Mercurial > hg-stable
changeset 50694:a41eeb877d07
branching: merge with stable
| author | Pierre-Yves David <pierre-yves.david@octobus.net> |
|---|---|
| date | Sun, 18 Jun 2023 00:09:39 +0200 |
| parents | b9a355763e76 (current diff) e7a3611181aa (diff) |
| children | 1c31b343e514 |
| files | hgext/clonebundles.py mercurial/bundlecaches.py mercurial/dirstatemap.py mercurial/revlog.py mercurial/revlogutils/deltas.py mercurial/transaction.py mercurial/wireprotov1server.py tests/test-clonebundles.t tests/test-contrib-perf.t tests/test-dirstate-version-fallback.t tests/test-transaction-rollback-on-revlog-split.t tests/test-upgrade-repo.t |
| diffstat | 20 files changed, 651 insertions(+), 264 deletions(-) [+] |
line wrap: on
line diff
--- a/.hgsigs Tue Jun 20 02:36:52 2023 +0200 +++ b/.hgsigs Sun Jun 18 00:09:39 2023 +0200 @@ -244,3 +244,4 @@ 83ea6ce48b4fd09fb79c4e34cc5750c805699a53 0 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 f952be90b0514a576dcc8bbe758ce3847faba9bb 0 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 fc445f8abcf90b33db7c463816a1b3560681767f 0 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 +da372c745e0f053bb7a64e74cccd15810d96341d 0 iQHNBAABCgA3FiEEH2b4zfZU6QXBHaBhoR4BzQ4F2VYFAmSB7WkZHGFscGhhcmVAcmFwaGFlbGdvbWVzLmRldgAKCRChHgHNDgXZVoy+C/4zwO+Wxc3wr0aEzjVqAss7FuGS5e66H+0T3WzVgKIRMqiiOmUmmiNf+XloXlX4TOwoh9j9GNEpoZfV6TSwFSqV0LALaVIRRwrkJBDhnqw4eNBZbK5aBWNa2/21dkHecxF4KG3ai9kLwy2mtHxkDIy8T2LPvdx8pfNcYT4PZ19x2itqZLouBJqiZYehsqeMLNF2vRqkq+rQ+D2sFGLljgPo0JlpkOZ4IL7S/cqTOBG1sQ6KJK+hAE1kF1lhvK796VhKKXVnWVgqJLyg7ZI6168gxeFv5cyCtb+FUXJJ/5SOkxaCKJf3mg3DIYi3G7xjwB5CfUGW8A2qexgEjXeV42Mu7/Mkmn/aeTdL0UcRK3oBVHJwqt/fJlGFqVWt4/9g9KW5mJvTDQYBo/zjLyvKFEbnSLzhEP+9SvthCrtX0UYkKxOGi2M2Z7e9wgBB0gY8a36kA739lkNu6r3vH/FVh0aPTMWukLToELS90WgfViNr16lDnCeDjMgg97OKxWdOW6U=
--- a/.hgtags Tue Jun 20 02:36:52 2023 +0200 +++ b/.hgtags Sun Jun 18 00:09:39 2023 +0200 @@ -260,3 +260,4 @@ 83ea6ce48b4fd09fb79c4e34cc5750c805699a53 6.4.1 f952be90b0514a576dcc8bbe758ce3847faba9bb 6.4.2 fc445f8abcf90b33db7c463816a1b3560681767f 6.4.3 +da372c745e0f053bb7a64e74cccd15810d96341d 6.4.4
--- a/hgext/clonebundles.py Tue Jun 20 02:36:52 2023 +0200 +++ b/hgext/clonebundles.py Sun Jun 18 00:09:39 2023 +0200 @@ -79,8 +79,10 @@ Bundle files can be generated with the :hg:`bundle` command. Typically :hg:`bundle --all` is used to produce a bundle of the entire repository. -:hg:`debugcreatestreamclonebundle` can be used to produce a special -*streaming clonebundle*. These are bundle files that are extremely efficient +The bundlespec option `stream` (see :hg:`help bundlespec`) +can be used to produce a special *streaming clonebundle*, typically using +:hg:`bundle --all --type="none-streamv2"`. +These are bundle files that are extremely efficient to produce and consume (read: fast). However, they are larger than traditional bundle formats and require that clients support the exact set of repository data store formats in use by the repository that created them.
--- a/mercurial/bundlecaches.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/bundlecaches.py Sun Jun 18 00:09:39 2023 +0200 @@ -52,6 +52,14 @@ return url +SUPPORTED_CLONEBUNDLE_SCHEMES = [ + b"http://", + b"https://", + b"largefile://", + CLONEBUNDLESCHEME, +] + + @attr.s class bundlespec: compression = attr.ib() @@ -384,7 +392,9 @@ return False -def filterclonebundleentries(repo, entries, streamclonerequested=False): +def filterclonebundleentries( + repo, entries, streamclonerequested=False, pullbundles=False +): """Remove incompatible clone bundle manifest entries. Accepts a list of entries parsed with ``parseclonebundlesmanifest`` @@ -396,6 +406,16 @@ """ newentries = [] for entry in entries: + url = entry.get(b'URL') + if not pullbundles and not any( + [url.startswith(scheme) for scheme in SUPPORTED_CLONEBUNDLE_SCHEMES] + ): + repo.ui.debug( + b'filtering %s because not a supported clonebundle scheme\n' + % url + ) + continue + spec = entry.get(b'BUNDLESPEC') if spec: try: @@ -405,8 +425,7 @@ # entries. if streamclonerequested and not isstreamclonespec(bundlespec): repo.ui.debug( - b'filtering %s because not a stream clone\n' - % entry[b'URL'] + b'filtering %s because not a stream clone\n' % url ) continue @@ -416,7 +435,7 @@ except error.UnsupportedBundleSpecification as e: repo.ui.debug( b'filtering %s because unsupported bundle ' - b'spec: %s\n' % (entry[b'URL'], stringutil.forcebytestr(e)) + b'spec: %s\n' % (url, stringutil.forcebytestr(e)) ) continue # If we don't have a spec and requested a stream clone, we don't know @@ -424,14 +443,12 @@ elif streamclonerequested: repo.ui.debug( b'filtering %s because cannot determine if a stream ' - b'clone bundle\n' % entry[b'URL'] + b'clone bundle\n' % url ) continue if b'REQUIRESNI' in entry and not sslutil.hassni: - repo.ui.debug( - b'filtering %s because SNI not supported\n' % entry[b'URL'] - ) + repo.ui.debug(b'filtering %s because SNI not supported\n' % url) continue if b'REQUIREDRAM' in entry: @@ -439,15 +456,14 @@ requiredram = util.sizetoint(entry[b'REQUIREDRAM']) except error.ParseError: repo.ui.debug( - b'filtering %s due to a bad REQUIREDRAM attribute\n' - % entry[b'URL'] + b'filtering %s due to a bad REQUIREDRAM attribute\n' % url ) continue actualram = repo.ui.estimatememory() if actualram is not None and actualram * 0.66 < requiredram: repo.ui.debug( b'filtering %s as it needs more than 2/3 of system memory\n' - % entry[b'URL'] + % url ) continue
