase for phase in allphases if phase > targetphase):
             # filter nodes that are not in a compatible phase already
-            nodes = [
-                n for n in nodes if self.phase(repo, repo[n].rev()) >= phase
-            ]
-            if not nodes:
+            revs = [rev for rev in revs if self.phase(repo, rev) >= phase]
+            if not revs:
                 break  # no roots to move anymore
 
             olds = self.phaseroots[phase]
 
-            affected = repo.revs(b'%ln::%ln', olds, nodes)
+            affected = repo.revs(b'%ln::%ld', olds, revs)
             changes.update(affected)
             if dryrun:
                 continue
             for r in affected:
                 _trackphasechange(
                     phasetracking, r, self.phase(repo, r), targetphase
                 )
 
             roots = {
                 ctx.node()
                 for ctx in repo.set(b'roots((%ln::) - %ld)', olds, affected)
             }
             if olds != roots:
                 self._updateroots(phase, roots, tr)
                 # some roots may need to be declared for lower phases
                 delroots.extend(olds - roots)
         if not dryrun:
             # declare deleted root in the target phase
             if targetphase != 0:
                 self._retractboundary(repo, tr, targetphase, delroots)
             repo.invalidatevolatilesets()
         return changes
 
     def retractboundary(self, repo, tr, targetphase, nodes):
         oldroots = {
             phase: revs
             for phase, revs in pycompat.iteritems(self.phaseroots)
             if phase <= targetphase
         }
         if tr is None:
             phasetracking = None
         else:
             phasetracking = tr.changes.get(b'phases')
         repo = repo.unfiltered()
         if (
             self._retractboundary(repo, tr, targetphase, nodes)
             and phasetracking is not None
         ):
 
             # find the affected revisions
             new = self.phaseroots[targetphase]
             old = oldroots[targetphase]
             affected = set(repo.revs(b'(%ln::) - (%ln::)', new, old))
 
             # find the phase of the affected revision
             for phase in pycompat.xrange(targetphase, -1, -1):
                 if phase:
                     roots = oldroots.get(phase, [])
                     revs = set(repo.revs(b'%ln::%ld', roots, affected))
                     affected -= revs
                 else:  # public phase
                     revs = affected
                 for r in sorted(revs):
                     _trackphasechange(phasetracking, r, phase, targetphase)
         repo.invalidatevolatilesets()
 
-    def _retractboundary(self, repo, tr, targetphase, nodes):
+    def _retractboundary(self, repo, tr, targetphase, nodes, revs=None):
         # Be careful to preserve shallow-copied values: do not update
         # phaseroots values, replace them.
+        if revs is None:
+            revs = []
         if targetphase in (archived, internal) and not supportinternal(repo):
             name = phasenames[targetphase]
             msg = b'this repository does not support the %s phase' % name
             raise error.ProgrammingError(msg)
 
         repo = repo.unfiltered()
         torev = repo.changelog.rev
         tonode = repo.changelog.node
         currentroots = {torev(node) for node in self.phaseroots[targetphase]}
         finalroots = oldroots = set(currentroots)
-        newroots = [torev(node) for node in nodes]
+        newroots = [torev(node) for node in nodes] + [r for r in revs]
         newroots = [
             rev for rev in newroots if self.phase(repo, rev) < targetphase
         ]
 
         if newroots:
             if nullrev in newroots:
                 raise error.Abort(_(b'cannot change null revision phase'))
             currentroots.update(newroots)
 
             # Only compute new roots for revs above the roots that are being
             # retracted.
             minnewroot = min(newroots)
             aboveroots = [rev for rev in currentroots if rev >= minnewroot]
             updatedroots = repo.revs(b'roots(%ld::)', aboveroots)
 
             finalroots = {rev for rev in currentroots if rev < minnewroot}
             finalroots.update(updatedroots)
         if finalroots != oldroots:
             self._updateroots(
                 targetphase, {tonode(rev) for rev in finalroots}, tr
             )
             return True
         return False
 
     def filterunknown(self, repo):
         """remove unknown nodes from the phase boundary
 
         Nothing is lost as unknown nodes only hold data for their descendants.
         """
         filtered = False
         has_node = repo.changelog.index.has_node  # to filter unknown nodes
         for phase, nodes in pycompat.iteritems(self.phaseroots):
             missing = sorted(node for node in nodes if not has_node(node))
             if missing:
                 for mnode in missing:
                     repo.ui.debug(
                         b'removing unknown node %s from %i-phase boundary\n'
                         % (short(mnode), phase)
                     )
                 nodes.symmetric_difference_update(missing)
                 filtered = True
         if filtered:
             self.dirty = True
         # filterunknown is called by repo.destroyed, we may have no changes in
         # root but _phasesets contents is certainly invalid (or at least we
         # have not proper way to check that). related to issue 3858.
         #
         # The other caller is __init__ that have no _phasesets initialized
         # anyway. If this change we should consider adding a dedicated
         # "destroyed" function to phasecache or a proper cache key mechanism
         # (see branchmap one)
         self.invalidate()
 
 
-def advanceboundary(repo, tr, targetphase, nodes, dryrun=None):
+def advanceboundary(repo, tr, targetphase, nodes, revs=None, dryrun=None):
     """Add nodes to a phase changing other nodes phases if necessary.
 