--- a/mercurial/dirstatemap.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/dirstatemap.py Sun Jun 18 00:09:39 2023 +0200 @@ -4,6 +4,7 @@ # GNU General Public License version 2 or any later version. +import struct from .i18n import _ from . import ( @@ -151,9 +152,15 @@ b'dirstate only has a docket in v2 format' ) self._set_identity() - self._docket = docketmod.DirstateDocket.parse( - self._readdirstatefile(), self._nodeconstants - ) + try: + self._docket = docketmod.DirstateDocket.parse( + self._readdirstatefile(), self._nodeconstants + ) + except struct.error: + self._ui.debug(b"failed to read dirstate-v2 data") + raise error.CorruptedDirstate( + b"failed to read dirstate-v2 data" + ) return self._docket def _read_v2_data(self): @@ -176,11 +183,23 @@ return self._opener.read(self.docket.data_filename()) def write_v2_no_append(self, tr, st, meta, packed): - old_docket = self.docket + try: + old_docket = self.docket + except error.CorruptedDirstate: + # This means we've identified a dirstate-v1 file on-disk when we + # were expecting a dirstate-v2 docket. We've managed to recover + # from that unexpected situation, and now we want to write back a + # dirstate-v2 file to make the on-disk situation right again. + # + # This shouldn't be triggered since `self.docket` is cached and + # we would have called parents() or read() first, but it's here + # just in case. + old_docket = None + new_docket = docketmod.DirstateDocket.with_new_uuid( self.parents(), len(packed), meta ) - if old_docket.uuid == new_docket.uuid: + if old_docket is not None and old_docket.uuid == new_docket.uuid: raise error.ProgrammingError(b'dirstate docket name collision') data_filename = new_docket.data_filename() self._opener.write(data_filename, packed) @@ -194,7 +213,7 @@ st.close() # Remove the old data file after the new docket pointing to # the new data file was written. - if old_docket.uuid: + if old_docket is not None and old_docket.uuid: data_filename = old_docket.data_filename() if tr is not None: tr.addbackup(data_filename, location=b'plain') @@ -211,28 +230,40 @@ def parents(self): if not self._parents: if self._use_dirstate_v2: - self._parents = self.docket.parents + try: + self.docket + except error.CorruptedDirstate as e: + # fall back to dirstate-v1 if we fail to read v2 + self._v1_parents(e) + else: + self._parents = self.docket.parents else: - read_len = self._nodelen * 2 - st = self._readdirstatefile(read_len) - l = len(st) - if l == read_len: - self._parents = ( - st[: self._nodelen], - st[self._nodelen : 2 * self._nodelen], - ) - elif l == 0: - self._parents = ( - self._nodeconstants.nullid, - self._nodeconstants.nullid, - ) - else: - raise error.Abort( - _(b'working directory state appears damaged!') - ) + self._v1_parents() return self._parents + def _v1_parents(self, from_v2_exception=None): + read_len = self._nodelen * 2 + st = self._readdirstatefile(read_len) + l = len(st) + if l == read_len: + self._parents = ( + st[: self._nodelen], + st[self._nodelen : 2 * self._nodelen], + ) + elif l == 0: + self._parents = ( + self._nodeconstants.nullid, + self._nodeconstants.nullid, + ) + else: + hint = None + if from_v2_exception is not None: + hint = _(b"falling back to dirstate-v1 from v2 also failed") + raise error.Abort( + _(b'working directory state appears damaged!'), hint + ) + class dirstatemap(_dirstatemapcommon): """Map encapsulating the dirstate's contents. @@ -330,11 +361,17 @@ def read(self): testing.wait_on_cfg(self._ui, b'dirstate.pre-read-file') if self._use_dirstate_v2: - - if not self.docket.uuid: - return - testing.wait_on_cfg(self._ui, b'dirstate.post-docket-read-file') - st = self._read_v2_data() + try: + self.docket + except error.CorruptedDirstate: + # fall back to dirstate-v1 if we fail to read v2 + self._set_identity() + st = self._readdirstatefile() + else: + if not self.docket.uuid: + return + testing.wait_on_cfg(self._ui, b'dirstate.post-docket-read-file') + st = self._read_v2_data() else: self._set_identity() st = self._readdirstatefile() @@ -365,10 +402,17 @@ # # (we cannot decorate the function directly since it is in a C module) if self._use_dirstate_v2: - p = self.docket.parents - meta = self.docket.tree_metadata - parse_dirstate = util.nogc(v2.parse_dirstate) - parse_dirstate(self._map, self.copymap, st, meta) + try: + self.docket + except error.CorruptedDirstate: + # fall back to dirstate-v1 if we fail to parse v2 + parse_dirstate = util.nogc(parsers.parse_dirstate) + p = parse_dirstate(self._map, self.copymap, st) + else: + p = self.docket.parents + meta = self.docket.tree_metadata + parse_dirstate = util.nogc(v2.parse_dirstate) + parse_dirstate(self._map, self.copymap, st, meta) else: parse_dirstate = util.nogc(parsers.parse_dirstate) p = parse_dirstate(self._map, self.copymap, st) @@ -597,38 +641,37 @@ testing.wait_on_cfg(self._ui, b'dirstate.pre-read-file') if self._use_dirstate_v2: - self.docket # load the data if needed - inode = ( - self.identity.stat.st_ino - if self.identity is not None - and self.identity.stat is not None - else None - ) - testing.wait_on_cfg(self._ui, b'dirstate.post-docket-read-file') - if not self.docket.uuid: - data = b'' - self._map = rustmod.DirstateMap.new_empty() + try: + self.docket + except error.CorruptedDirstate as e: + # fall back to dirstate-v1 if we fail to read v2 + parents = self._v1_map(e) else: - data = self._read_v2_data() - self._map = rustmod.DirstateMap.new_v2( - data, - self.docket.data_size, - self.docket.tree_metadata, - self.docket.uuid, - inode, + parents = self.docket.parents + inode = ( + self.identity.stat.st_ino + if self.identity is not None + and self.identity.stat is not None + else None + ) + testing.wait_on_cfg( + self._ui, b'dirstate.post-docket-read-file' ) - parents = self.docket.parents + if not self.docket.uuid: + data = b'' + self._map = rustmod.DirstateMap.new_empty() + else: + data = self._read_v2_data() + self._map = rustmod.DirstateMap.new_v2( + data, + self.docket.data_size, + self.docket.tree_metadata, + self.docket.uuid, + inode, + ) + parents = self.docket.parents else: - self._set_identity() - inode = ( - self.identity.stat.st_ino - if self.identity is not None - and self.identity.stat is not None - else None - ) - self._map, parents = rustmod.DirstateMap.new_v1( - self._readdirstatefile(), inode - ) + parents = self._v1_map() if parents and not self._dirtyparents: self.setparents(*parents) @@ -638,6 +681,23 @@ self.get = self._map.get return self._map + def _v1_map(self, from_v2_exception=None): + self._set_identity() + inode = ( + self.identity.stat.st_ino + if self.identity is not None and self.identity.stat is not None + else None + ) + try: + self._map, parents = rustmod.DirstateMap.new_v1( + self._readdirstatefile(), inode + ) + except OSError as e: + if from_v2_exception is not None: + raise e from from_v2_exception + raise + return parents + @property def copymap(self): return self._map.copymap() @@ -696,9 +756,15 @@ self._dirtyparents = False return + write_mode = self._write_mode + try: + docket = self.docket + except error.CorruptedDirstate: + # fall back to dirstate-v1 if we fail to parse v2 + docket = None + # We can only append to an existing data file if there is one - write_mode = self._write_mode - if self.docket.uuid is None: + if docket is None or docket.uuid is None: write_mode = WRITE_MODE_FORCE_NEW packed, meta, append = self._map.write_v2(write_mode) if append:
--- a/mercurial/error.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/error.py Sun Jun 18 00:09:39 2023 +0200 @@ -650,6 +650,13 @@ __bytes__ = _tobytes +class CorruptedDirstate(Exception): + """error raised the dirstate appears corrupted on-disk. It may be due to + a dirstate version mismatch (i.e. expecting v2 and finding v1 on disk).""" + + __bytes__ = _tobytes + + class PeerTransportError(Abort): """Transport-level I/O error when communicating with a peer repo."""
--- a/mercurial/helptext/bundlespec.txt Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/helptext/bundlespec.txt Sun Jun 18 00:09:39 2023 +0200 @@ -67,6 +67,10 @@ .. bundlecompressionmarker +The compression engines can be prepended with ``stream`` to create a streaming bundle. +These are bundles that are extremely efficient to produce and consume, +but do not have guaranteed compatibility with older clients. + Available Options ================= @@ -89,7 +93,6 @@ revbranchcache Include the "tags-fnodes" cache inside the bundle. - tagsfnodescache Include the "tags-fnodes" cache inside the bundle. @@ -109,3 +112,10 @@ ``zstd-v1`` This errors because ``zstd`` is not supported for ``v1`` types. + +``none-streamv2`` + Produce a ``v2`` streaming bundle with no compression. + +``zstd-v2;obsolescence=true;phases=true`` + Produce a ``v2`` bundle with zstandard compression which includes + obsolescence markers and phases.
--- a/mercurial/revlog.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/revlog.py Sun Jun 18 00:09:39 2023 +0200 @@ -615,8 +615,8 @@ entry_point = b'%s.i.%s' % (self.radix, self.postfix) elif self._trypending and self.opener.exists(b'%s.i.a' % self.radix): entry_point = b'%s.i.a' % self.radix - elif self._try_split and self.opener.exists(b'%s.i.s' % self.radix): - entry_point = b'%s.i.s' % self.radix + elif self._try_split and self.opener.exists(self._split_index_file): + entry_point = self._split_index_file else: entry_point = b'%s.i' % self.radix @@ -2125,6 +2125,22 @@ raise error.CensoredNodeError(self.display_id, node, text) raise + @property + def _split_index_file(self): + """the path where to expect the index of an ongoing splitting operation + + The file will only exist if a splitting operation is in progress, but + it is always expected at the same location.""" + parts = os.path.split(self.radix) + if len(parts) > 1: + # adds a '-s' prefix to the ``data/` or `meta/` base + head = parts[0] + b'-s' + return os.path.join(head, *parts[1:]) + else: + # the revlog is stored at the root of the store (changelog or + # manifest), no risk of collision. + return self.radix + b'.i.s' + def _enforceinlinesize(self, tr, side_write=True): """Check if the revlog is too big for inline and convert if so. @@ -2161,7 +2177,7 @@ # this code if side_write: old_index_file_path = self._indexfile - new_index_file_path = self._indexfile + b'.s' + new_index_file_path = self._split_index_file opener = self.opener weak_self = weakref.ref(self)
--- a/mercurial/revlogutils/deltas.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/revlogutils/deltas.py Sun Jun 18 00:09:39 2023 +0200 @@ -1087,10 +1087,17 @@ ): self.revlog = revlog self._write_debug = write_debug - self._debug_search = debug_search + if write_debug is None: + self._debug_search = False + else: + self._debug_search = debug_search self._debug_info = debug_info self._snapshot_cache = SnapshotCache() + @property + def _gather_debug(self): + return self._write_debug is not None or self._debug_info is not None + def buildtext(self, revinfo, fh): """Builds a fulltext version of a revision @@ -1136,7 +1143,6 @@ def _builddeltainfo(self, revinfo, base, fh, target_rev=None): # can we use the cached delta? revlog = self.revlog - debug_search = self._write_debug is not None and self._debug_search chainbase = revlog.chainbase(base) if revlog._generaldelta: deltabase = base @@ -1173,7 +1179,7 @@ delta = revinfo.cachedelta[1] if delta is None: delta = self._builddeltadiff(base, revinfo, fh) - if debug_search: + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: uncompressed-delta-size=%d\n" msg %= len(delta) self._write_debug(msg) @@ -1181,17 +1187,17 @@ if revlog.upperboundcomp is not None and snapshotdepth: lowestrealisticdeltalen = len(delta) // revlog.upperboundcomp snapshotlimit = revinfo.textlen >> snapshotdepth - if debug_search: + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: projected-lower-size=%d\n" msg %= lowestrealisticdeltalen self._write_debug(msg) if snapshotlimit < lowestrealisticdeltalen: - if debug_search: + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: DISCARDED (snapshot limit)\n" self._write_debug(msg) return None if revlog.length(base) < lowestrealisticdeltalen: - if debug_search: + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: DISCARDED (prev size)\n" self._write_debug(msg) return None @@ -1253,41 +1259,34 @@ if target_rev is None: target_rev = len(self.revlog) - if not revinfo.textlen: - return self._fullsnapshotinfo(fh, revinfo, target_rev) + gather_debug = self._gather_debug + cachedelta = revinfo.cachedelta + revlog = self.revlog + p1r = p2r = None if excluded_bases is None: excluded_bases = set() - # no delta for flag processor revision (see "candelta" for why) - # not calling candelta since only one revision needs test, also to - # avoid overhead fetching flags again. - if revinfo.flags & REVIDX_RAWTEXT_CHANGING_FLAGS: - return self._fullsnapshotinfo(fh, revinfo, target_rev) - - gather_debug = ( - self._write_debug is not None or self._debug_info is not None - ) - debug_search = self._write_debug is not None and self._debug_search - if gather_debug: start = util.timer() - - # count the number of different delta we tried (for debug purpose) - dbg_try_count = 0 - # count the number of "search round" we did. (for debug purpose) - dbg_try_rounds = 0 - dbg_type = b'unknown' - - cachedelta = revinfo.cachedelta - p1 = revinfo.p1 - p2 = revinfo.p2 - revlog = self.revlog - - deltainfo = None - p1r, p2r = revlog.rev(p1), revlog.rev(p2) - - if gather_debug: + dbg = self._one_dbg_data() + dbg['revision'] = target_rev + target_revlog = b"UNKNOWN" + target_type = self.revlog.target[0] + target_key = self.revlog.target[1] + if target_type == KIND_CHANGELOG: + target_revlog = b'CHANGELOG:' + elif target_type == KIND_MANIFESTLOG: + target_revlog = b'MANIFESTLOG:' + if target_key: + target_revlog += b'%s:' % target_key + elif target_type == KIND_FILELOG: + target_revlog = b'FILELOG:' + if target_key: + target_revlog += b'%s:' % target_key + dbg['target-revlog'] = target_revlog + p1r = revlog.rev(revinfo.p1) + p2r = revlog.rev(revinfo.p2) if p1r != nullrev: p1_chain_len = revlog._chaininfo(p1r)[0] else: @@ -1296,7 +1295,109 @@ p2_chain_len = revlog._chaininfo(p2r)[0] else: p2_chain_len = -1 - if debug_search: + dbg['p1-chain-len'] = p1_chain_len + dbg['p2-chain-len'] = p2_chain_len + + # 1) if the revision is empty, no amount of delta can beat it + # + # 2) no delta for flag processor revision (see "candelta" for why) + # not calling candelta since only one revision needs test, also to + # avoid overhead fetching flags again. + if not revinfo.textlen or revinfo.flags & REVIDX_RAWTEXT_CHANGING_FLAGS: + deltainfo = self._fullsnapshotinfo(fh, revinfo, target_rev) + if gather_debug: + end = util.timer() + dbg['duration'] = end - start + dbg[ + 'delta-base' + ] = deltainfo.base # pytype: disable=attribute-error + dbg['search_round_count'] = 0 + dbg['using-cached-base'] = False + dbg['delta_try_count'] = 0 + dbg['type'] = b"full" + dbg['snapshot-depth'] = 0 + self._dbg_process_data(dbg) + return deltainfo + + deltainfo = None + + # If this source delta are to be forcibly reuse, let us comply early. + if ( + revlog._generaldelta + and revinfo.cachedelta is not None + and revinfo.cachedelta[2] == DELTA_BASE_REUSE_FORCE + ): + base = revinfo.cachedelta[0] + if base == nullrev: + dbg_type = b"full" + deltainfo = self._fullsnapshotinfo(fh, revinfo, target_rev) + if gather_debug: + snapshotdepth = 0 + elif base not in excluded_bases: + delta = revinfo.cachedelta[1] + header, data = revlog.compress(delta) + deltalen = len(header) + len(data) + if gather_debug: + offset = revlog.end(len(revlog) - 1) + chainbase = revlog.chainbase(base) + distance = deltalen + offset - revlog.start(chainbase) + chainlen, compresseddeltalen = revlog._chaininfo(base) + chainlen += 1 + compresseddeltalen += deltalen + if base == p1r or base == p2r: + dbg_type = b"delta" + snapshotdepth = None + elif not revlog.issnapshot(base): + snapshotdepth = None + else: + dbg_type = b"snapshot" + snapshotdepth = revlog.snapshotdepth(base) + 1 + else: + distance = None + chainbase = None + chainlen = None + compresseddeltalen = None + snapshotdepth = None + deltainfo = _deltainfo( + distance=distance, + deltalen=deltalen, + data=(header, data), + base=base, + chainbase=chainbase, + chainlen=chainlen, + compresseddeltalen=compresseddeltalen, + snapshotdepth=snapshotdepth, + ) + + if deltainfo is not None: + if gather_debug: + end = util.timer() + dbg['duration'] = end - start + dbg[ + 'delta-base' + ] = deltainfo.base # pytype: disable=attribute-error + dbg['search_round_count'] = 0 + dbg['using-cached-base'] = True + dbg['delta_try_count'] = 0 + dbg['type'] = b"full" + if snapshotdepth is None: + dbg['snapshot-depth'] = 0 + else: + dbg['snapshot-depth'] = snapshotdepth + self._dbg_process_data(dbg) + return deltainfo + + # count the number of different delta we tried (for debug purpose) + dbg_try_count = 0 + # count the number of "search round" we did. (for debug purpose) + dbg_try_rounds = 0 + dbg_type = b'unknown' + + if p1r is None: + p1r = revlog.rev(revinfo.p1) + p2r = revlog.rev(revinfo.p2) + + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: SEARCH rev=%d\n" msg %= target_rev self._write_debug(msg) @@ -1314,7 +1415,7 @@ candidaterevs = next(groups) while candidaterevs is not None: dbg_try_rounds += 1 - if debug_search: + if self._debug_search: prev = None if deltainfo is not None: prev = deltainfo.base @@ -1325,7 +1426,7 @@ and cachedelta[0] in candidaterevs ): round_type = b"cached-delta" - elif p1 in candidaterevs or p2 in candidaterevs: + elif p1r in candidaterevs or p2r in candidaterevs: round_type = b"parents" elif prev is not None and all(c < prev for c in candidaterevs): round_type = b"refine-down" @@ -1338,7 +1439,7 @@ self._write_debug(msg) nominateddeltas = [] if deltainfo is not None: - if debug_search: + if self._debug_search: msg = ( b"DBG-DELTAS-SEARCH: CONTENDER: rev=%d - length=%d\n" ) @@ -1348,14 +1449,14 @@ # challenge it against refined candidates nominateddeltas.append(deltainfo) for candidaterev in candidaterevs: - if