     This function move boundary *forward* this means that all nodes
     are set in the target phase or kept in a *lower* phase.
 
     Simplify boundary to contains phase roots only.
 
     If dryrun is True, no actions will be performed
 
     Returns a set of revs whose phase is changed or should be changed
     """
+    if revs is None:
+        revs = []
     phcache = repo._phasecache.copy()
     changes = phcache.advanceboundary(
-        repo, tr, targetphase, nodes, dryrun=dryrun
+        repo, tr, targetphase, nodes, revs=revs, dryrun=dryrun
     )
     if not dryrun:
         repo._phasecache.replace(phcache)
     return changes
 
 
 def retractboundary(repo, tr, targetphase, nodes):
     """Set nodes back to a phase changing other nodes phases if
     necessary.
 
     This function move boundary *backward* this means that all nodes
     are set in the target phase or kept in a *higher* phase.
 
     Simplify boundary to contains phase roots only."""
     phcache = repo._phasecache.copy()
     phcache.retractboundary(repo, tr, targetphase, nodes)
     repo._phasecache.replace(phcache)
 
 
-def registernew(repo, tr, targetphase, nodes):
+def registernew(repo, tr, targetphase, nodes, revs=None):
     """register a new revision and its phase
 
     Code adding revisions to the repository should use this function to
     set new changeset in their target phase (or higher).
     """
+    if revs is None:
+        revs = []
     phcache = repo._phasecache.copy()
-    phcache.registernew(repo, tr, targetphase, nodes)
+    phcache.registernew(repo, tr, targetphase, nodes, revs=revs)
     repo._phasecache.replace(phcache)
 
 
 def listphases(repo):
     """List phases root for serialization over pushkey"""
     # Use ordered dictionary so behavior is deterministic.
     keys = util.sortdict()
     value = b'%i' % draft
     cl = repo.unfiltered().changelog
     for root in repo._phasecache.phaseroots[draft]:
         if repo._phasecache.phase(repo, cl.rev(root)) <= draft:
             keys[hex(root)] = value
 
     if repo.publishing():
         # Add an extra data to let remote know we are a publishing
         # repo. Publishing repo can't just pretend they are old repo.
         # When pushing to a publishing repo, the client still need to
         # push phase boundary
         #
         # Push do not only push changeset. It also push phase data.
         # New phase data may apply to common changeset which won't be
         # push (as they are common). Here is a very simple example:
         #
         # 1) repo A push changeset X as draft to repo B
         # 2) repo B make changeset X public
         # 3) repo B push to repo A. X is not pushed but the data that
         #    X as now public should
         #
         # The server can't handle it on it's own as it has no idea of
         # client phase data.
         keys[b'publishing'] = b'True'
     return keys
 
 
 def pushphase(repo, nhex, oldphasestr, newphasestr):
     """List phases root for serialization over pushkey"""
     repo = repo.unfiltered()
     with repo.lock():
         currentphase = repo[nhex].phase()
         newphase = abs(int(newphasestr))  # let's avoid negative index surprise
         oldphase = abs(int(oldphasestr))  # let's avoid negative index surprise
         if currentphase == oldphase and newphase < oldphase:
             with repo.transaction(b'pushkey-phase') as tr:
                 advanceboundary(repo, tr, newphase, [bin(nhex)])
             return True
         elif currentphase == newphase:
             # raced, but got correct result
             return True
         else:
             return False
 
 
 def subsetphaseheads(repo, subset):
     """Finds the phase heads for a subset of a history
 
     Returns a list indexed by phase number where each item is a list of phase
     head nodes.
     """
     cl = repo.changelog
 
     headsbyphase = {i: [] for i in allphases}
     # No need to keep track of secret phase; any heads in the subset that
     # are not mentioned are implicitly secret.
     for phase in allphases[:secret]:
         revset = b"heads(%%ln & %s())" % phasenames[phase]
         headsbyphase[phase] = [cl.node(r) for r in repo.revs(revset, subset)]
     return headsbyphase
 
 
 def updatephases(repo, trgetter, headsbyphase):
     """Updates the repo with the given phase heads"""
     # Now advance phase boundaries of all phases
     #
     # run the update (and fetch transaction) only if there are actually things
     # to update. This avoid creating empty transaction during no-op operation.
 