debug_search: + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: CANDIDATE: rev=%d\n" msg %= candidaterev self._write_debug(msg) candidate_type = None - if candidaterev == p1: + if candidaterev == p1r: candidate_type = b"p1" - elif candidaterev == p2: + elif candidaterev == p2r: candidate_type = b"p2" elif self.revlog.issnapshot(candidaterev): candidate_type = b"snapshot-%d" @@ -1376,7 +1477,7 @@ dbg_try_count += 1 - if debug_search: + if self._debug_search: delta_start = util.timer() candidatedelta = self._builddeltainfo( revinfo, @@ -1384,23 +1485,23 @@ fh, target_rev=target_rev, ) - if debug_search: + if self._debug_search: delta_end = util.timer() msg = b"DBG-DELTAS-SEARCH: delta-search-time=%f\n" msg %= delta_end - delta_start self._write_debug(msg) if candidatedelta is not None: if is_good_delta_info(self.revlog, candidatedelta, revinfo): - if debug_search: + if self._debug_search: msg = b"DBG-DELTAS-SEARCH: DELTA: length=%d (GOOD)\n" msg %= candidatedelta.deltalen self._write_debug(msg) nominateddeltas.append(candidatedelta) - elif debug_search: + elif self._debug_search: msg = b"DBG-DELTAS-SEARCH: DELTA: length=%d (BAD)\n" msg %= candidatedelta.deltalen self._write_debug(msg) - elif debug_search: + elif self._debug_search: msg = b"DBG-DELTAS-SEARCH: NO-DELTA\n" self._write_debug(msg) if nominateddeltas: @@ -1434,17 +1535,14 @@ and dbg_try_count == 1 and deltainfo.base == cachedelta[0] ) - dbg = { - 'duration': end - start, - 'revision': target_rev, - 'delta-base': deltainfo.base, # pytype: disable=attribute-error - 'search_round_count': dbg_try_rounds, - 'using-cached-base': used_cached, - 'delta_try_count': dbg_try_count, - 'type': dbg_type, - 'p1-chain-len': p1_chain_len, - 'p2-chain-len': p2_chain_len, - } + dbg['duration'] = end - start + dbg[ + 'delta-base' + ] = deltainfo.base # pytype: disable=attribute-error + dbg['search_round_count'] = dbg_try_rounds + dbg['using-cached-base'] = used_cached + dbg['delta_try_count'] = dbg_try_count + dbg['type'] = dbg_type if ( deltainfo.snapshotdepth # pytype: disable=attribute-error is not None @@ -1454,55 +1552,58 @@ ] = deltainfo.snapshotdepth # pytype: disable=attribute-error else: dbg['snapshot-depth'] = 0 - target_revlog = b"UNKNOWN" - target_type = self.revlog.target[0] - target_key = self.revlog.target[1] - if target_type == KIND_CHANGELOG: - target_revlog = b'CHANGELOG:' - elif target_type == KIND_MANIFESTLOG: - target_revlog = b'MANIFESTLOG:' - if target_key: - target_revlog += b'%s:' % target_key - elif target_type == KIND_FILELOG: - target_revlog = b'FILELOG:' - if target_key: - target_revlog += b'%s:' % target_key - dbg['target-revlog'] = target_revlog + self._dbg_process_data(dbg) + return deltainfo - if self._debug_info is not None: - self._debug_info.append(dbg) + def _one_dbg_data(self): + return { + 'duration': None, + 'revision': None, + 'delta-base': None, + 'search_round_count': None, + 'using-cached-base': None, + 'delta_try_count': None, + 'type': None, + 'p1-chain-len': None, + 'p2-chain-len': None, + 'snapshot-depth': None, + 'target-revlog': None, + } + + def _dbg_process_data(self, dbg): + if self._debug_info is not None: + self._debug_info.append(dbg) - if self._write_debug is not None: - msg = ( - b"DBG-DELTAS:" - b" %-12s" - b" rev=%d:" - b" delta-base=%d" - b" is-cached=%d" - b" - search-rounds=%d" - b" try-count=%d" - b" - delta-type=%-6s" - b" snap-depth=%d" - b" - p1-chain-length=%d" - b" p2-chain-length=%d" - b" - duration=%f" - b"\n" - ) - msg %= ( - dbg["target-revlog"], - dbg["revision"], - dbg["delta-base"], - dbg["using-cached-base"], - dbg["search_round_count"], - dbg["delta_try_count"], - dbg["type"], - dbg["snapshot-depth"], - dbg["p1-chain-len"], - dbg["p2-chain-len"], - dbg["duration"], - ) - self._write_debug(msg) - return deltainfo + if self._write_debug is not None: + msg = ( + b"DBG-DELTAS:" + b" %-12s" + b" rev=%d:" + b" delta-base=%d" + b" is-cached=%d" + b" - search-rounds=%d" + b" try-count=%d" + b" - delta-type=%-6s" + b" snap-depth=%d" + b" - p1-chain-length=%d" + b" p2-chain-length=%d" + b" - duration=%f" + b"\n" + ) + msg %= ( + dbg["target-revlog"], + dbg["revision"], + dbg["delta-base"], + dbg["using-cached-base"], + dbg["search_round_count"], + dbg["delta_try_count"], + dbg["type"], + dbg["snapshot-depth"], + dbg["p1-chain-len"], + dbg["p2-chain-len"], + dbg["duration"], + ) + self._write_debug(msg) def delta_compression(default_compression_header, deltainfo):
--- a/mercurial/templatefuncs.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/templatefuncs.py Sun Jun 18 00:09:39 2023 +0200 @@ -50,8 +50,8 @@ @templatefunc(b'date(date[, fmt])') def date(context, mapping, args): - """Format a date. See :hg:`help dates` for formatting - strings. The default is a Unix date format, including the timezone: + """Format a date. The format string uses the Python strftime format. + The default is a Unix date format, including the timezone: "Mon Sep 04 15:13:13 2006 0700".""" if not (1 <= len(args) <= 2): # i18n: "date" is a keyword
--- a/mercurial/transaction.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/transaction.py Sun Jun 18 00:09:39 2023 +0200 @@ -290,6 +290,8 @@ self._backupjournal = b"%s.backupfiles" % self._journal self._backupsfile = opener.open(self._backupjournal, b'w') self._backupsfile.write(b'%d\n' % version) + # the set of temporary files + self._tmp_files = set() if createmode is not None: opener.chmod(self._journal, createmode & 0o666) @@ -354,6 +356,7 @@ file in self._newfiles or file in self._offsetmap or file in self._backupmap + or file in self._tmp_files ): return if self._queue: @@ -368,6 +371,7 @@ file in self._newfiles or file in self._offsetmap or file in self._backupmap + or file in self._tmp_files ): return if offset: @@ -439,6 +443,7 @@ Such files will be deleted when the transaction exits (on both failure and success). """ + self._tmp_files.add(tmpfile) self._addbackupentry((location, b'', tmpfile, False)) @active
--- a/mercurial/wireprotov1server.py Tue Jun 20 02:36:52 2023 +0200 +++ b/mercurial/wireprotov1server.py Sun Jun 18 00:09:39 2023 +0200 @@ -437,7 +437,7 @@ if not manifest: return None res = bundlecaches.parseclonebundlesmanifest(repo, manifest) - res = bundlecaches.filterclonebundleentries(repo, res) + res = bundlecaches.filterclonebundleentries(repo, res, pullbundles=True) if not res: return None cl = repo.unfiltered().changelog
--- a/relnotes/6.4 Tue Jun 20 02:36:52 2023 +0200 +++ b/relnotes/6.4 Sun Jun 18 00:09:39 2023 +0200 @@ -1,3 +1,14 @@ += Mercurial 6.4.4 = + + * clonebundles: filter out invalid schemes instead of failing on them + * doc: format argument for date uses strftime format string (issue6818) + * test: make test-contrib-perf.t more robust + * revlog: fix a bug in revlog splitting + * bundles: clarify streaming v2 bundle usage + * delta-find: fix pulled-delta-reuse-policy=forced behavior + * dirstate: fall back to v1 if reading v2 failed + * revlog: avoid possible collision between directory and temporary index + = Mercurial 6.4.3 = * chg: declare environ (issue6812)
--- a/rust/hg-core/src/repo.rs Tue Jun 20 02:36:52 2023 +0200 +++ b/rust/hg-core/src/repo.rs Sun Jun 18 00:09:39 2023 +0200 @@ -232,7 +232,17 @@ try_with_lock_no_wait(self.hg_vfs(), "wlock", f) } - pub fn has_dirstate_v2(&self) -> bool { + /// Whether this repo should use dirstate-v2. + /// The presence of `dirstate-v2` in the requirements does not mean that + /// the on-disk dirstate is necessarily in version 2. In most cases, + /// a dirstate-v2 file will indeed be found, but in rare cases (like the + /// upgrade mechanism being cut short), the on-disk version will be a + /// v1 file. + /// Semantically, having a requirement only means that a client cannot + /// properly understand or properly update the repo if it lacks the support + /// for the required feature, but not that that feature is actually used + /// in all occasions. + pub fn use_dirstate_v2(&self) -> bool { self.requirements .contains(requirements::DIRSTATE_V2_REQUIREMENT) } @@ -277,10 +287,21 @@ let dirstate = self.dirstate_file_contents()?; let parents = if dirstate.is_empty() { DirstateParents::NULL - } else if self.has_dirstate_v2() { - let docket = - crate::dirstate_tree::on_disk::read_docket(&dirstate)?; - docket.parents() + } else if self.use_dirstate_v2() { + let docket_res = + crate::dirstate_tree::on_disk::read_docket(&dirstate); + match docket_res { + Ok(docket) => docket.parents(), + Err(_) => { + log::info!( + "Parsing dirstate docket failed, \ + falling back to dirstate-v1" + ); + *crate::dirstate::parsers::parse_dirstate_parents( + &dirstate, + )? + } + } } else { *crate::dirstate::parsers::parse_dirstate_parents(&dirstate)? }; @@ -296,7 +317,7 @@ &self, ) -> Result<DirstateMapIdentity, HgError> { assert!( - self.has_dirstate_v2(), + self.use_dirstate_v2(), "accessing dirstate data file ID without dirstate-v2" ); // Get the identity before the contents since we could have a race @@ -308,15 +329,35 @@ self.dirstate_parents.set(DirstateParents::NULL); Ok((identity, None, 0)) } else { - let docket = - crate::dirstate_tree::on_disk::read_docket(&dirstate)?; - self.dirstate_parents.set(docket.parents()); - Ok((identity, Some(docket.uuid.to_owned()), docket.data_size())) + let docket_res = + crate::dirstate_tree::on_disk::read_docket(&dirstate); + match docket_res { + Ok(docket) => { + self.dirstate_parents.set(docket.parents()); + Ok(( + identity, + Some(docket.uuid.to_owned()), + docket.data_size(), + )) + } + Err(_) => { + log::info!( + "Parsing dirstate docket failed, \ + falling back to dirstate-v1" + ); + let parents = + *crate::dirstate::parsers::parse_dirstate_parents( + &dirstate, + )?; + self.dirstate_parents.set(parents); + Ok((identity, None, 0)) + } + } } } fn new_dirstate_map(&self) -> Result<OwningDirstateMap, DirstateError> { - if self.has_dirstate_v2() { + if self.use_dirstate_v2() { // The v2 dirstate is split into a docket and a data file. // Since we don't always take the `wlock` to read it // (like in `hg status`), it is susceptible to races. @@ -343,7 +384,13 @@ ); continue; } - _ => return Err(e), + _ => { + log::info!( + "Reading dirstate v2 failed, \ + falling back to v1" + ); + return self.new_dirstate_map_v1(); + } }, } } @@ -354,23 +401,22 @@ ); Err(DirstateError::Common(error)) } else { - debug_wait_for_file_or_print( - self.config(), - "dirstate.pre-read-file", - ); - let identity = self.dirstate_identity()?; - let dirstate_file_contents = self.dirstate_file_contents()?; - if dirstate_file_contents.is_empty() { - self.dirstate_parents.set(DirstateParents::NULL); - Ok(OwningDirstateMap::new_empty(Vec::new())) - } else { - let (map, parents) = OwningDirstateMap::new_v1( - dirstate_file_contents, - identity, - )?; - self.dirstate_parents.set(parents); - Ok(map) - } + self.new_dirstate_map_v1() + } + } + + fn new_dirstate_map_v1(&self) -> Result<OwningDirstateMap, DirstateError> { + debug_wait_for_file_or_print(self.config(), "dirstate.pre-read-file"); + let identity = self.dirstate_identity()?; + let dirstate_file_contents = self.dirstate_file_contents()?; + if dirstate_file_contents.is_empty() { + self.dirstate_parents.set(DirstateParents::NULL); + Ok(OwningDirstateMap::new_empty(Vec::new())) + } else { + let (map, parents) = + OwningDirstateMap::new_v1(dirstate_file_contents, identity)?; + self.dirstate_parents.set(parents); + Ok(map) } } @@ -550,7 +596,7 @@ // TODO: Maintain a `DirstateMap::dirty` flag, and return early here if // it’s unset let parents = self.dirstate_parents()?; - let (packed_dirstate, old_uuid_to_remove) = if self.has_dirstate_v2() { + let (packed_dirstate, old_uuid_to_remove) = if self.use_dirstate_v2() { let (identity, uuid, data_size) = self.get_dirstate_data_file_integrity()?; let identity_changed = identity != map.old_identity();