     for phase in allphases:
         revset = b'%ln - _phase(%s)'
         heads = [c.node() for c in repo.set(revset, headsbyphase[phase], phase)]
         if heads:
             advanceboundary(repo, trgetter(), phase, heads)
 
 
 def analyzeremotephases(repo, subset, roots):
     """Compute phases heads and root in a subset of node from root dict
 
     * subset is heads of the subset
     * roots is {<nodeid> => phase} mapping. key and value are string.
 
     Accept unknown element input
     """
     repo = repo.unfiltered()
     # build list from dictionary
     draftroots = []
     has_node = repo.changelog.index.has_node  # to filter unknown nodes
     for nhex, phase in pycompat.iteritems(roots):
         if nhex == b'publishing':  # ignore data related to publish option
             continue
         node = bin(nhex)
         phase = int(phase)
         if phase == public:
             if node != nullid:
                 repo.ui.warn(
                     _(
                         b'ignoring inconsistent public root'
                         b' from remote: %s\n'
                     )
                     % nhex
                 )
         elif phase == draft:
             if has_node(node):
                 draftroots.append(node)
         else:
             repo.ui.warn(
                 _(b'ignoring unexpected root from remote: %i %s\n')
                 % (phase, nhex)
             )
     # compute heads
     publicheads = newheads(repo, subset, draftroots)
     return publicheads, draftroots
 
 
 class remotephasessummary(object):
     """summarize phase information on the remote side
 
     :publishing: True is the remote is publishing
     :publicheads: list of remote public phase heads (nodes)
     :draftheads: list of remote draft phase heads (nodes)
     :draftroots: list of remote draft phase root (nodes)
     """
 
     def __init__(self, repo, remotesubset, remoteroots):
         unfi = repo.unfiltered()
         self._allremoteroots = remoteroots
 
         self.publishing = remoteroots.get(b'publishing', False)
 
         ana = analyzeremotephases(repo, remotesubset, remoteroots)
         self.publicheads, self.draftroots = ana
         # Get the list of all "heads" revs draft on remote
         dheads = unfi.set(b'heads(%ln::%ln)', self.draftroots, remotesubset)
         self.draftheads = [c.node() for c in dheads]
 
 
 def newheads(repo, heads, roots):
     """compute new head of a subset minus another
 
     * `heads`: define the first subset
     * `roots`: define the second we subtract from the first"""
     # prevent an import cycle
     # phases > dagop > patch > copies > scmutil > obsolete > obsutil > phases
     from . import dagop
 
     repo = repo.unfiltered()
     cl = repo.changelog
     rev = cl.index.get_rev
     if not roots:
         return heads
     if not heads or heads == [nullid]:
         return []
     # The logic operated on revisions, convert arguments early for convenience
     new_heads = {rev(n) for n in heads if n != nullid}
     roots = [rev(n) for n in roots]
     # compute the area we need to remove
     affected_zone = repo.revs(b"(%ld::%ld)", roots, new_heads)
     # heads in the area are no longer heads
     new_heads.difference_update(affected_zone)
     # revisions in the area have children outside of it,
     # They might be new heads
     candidates = repo.revs(
         b"parents(%ld + (%ld and merge())) and not null", roots, affected_zone
     )
     candidates -= affected_zone
     if new_heads or candidates:
         # remove candidate that are ancestors of other heads
         new_heads.update(candidates)
         prunestart = repo.revs(b"parents(%ld) and not null", new_heads)
         pruned = dagop.reachableroots(repo, candidates, prunestart)
         new_heads.difference_update(pruned)
 
     return pycompat.maplist(cl.node, sorted(new_heads))
 
 
 def newcommitphase(ui):
     """helper to get the target phase of new commit
 
     Handle all possible values for the phases.new-commit options.
 
     """
     v = ui.config(b'phases', b'new-commit')
     try:
         return phasenumber2[v]
     except KeyError:
         raise error.ConfigError(
             _(b"phases.new-commit: not a valid phase name ('%s')") % v
         )
 
 
 def hassecret(repo):
     """utility function that check if a repo have any secret changeset."""
     return bool(repo._phasecache.phaseroots[secret])
 
 
 def preparehookargs(node, old, new):
     if old is None:
         old = b''
     else:
         old = phasenames[old]
     return {b'node': node, b'oldphase': old, b'phase': phasenames[new]}