--- a/tests/test-clonebundles.t Tue Jun 20 02:36:52 2023 +0200 +++ b/tests/test-clonebundles.t Sun Jun 18 00:09:39 2023 +0200 @@ -59,6 +59,20 @@ (if this error persists, consider contacting the server operator or disable clone bundles via "--config ui.clonebundles=false") [255] +Manifest file with URL with unknown scheme skips the URL + $ echo 'weirdscheme://does.not.exist/bundle.hg' > server/.hg/clonebundles.manifest + $ hg clone http://localhost:$HGPORT unknown-scheme + no compatible clone bundles available on server; falling back to regular clone + (you may want to report this to the server operator) + requesting all changes + adding changesets + adding manifests + adding file changes + added 2 changesets with 2 changes to 2 files + new changesets 53245c60e682:aaff8d2ffbbf + updating to branch default + 2 files updated, 0 files merged, 0 files removed, 0 files unresolved + Server is not running aborts $ echo "http://localhost:$HGPORT1/bundle.hg" > server/.hg/clonebundles.manifest
--- a/tests/test-contrib-perf.t Tue Jun 20 02:36:52 2023 +0200 +++ b/tests/test-contrib-perf.t Sun Jun 18 00:09:39 2023 +0200 @@ -304,20 +304,20 @@ Multiple entries - $ hg perfparents --config perf.stub=no --config perf.run-limits='500000-1, 0.000000001-5' - ! wall * comb * user * sys * (best of 5) (glob) + $ hg perfparents --config perf.stub=no --config perf.run-limits='500000-1, 0.000000001-50' + ! wall * comb * user * sys * (best of 50) (glob) error case are ignored - $ hg perfparents --config perf.stub=no --config perf.run-limits='500, 0.000000001-5' + $ hg perfparents --config perf.stub=no --config perf.run-limits='500, 0.000000001-50' malformatted run limit entry, missing "-": 500 - ! wall * comb * user * sys * (best of 5) (glob) - $ hg perfparents --config perf.stub=no --config perf.run-limits='aaa-12, 0.000000001-5' - malformatted run limit entry, could not convert string to float: 'aaa': aaa-12 - ! wall * comb * user * sys * (best of 5) (glob) - $ hg perfparents --config perf.stub=no --config perf.run-limits='12-aaaaaa, 0.000000001-5' - malformatted run limit entry, invalid literal for int() with base 10: 'aaaaaa': 12-aaaaaa - ! wall * comb * user * sys * (best of 5) (glob) + ! wall * comb * user * sys * (best of 50) (glob) + $ hg perfparents --config perf.stub=no --config perf.run-limits='aaa-120, 0.000000001-50' + malformatted run limit entry, could not convert string to float: 'aaa': aaa-120 + ! wall * comb * user * sys * (best of 50) (glob) + $ hg perfparents --config perf.stub=no --config perf.run-limits='120-aaaaaa, 0.000000001-50' + malformatted run limit entry, invalid literal for int() with base 10: 'aaaaaa': 120-aaaaaa + ! wall * comb * user * sys * (best of 50) (glob) test actual output ------------------
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/tests/test-dirstate-version-fallback.t Sun Jun 18 00:09:39 2023 +0200 @@ -0,0 +1,51 @@ + $ cat >> $HGRCPATH << EOF + > [storage] + > dirstate-v2.slow-path=allow + > [format] + > use-dirstate-v2=no + > EOF + +Set up a v1 repo + + $ hg init repo + $ cd repo + $ echo a > a + $ hg add a + $ hg commit -m a + $ hg debugrequires | grep dirstate + [1] + $ ls -1 .hg/dirstate* + .hg/dirstate + +Copy v1 dirstate + $ cp .hg/dirstate $TESTTMP/dirstate-v1-backup + +Upgrade it to v2 + + $ hg debugupgraderepo -q --config format.use-dirstate-v2=1 --run | egrep 'added:|removed:' + added: dirstate-v2 + $ hg debugrequires | grep dirstate + dirstate-v2 + $ ls -1 .hg/dirstate* + .hg/dirstate + .hg/dirstate.* (glob) + +Manually reset to dirstate v1 to simulate an incomplete dirstate-v2 upgrade + + $ rm .hg/dirstate* + $ cp $TESTTMP/dirstate-v1-backup .hg/dirstate + +There should be no errors, but a v2 dirstate should be written back to disk + $ hg st + $ ls -1 .hg/dirstate* + .hg/dirstate + .hg/dirstate.* (glob) + +Corrupt the dirstate to see how the errors show up to the user + $ echo "I ate your data" > .hg/dirstate + + $ hg st + abort: working directory state appears damaged! (no-rhg !) + (falling back to dirstate-v1 from v2 also failed) (no-rhg !) + abort: Too little data for dirstate: 16 bytes. (rhg !) + [255]
--- a/tests/test-revlog-delta-find.t Tue Jun 20 02:36:52 2023 +0200 +++ b/tests/test-revlog-delta-find.t Sun Jun 18 00:09:39 2023 +0200 @@ -329,5 +329,34 @@ DBG-DELTAS: CHANGELOG: * (glob) DBG-DELTAS: MANIFESTLOG: * (glob) DBG-DELTAS: MANIFESTLOG: * (glob) - DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 * (glob) - DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=3 * (glob) + DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=2 is-cached=1 *search-rounds=0 try-count=0* (glob) + DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=3 is-cached=1 *search-rounds=0 try-count=0* (glob) + +Check that running "forced" on a non-general delta repository does not corrupt it +--------------------------------------------------------------------------------- + +Even if requested to be used, some of the delta in the revlog cannot be stored on a non-general delta repository. We check that the bundle application was correct. + + $ hg init \ + > --config format.usegeneraldelta=no \ + > --config format.sparse-revlog=no \ + > local-forced-full-p1-no-gd + $ hg debugformat -R local-forced-full-p1-no-gd | grep generaldelta + generaldelta: no + $ hg -R local-forced-full-p1-no-gd pull --quiet local-pre-pull-full \ + > --config debug.revlog.debug-delta=no + $ hg -R local-forced-full-p1-no-gd pull --quiet \ + > --config 'paths.*:pulled-delta-reuse-policy=forced' all-p1.hg + DBG-DELTAS: CHANGELOG: * (glob) + DBG-DELTAS: CHANGELOG: * (glob) + DBG-DELTAS: MANIFESTLOG: * (glob) + DBG-DELTAS: MANIFESTLOG: * (glob) + DBG-DELTAS: FILELOG:my-file.txt: rev=3: delta-base=0 * - search-rounds=1 try-count=1 * (glob) + DBG-DELTAS: FILELOG:my-file.txt: rev=4: delta-base=4 * - search-rounds=1 try-count=1 * (glob) + $ hg -R local-forced-full-p1-no-gd verify + checking changesets + checking manifests + crosschecking files in changesets and manifests + checking files + checking dirstate + checked 5 changesets with 5 changes to 1 files
--- a/tests/test-transaction-rollback-on-revlog-split.t Tue Jun 20 02:36:52 2023 +0200 +++ b/tests/test-transaction-rollback-on-revlog-split.t Sun Jun 18 00:09:39 2023 +0200 @@ -84,6 +84,8 @@ > Directory_With,Special%Char/Complex_File.babar > foo/bar/babar_celeste/foo > 1234567890/1234567890/1234567890/1234567890/1234567890/1234567890/1234567890/1234567890/1234567890/1234567890/f + > some_dir/sub_dir/foo_bar + > some_dir/sub_dir/foo_bar.i.s/tutu > " $ for f in $files; do > mkdir -p `dirname $f` @@ -104,13 +106,17 @@ > dd if=/dev/zero of=$f bs=1k count=128 > /dev/null 2>&1 > done $ hg commit -AqmD --traceback + $ for f in $files; do + > dd if=/dev/zero of=$f bs=1k count=132 > /dev/null 2>&1 + > done + $ hg commit -AqmD --traceback Reference size: $ f -s file - file: size=131072 - $ f -s .hg/store/data/file* - .hg/store/data/file.d: size=132139 - .hg/store/data/file.i: size=256 + file: size=135168 + $ f -s .hg/store/data*/file* + .hg/store/data/file.d: size=267307 + .hg/store/data/file.i: size=320 $ cd .. @@ -134,16 +140,16 @@ adding changesets adding manifests adding file changes - added 2 changesets with 8 changes to 4 files - new changesets 16a630ece54e:8437c461d70a + added 3 changesets with 18 changes to 6 files + new changesets c99a94cae9b1:64874a3b0160 (run 'hg update' to get a working copy) The inline revlog has been replaced $ f -s .hg/store/data/file* - .hg/store/data/file.d: size=132139 - .hg/store/data/file.i: size=256 + .hg/store/data/file.d: size=267307 + .hg/store/data/file.i: size=320 $ hg verify -q @@ -171,7 +177,7 @@ Reference size: $ f -s file file: size=1024 - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.i: size=1174 $ cat > .hg/hgrc <<EOF @@ -192,10 +198,13 @@ The inline revlog still exist, but a split version exist next to it - $ f -s .hg/store/data/file* - .hg/store/data/file.d: size=132139 + $ cat .hg/store/journal | tr '\0' ' ' | grep '\.s' + data/some_dir/sub_dir/foo_bar.i.s/tutu.i 1174 + data/some_dir/sub_dir/foo_bar.i.s/tutu.d 0 + $ f -s .hg/store/data*/file* + .hg/store/data-s/file: size=320 + .hg/store/data/file.d: size=267307 .hg/store/data/file.i: size=132395 - .hg/store/data/file.i.s: size=256 The first file.i entry should match the "Reference size" above. @@ -206,19 +215,19 @@ $ cat .hg/store/journal | tr -s '\000' ' ' | grep data/file data/file.i 1174 data/file.d 0 - $ cat .hg/store/journal.backupfiles | tr -s '\000' ' ' | tr -s '\00' ' '| grep data/file + $ cat .hg/store/journal.backupfiles | tr -s '\000' ' ' | tr -s '\00' ' '| grep 'data.*/file' data/file.i data/journal.backup.file.i.bck 0 - data/file.i.s 0 + data-s/file 0 recover is rolling the split back, the fncache is still valid $ hg recover rolling back interrupted transaction (verify step skipped, run `hg verify` to check your repository content) - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.i: size=1174 $ hg tip - changeset: 1:cc8dfb126534 + changeset: 1:64b04c8dc267 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 @@ -243,7 +252,7 @@ Reference size: $ f -s file file: size=1024 - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.i: size=1174 $ cat > .hg/hgrc <<EOF @@ -271,12 +280,12 @@ The inline revlog still exist, but a split version exist next to it - $ f -s .hg/store/data/file* - .hg/store/data/file.d: size=132139 + $ f -s .hg/store/data*/file* + .hg/store/data-s/file: size=320 + .hg/store/data/file.d: size=267307 .hg/store/data/file.i: size=132395 - .hg/store/data/file.i.s: size=256 - $ cat .hg/store/journal | tr -s '\000' ' ' | grep data/file + $ cat .hg/store/journal | tr -s '\000' ' ' | grep 'data.*/file' data/file.i 1174 data/file.d 0 @@ -285,10 +294,10 @@ $ hg recover rolling back interrupted transaction (verify step skipped, run `hg verify` to check your repository content) - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.i: size=1174 $ hg tip - changeset: 1:cc8dfb126534 + changeset: 1:64b04c8dc267 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 @@ -308,7 +317,7 @@ Reference size: $ f -s file file: size=1024 - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.i: size=1174 $ cat > .hg/hgrc <<EOF @@ -336,11 +345,11 @@ The inline revlog was over written on disk - $ f -s .hg/store/data/file* - .hg/store/data/file.d: size=132139 - .hg/store/data/file.i: size=256 + $ f -s .hg/store/data*/file* + .hg/store/data/file.d: size=267307 + .hg/store/data/file.i: size=320 - $ cat .hg/store/journal | tr -s '\000' ' ' | grep data/file + $ cat .hg/store/journal | tr -s '\000' ' ' | grep 'data.*/file' data/file.i 1174 data/file.d 0 @@ -349,10 +358,10 @@ $ hg recover rolling back interrupted transaction (verify step skipped, run `hg verify` to check your repository content) - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.i: size=1174 $ hg tip - changeset: 1:cc8dfb126534 + changeset: 1:64b04c8dc267 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 @@ -386,13 +395,13 @@ The split was rollback - $ f -s .hg/store/data/file* + $ f -s .hg/store/data*/file* .hg/store/data/file.d: size=0 .hg/store/data/file.i: size=1174 $ hg tip - changeset: 1:cc8dfb126534 + changeset: 1:64b04c8dc267 tag: tip user: test date: Thu Jan 01 00:00:00 1970 +0000 @@ -472,7 +481,7 @@ adding changesets adding manifests adding file changes - size=131072 + size=135168 transaction abort! rollback completed abort: pretxnclose.03-abort hook exited with status 1
--- a/tests/test-upgrade-repo.t Tue Jun 20 02:36:52 2023 +0200 +++ b/tests/test-upgrade-repo.t Sun Jun 18 00:09:39 2023 +0200 @@ -839,6 +839,7 @@ 00changelog.i 00manifest.i data + data-s fncache phaseroots requires @@ -862,6 +863,7 @@ 00changelog.i 00manifest.i data + data-s fncache phaseroots requires